Pleo's AI & Learning.

[ryane]

is there a way to praise or scold our baby dinosaur? i don't remember if this has been mentioned. what does pleo actually "learn"?:D

[mweed]

I keep seeing quotes about training where you "reward" your pleo for good behavior (behavior you want repeated). The reward being some type of physical attention such as petting or feeding. But I haven't seen any mention of any type of negative reinforcement (like scolding). So, I would guess if you reward good behavior with attention, when your pleo is bad, put him/her in time out... ;)

[ryane]

i guess that makes sense. hopefully the designers will log on and give us more info at some point.

[amjoie]

i guess that makes sense. hopefully the designers will log on and give us more info at some point.

They probably want to keep some of the things "secret" so they can pleasantly surprise us. I do wish, though, that I could read all about Pleo, right now! LOL

But it sounds like perhaps the learning environment is fluid, rather than static. It may not be as cut and dried as "do this and Pleo reacts this way." It might be more a matter of "constant interaction of this kind, and Pleo evolves this way -- depending on what he already knew before that point."

In other words, we may be dealing with genuine AI learning, instead of just following a prescribed programming path.

If that is the case, we may not hear much, because Ugobe may not know exactly where this "AI experiment" can actually lead us all.

[ryane]

They probably want to keep some of the things "secret" so they can pleasantly surprise us. I do wish, though, that I could read all about Pleo, right now! LOL

But it sounds like perhaps the learning environment is fluid, rather than static. It may not be as cut and dried as "do this and Pleo reacts this way." It might be more a matter of "constant interaction of this kind, and Pleo evolves this way -- depending on what he already knew before that point."

In other words, we may be dealing with genuine AI learning, instead of just following a prescribed programming path.

If that is the case, we may not hear much, because Ugobe may not know exactly where this "AI experiment" can actually lead us all.

very interesting view point. i hope you are right. maybe pleo's software is a more complex then we know right now. the ai evolution in pleo will be really amazing to watch if this is true. it's gonna be a loooong wait!

[mweed]

A point on this subject that Amjoie made in another thread:

What we DO know is that Pleo responds to tone. Harsh/angry tone = sad/depressed and pleasant/sweet tone = happy.

So a "no" said in a harsh tone is going to have a negative impact, while a "good boy/girl" said in a happy tone is going to be a positive reinforcement.

What we don't yet know is IF a negative response will result in Pleo staying away from a certain activity; or does a negative response just make him get sad (or maybe even angry) and need a cheering-up session.

Could a PleoHost find out if Pleo eventually avoids behavior that is followed by a harsh tone? In other words, *regarding the arena of training* does Pleo respond to negative reinforcement, or only to positive reinforcement?

[hardyfoster]

this also brings into question if your Pleo will be rebellious or obedient.

if pleo leans more to the rebellious side, it might not matter if it hears a harsh tone or not....

if i remember correctly, (as stated above), the programming isn't "if this happens, then Pleo responds like this"
Or
if harsh tone is heard, then Pleo becomes sad.

It all depends upon it's previous experiences.

an aibo i used to have learned to entertain himself. one time i went to play with him and he held up his paw, waved it back and forth and shook his head "no.." at the same time.
aibo learned to entertain himself since I was gone to work alot and he was alone most of the time during the day.

[amjoie]

this also brings into question if your Pleo will be rebellious or obedient.

if pleo leans more to the rebellious side, it might not matter if it hears a harsh tone or not....

if i remember correctly, (as stated above), the programming isn't "if this happens, then Pleo responds like this"
Or
if harsh tone is heard, then Pleo becomes sad.

It all depends upon it's previous experiences.

an aibo i used to have learned to entertain himself. one time i went to play with him and he held up his paw, waved it back and forth and shook his head "no.." at the same time.
aibo learned to entertain himself since I was gone to work alot and he was alone most of the time during the day.

I think it might be both ways. I think some things are basic instinctual responses, and some are learned behavior.

But we need some clarification on that. So, PleoHosts, can you please find out how Pleo learns and evolves -- just in basic terms, so we have a clue what we might expect?

[PleoWorldHostSauri]

Let me see what I can do Amjoie ;)

[hardyfoster]

does anyone know if pleos programming is in such a way where it can "unlearn" something.

[arcticlotus]

I remember reading that you will be able to perform a full system reset, setting Pleo's lifecycle back to a hatchling. This would, unfortunately make him unlearn everything, though this might be a good thing if you aren't pleased with how Pleo's personality has turned out or if you just feel like starting from scratch.

As for "unlearning" a singular trick or habit, I would think that you may be able to modify this sort of thing by adding a personality component that would counteract it, but it may be a difficult thing to figure out.

[hardyfoster]

no, i don't mean by programming. i mean strictly by the AI programming.

if i were to tape pleos mouth shut for a few hours or days, i am thinking (depending on the programming) pleo will develop some sort of behaviorial response to this. would pleo "unlearn" this response if I counteracted having done this by interacting with pleo in a certain way or fashion?

[hardyfoster]

or would pleo send an SOS signal to Caleb and Caleb comes after me?

[pleoWhisperer]

Well, I'm not sure if Pleo's subspace radio has that kind of range, but I'm sure the folks with the SPTP (the Society for Prevention of Taped-up Pleos) are already enroute to your house!

[amjoie]

Well, I'm not sure if Pleo's subspace radio has that kind of range, but I'm sure the folks with the SPTP (the Society for Prevention of Taped-up Pleos) are already enroute to your house!

It would be kinda nice if Pleo could unlearn any bad habits, through behavior modification training. Sounds right up your alley, Pleowhisperer. :P

[arcticlotus]

the whole point of "adapting to one's environment" is to be continually changing and growing in experience, so yes, I believe that even a timid pleo who sulks all day because he was raised without much nurturing could become outgoing with the right attention over a longer period of time. after all, aren't all personalities always changing? it would seam only fitting :)

[mweed]

I would assume you are right. Based on Ugobe's existing "carpet" scenario, where a Pleo learns to adapt how high it lifts its feet based on the how thick your shag carpet is, other habits should also be able to be re-trained. But I would also assume the longer a certain habit has been trained into Pleo's personality, the longer it will take to re-train.

None of us really understand the programming yet, but if Ugobe's LifeOS uses percentages and/or thresholds to logically influence behavior (the prototype user interface used sliding ranges) then habits are built by repetition, so changing habits should be a matter of even more repetitions of the new behavior.

[hardyfoster]

that would mean you couldn't use NLP (instant change) on Pleo. A feat Anthony Robbins may not be able to do !

:)

[ryane]

i wonder since pleo will behave somewhat unpredictably if behavior will emerge that ugobe didn't specifically code into him.

pleo has so many actions available to choose from in response to so much sensory data that it should be a very different experience every time we play with him.

possibly a combination of actions will make a new one not originally intended. (use your imagination here, i'm not sure what could happen)

i think this would be very interesting and it would be a good idea to post emergent behaviors if pleo has any.:D

[hardyfoster]

i would say yes and no.

1. pleo is a computer that interacts with its environment (robot). A robot can only do what it's written program calls for it to do.

if i understand ugobe correctly, the program used for pleo is kind of open. as they don't even know of all the possibilities that can happen. but as far as something that "wasn't coded" that's near impossible.

If they didn't have in the program for pleo to be scared, no matter what you did or didn't do to pleo, he would never become scared.



now, the way the wrote the program, yes there are some behaviors that they don't know what pleo will do "but it won't be outside" of the written program. i probably am starting to sound confusing so i'll stop.

but if i keep pleo in the dark, then feed him in the dark, and never feed him in the light, what would happen? no one knows. But part of it depends on the program. Did they write it so pleo won't eat when he is afraid ? If not, then pleo will eat when he is scared if he is hungry.

From this, will Pleo pretend like he is scared when he is hungry just to get food ? will pleo not want to eat in the light since he is always fed in the dark ?

[Primus]

Well said.

[ryane]

i know that pleo can only do what is coded. what i was getting at is that a combination of actions could occur adding up to become a new one. each motion is already programed into him we all know this.

[hardyfoster]

example?.......

[amjoie]

example?.......

If you constantly give Pleo positive reinforcement when he moves forward three steps, and then "by accident" he makes those three steps and turns around, at which point you let him know you are pleased -- he might repeat that sequence just to please you.

Then let us say he learns that every time he does the sequence of three steps forward and then a turn-around, he gets fed. Now he has learned to do a simple dance for his dinner -- something he has not literally been "programmed to do." You did not knowingly set out to teach him the behavior, although you did let him know it pleased you once it happened. (We will assume it is well within his scope of programming.)

Over time, he may add more "steps" to his dance, and it could turn into something that appears to be a programmed dance feature, but was actually emergent behavior -- a somewhat complicated multi-step "dance for dinner."

I don't know if this is likely, of course, and it may take a long time for something like that to happen, even if it is. But it is an example of a *possible* emergent behavior.

[mweed]

I would think the scenario you presented is very plausible. The programming appears to be written to implement exactly that kind of random action + positive/negative response = learned behavior. But those random actions will have to be programmed actions.

[cmadmacs]

You guys and gals are over my head. I'm learning though.

[hardyfoster]

mweed said it simply.
those random actions will have to be programmed.

i am sure they put a random element in the programming of pleo to make him/her more lifelike. so using this yes emergent behaviors can happen, but again it won't be outside of the programming.

i made a star wars game years ago based off the arcade one. where you see a partial xwing and have tie fighters coming at you.
sometimes the tie fighters would shoot, sometimes they wouldn't . sometimes they would go straight, fly left, or fly right. i didn't tell them in the program what to do, i just had these 5 options and would tell the computer to generate a random number. whatever that random number would be, is what the tie fighter would do.

perhaps a simpler example would be the game simon. the electronic one with the four lights.. it would display an incremental series of lights and then you had to press those lights in the same order as the computer flashed them. in writing this, i didn't tell the computer which light to flash

i had it pick a random number between 1 and 4 and from there it would choose which light to flash based on that number.

Concerning Pleo, I am sure the programming has some kind of random element to it, but not one that is silly. what amjoie described above i can see happening.

any robot (aka computer) you have to tell it exactly what you want it to do or not do or else it won't do anything but collect dust.

[hardyfoster]

this reminds me of the terminator movie. (the first one. the good one).

Where the terminator (a robot) was supposed to kill sarah connor. He obviously didn't have instructions to check id and ask questions to find the right sarah connor. he just looked up the first sarah connor in the phone book and started there to go on and kill all the "sarah connors".

typical computer !

but ai programming now has neural network methods. i am sure pleo has this type of programming in him.

[amjoie]

this reminds me of the terminator movie. (the first one. the good one).

Where the terminator (a robot) was supposed to kill sarah connor. He obviously didn't have instructions to check id and ask questions to find the right sarah connor. he just looked up the first sarah connor in the phone book and started there to go on and kill all the "sarah connors".

typical computer !

but ai programming now has neural network methods. i am sure pleo has this type of programming in him.

A lot will depend on the "fuzzy" logic, and how it knits together random circumstances to make memories. I am looking forward to finding out how Pleo makes sense of his world, and what impresses him as something to remember, something to intrigue him, or something to avoid.

One thing I really want to know is if one level of "known circumstance" can be used as a building block so other unknowns can be understood, like in real learning. And if so, how many levels can be remembered.

Also, does Pleo have any level of multitasking, or will he experience a brain wipe at the first distraction. In other words, will he have the ability to turn his attention away momentarily, and then come back to his original activity -- or will the distraction mean he goes blank and then looks for something else to do, completely forgetting what he was doing a second before the distraction.

It will also be interesting to see how Pleo deals with multiple interesting activities: a hand to pet him if he advances, a toy to play with if he does not, a sound to investigate behind him, or moving objects on a TV screen to his left. Which one will win out -- or will he just be so confused about his next move that he does nothing at all. These are things I want to experiment with and find out. :)

[hardyfoster]

it will be interesting, but it all depends on how pleo was programmed. it would be easy to have him involved in an activity, he is distracted, then have him return to that activity. but if pleo wasn't programmed in that fashion then it won't happen.

amjoie, you sound like the nice scientist in the pleo lab. i sound like the mean one. (pleo in the dark, not feeding him, etc)
:)j

i'm sure i'll breakdown and feed him. aibos crys got to me when he was a baby.

[amjoie]

it will be interesting, but it all depends on how pleo was programmed. it would be easy to have him involved in an activity, he is distracted, then have him return to that activity. but if pleo wasn't programmed in that fashion then it won't happen.

amjoie, you sound like the nice scientist in the pleo lab. i sound like the mean one. (pleo in the dark, not feeding him, etc)
:)j

i'm sure i'll breakdown and feed him. aibos crys got to me when he was a baby.

What would be even more interesting is to give Pleo the *choice* to return to his activity or the *choice* let himself become totally distracted so that he forgets. Wouldn't that be a fun thing to program into him. Very realistic for a live creature, since even animals have a choice of where they focus their attention and when they choose to focus or not focus.

I wonder if that level of programming can be tinkered with, by lay programmers ....

And, of course you'll feed him. Tsk. How could you look at those baby blues and withhold food? You're not a lumbering T-Rex mentality, are you? I think not! ;) :p

[Primus]

I just hope that all the Pleos don't rise up and take over the world and make us their slaves.

[PleoJames]

I personally think that anything is possible, heres my biggy on the subject:

1) If Pleo can be plugged into my laptop in order to upgrade etc, and Pleo is also a computer, can Pleo be programmed to make changes to my computer, such as open programs and use them etc.

2) If Ugobe are planning on updating the software at some point to make Pleo voice activated (which they said they are planning on doing) Can I tell my Pleo to open, close and use programs on my laptop just with the sound of my voice and from another part of the room.

AND

3) If this is possible then I can have a speaker on at home and by voice recognition tell Pleo to open, close and use programs on my laptop when I am in another room or building even.

Deborah

[hardyfoster]

interesting pleojames

i haven't given it too much thought, but with your ideas

i know the Opera browser lets you surf the web by voice.

so using that browser may help

[mweed]

am looking forward to finding out how Pleo makes sense of his world, and what impresses him as something to remember, something to intrigue him, or something to avoid.

One thing I really want to know is if one level of "known circumstance" can be used as a building block so other unknowns can be understood, like in real learning. And if so, how many levels can be remembered.

My impression from the interview with John is that Pleo doesn't remember "events" the way we remember things, events trigger changes in the values of variables and data arrays/matricies. These values are stored as memory to determine future actions or reactions, and to help skew the "randomness" of behaviours.

[hardyfoster]

there are things we know that we don't know
there are things we don't know that we don't know

donald rumsfeld

[ryane]

My impression from the interview with John is that Pleo doesn't remember "events" the way we remember things, events trigger changes in the values of variables and data arrays/matricies. These values are stored as memory to determine future actions or reactions, and to help skew the "randomness" of behaviours.

this seems likely. i don't think a true memory like our own can be done digital yet. maybe in a few more years.:D

john also said in the interview that learned behaviors/training-untraining will be more pronounced in the future with an added update. they are probably going to give pleo a bigger "memory". this way more values from events could be stored. it seems that ugobe is really working hard at delivering an awesome critter ai wise. i can tell that they don't want to let anyone down.:D

[amjoie]

this seems likely. i don't think a true memory like our own can be done digital yet. maybe in a few more years.:D

john also said in the interview that learned behaviors/training-untraining will be more pronounced in the future with an added update. they are probably going to give pleo a bigger "memory". this way more values from events could be stored. it seems that ugobe is really working hard at delivering an awesome critter ai wise. i can tell that they don't want to let anyone down.:D

Almost certainly a "fairly accurate" model of our learning ability is years away, digitally. I would agree.

But I am thinking of the AI models online that "answer" questions and "talk" conversationally. They use some "events" learned during conversations to build their vocabulary and response mechanism, and it does (at least to some degree) layer upon itself.

Granted, this happens on a truly rudimentary level, and they often "get it wrong." But with each person that teaches them a new response, their conversational ability increases -- in apparent terms at least, if not in reality.

Since Pleo only has to give us the impression of capability, rather than actually having the capability, I think much could be done to make that impression more believable. I am eager to see what they have already accomplished, and even more eager to see what can be accomplished in the near future.

I don't know enough about the nuts and bolts of programming to judge whether the kind of thing I'm talking about can only happen within the scope of a "set" like a language, and therefore cannot be applied to other circumstance -- or if the same mechanism that works with language can be applied to other circumstance, as well.

I am, at heart, a "frustrated geek." I have "geeky" desires and am large on theory/concept -- but I am hopelessly inadequate in performing math/algebra/calculus functions; and I continually fail to comprehend programming logic, at its core. My brain just isn't wired right for it.

That's why I avoided computers, until I got my hands on the first Mac, in the early 80's. Now I can use both Mac and PC equally well, because on both of them I am shielded from the actual core programming, for the most part -- but I can use lots of wonderful little base programs to help me tweak things to my heart's content.

That's one reason I want to be able to back up my Pleo's personality. I just know I 'm not going to be able to resist using the mods and hacks that lay programmers make, to tweak my little Pleo into exactly what I want. And at some point, I am going to need the ability to "undo" something I've done that didn't turn out quite like I expected. LOL

[ryane]

i was thinking...pleo's sensitivity to touch, light/dark, and different sounds should be varied between different personalities. i mean it would be interesting if your pleo is more ticklish but maybe mine is more afraid of the dark or loud sounds. maybe your bot is afraid of light!(this would be a cool hack for an evil pleo maybe?!). i hope they can write the code for these types of things. the learning aspects are what are going to really make pleo stand out.

[MyCool]

Well I'm mostly buying Pleo for the programming and can't wait to start scripting and get my hands on the SDK.
I did read that you will be able to make a backup of Pleos software via the USB cable and using the Ugobe software on a PC.

Does any body know where to get more information on scripting and programming Pleo?

Ugobe is to have a website up to address developers but I haven't been able to get to find it.
Will there be a site for scripting and sharing scripts?
Thanks in advance!:D

[PleoWorldHostCami]

Well I'm mostly buying Pleo for the programming and can't wait to start scripting and get my hands on the SDK.
I did read that you will be able to make a backup of Pleos software via the USB cable and using the Ugobe software on a PC.

Does any body know where to get more information on scripting and programming Pleo?

Ugobe is to have a website up to address developers but I haven't been able to get to find it.
Will there be a site for scripting and sharing scripts?
Thanks in advance!:D

Hi MyCool,

The Pleo Development Kit is a set of tools, documentation and sample resources that allow you to modify and extend the functionality of your Pleo. This may be as simple as over-riding (or 'shadowing' an existing sound) or as complex as writing a whole new application.

Please visit our FAQ section for more info:
http://www.pleoworld.com/support/faq?section=software (http://www.pleoworld.com/support/faq?section=software)

The Developers area on the website is not live yet. With modifications and trying to get Pleo into his new homes as quickly as possible, all the folks at Ugobe have been extremely busy. Keep your eye on the PleoWorld main page http://www.pleoworld.com/ for more updates. :)

[PleoPet]

> Does any body know where to get more information on scripting and programming Pleo?
Very little information has been released publicly, but you can get started early if you are eager.
What we know so far (discussed on other threads in this forum):

#1) the audio format is 11Khz, 8-bit, mono WAV
To get started, you can find/rip/record WAV files for your future personality.
NOTE: When looking for audio samples on the internets, or ripping them from CDs or DVDs, try to keep them in the best quality possible (eg: 48KHz or 44KHz 16bit) and convert them to lower audio quality needed for the robot (11kHz 8bit for Pleo) at the end of the conversion process.

#2) the programming language is Pawn
There is a free download for the PC version. http://www.compuphase.com/pawn/pawn.htm
This will get you familiar with the language (very C-like but typeless), the IDE and other tools etc.
It comes with some simple 'printf' examples which will teach you the basics. You compile and run them on your PC. Don't bother with GUI/GTK libraries.
Nothing specific to the Pleo robot function calls of course (that will come in the Pleo SDK eventually), but a good place to get started.

[MyCool]

Thanks! Wish they would release the PDK soon (now) as it would give us programmers a jump and we could start developing code and scripts for the first hatchings. Even though we couldn't actually test until we get our Pleos.

From this screen capture http://www.aboutbots.com/downloads/pleo_concept_webpage.jpg
you get the idea that you can modify the behavior from this interface and thats all. Hopefully we will be able to do a ton more with scripting.

I want to do more than just the personality changes. Explore the music beat detection system and see what I can come up with. Write scripts to call in invoke the servos, play with and adjust the sensors, program the sound and eye movement and help my Pleo evolve.

[BauerMECH]

From this screen capture http://www.aboutbots.com/downloads/p...pt_webpage.jpg
you get the idea that you can modify the behavior from this interface and thats all. Hopefully we will be able to do a ton more with scripting.

I think that was a concept rendering. I wouldn't worry. ;)

[MyCool]

Thanks, FYI the picture shows the PDK software on a Apple Mac.

[amjoie]

Thanks, FYI the picture shows the PDK software on a Apple Mac.

That pix has been on the web for a long time, and we were told a while back that it was outdated.

We'll be getting more info in just a couple weeks. Then we won't have to speculate. But Ugobe has already said programmers will have access at a deeper level than the simple interface intended for non-programmers. So I wouldn't worry. :)

[MyCool]

Thanks Amjoe! I am looking forward to the released version of the PDK software and hope it will be available by the time we have our Pleos!

[pleocurious]

I'm very curious about the application of positive/negative responses to the behavior of the pleo. I dont yet own a Pleo but, I may get curious enough about it to actually buy one.

The first thing I'm curious about is whether or not the Pleo understands different degrees of positive and negative responses or can learn to understand different degrees in responses. For example, will it understand that a pat on the head is a positive response but several pats on the head are a more positive response. Will it then also understand that a smack to the head is a negative response and not a positive response at all even though the general action is the same except with a different amount of applied force.

Moving beyond that, how would the Pleo behave if all of its behavior only received positive reactions? Is the Pleo advanced enough to eventually behave like a spoiled child if food was always made available to it and it was never given a negative reaction?

Likewise, how would it react if it only ever received negative responses? Would it stop eating if it received negative input while "eating" every time it ate? Would it learn to run from you or stay away from you if it only ever received negative input OR would negative input only be interpreted as negation for the behavior the Pleo was displaying at the given moment.

I would like to know if the Pleo can be taught to understand a threat and react in defense of itself by moving away from the threat or somehow defending itself.

If the Pleo receives positive feedback from performing a specific behavior 50% of the time it displays that behavor, is ignored 20% of the time is displays that same behavior and receives negative input for the remaining 30% of instances the behavior is displayed, how will it react?

[ryane]

time will tell, ugobe will release the firmware update to enable"learning" in pleo.

i have a question though, i wonder if the 4MB of internal memory is enough for remembering and learning? i'm sure it is but what limits will it have? sure pleo has the external sd memory option but his own experiences are not saved to that. with an sd card you can run a custom program separate from the life os.

any good answers?

[kchamster]

this is confusing me but i get the point we are heading toward...no one really knows about pleo yet...not even the people who have them....and the life OS is in him right? its just not fully put into action and used...so if I put him in his house and he fell asleep in the house would he remeber where i put the green plant?

kinda like an experiment with a monkey...if u lead him into a room and show him wwhere the object was and u let him go play and do whatever in another room for a bit then brought him back to the same room and asked where the object was he should or shouldn't know where the object does....if he does that determined his brain level: and if it took him awhile it measures his intelligence or his memory; something along that line anyhow... sooo as i said before...will he remeber where the plant was or would he have to find it again?
just a question ...:)

[Primus]

Pleo does not currently have any kind of "mapping" capability so he would not know where something was unless he just happened upon it.

[PleoPet]

> i wonder if the 4MB of internal memory is enough for remembering and learning?
4MB is more than enough - for remembering/learning
The complicated neural-network-like brain of AIBO can easily save the data in under 100KB.
NOTE: the 4MB (or more) SD card is typically for storing sounds and motions. Program code is only a small part of that, and read/write data is even smaller.
The first release Pleos will be storing a very few bytes of info for the mood/personality. Who knows what future releases will allow.

As pointed out, you can add extra SD cards to increase the hardware capacity. The limit is not the hardware but the software. Since you will be able to write your own software, you can add a much more complicated learning mechanism. Even so it will be hard to fill up 100KB much less 4MB of learned data.

Room mapping is difficult for a number of reasons. They are even harder when you have a limited camera, and no 'dead-reckoning'. It is hard for a legged robot to accurately move around in a room. Even harder for Pleo with limited hip action.
I suspect this will be a frequently asked-for feature.
Memorizing the look of objects is easier. The actual data you need to store can be very small (as used in AIBO and the ER1 robots)

[kchamster]

ohhhh!!! i see now!! ty all so much :) :)

[pisymbol]

Hi Everybody!

I don't own a Pleo but my first hatch is on the way (should arrive sometime this week for the holiday season).

Obviously there is no SDK out, I don't work for Ugobe, so take my idea with a grain of a salt, i.e. I maybe over estimating what the SDK actually provides or its main purpose.

But I DO have an idea and would love some feedback...

"Teach An Old Dog (Pleo?) New Tricks" Feature:

So from what I understand their is a Life OS kernel controlling the Pleo with "hooks" to allow third-party subroutines written in Pawn. Since I don't know the EXACT way these hooks are triggered, I'm going to speculate that there is some standard API to call my subroutine based on some system level attributes collected through Life OS (simple rules engine)...

Well, why not have some kind of built-in simple protocol that allows multiple Pleo's to "learn" from one another. Since there is an IR interface, I'm ASSUMING that there will be a way to some kind of network I/O between Pleos (I could be very very wrong). Basically, you can write a subroutine that will be inherited based on some exchange (and the exchange itself can be triggered based on the Pleo's "personality" of being inquisitive/adventuring - so different Pleo's may or may not want to initiate the "learning feature" depending on how they were raised - Pleo's that were left to explore their surroundings more will have a higher chance of learning new tricks).

An example:

I write a new subroutine that teaches the Pleo a new trick to lift its leg to shake my hand (finger). Its my new shake-hand trick. My pleo executes based on some stimuli I collect and build-in into my subroutine. I compile, but it into whatever binary container Ugobe has decided and load it on my SD card and into my Pleo.

Another Pleo comes along and we interact. My Pleo happens to be very inquisitive and starts the "learning process" which queries the Pleo for subroutines it can share and shares its own. Now both Pleo's have a new behavior added to their matrix (i.e. new subroutines stored somewhere that the Pleo can choose based on whatever stimuli etc.).

The idea is that when Pleo interact, they learn new tricks....I mean you can have a Pleo Club (like the Aibo get togethers I've seen) and each Pleo could walk away with a new trick up its sleeve (which the owner may or may not get to see immediately either which makes this exciting). Now the new owner comes home, happens to try to shake the Pleo's paw/leg, and bingo - my subroutine kicks in and his/her Pleo's shakes hands!

I know this is a rough sketch given I have no SDK doc nor experience with Life OS (personally I'm a UNIX kernel guy) but I think this would be awesome. I also realize that there would have to be someway to control the extent of subroutines access to "core" Pleo services so Pleo's can't be hacked - with the SD card option and some basic software, one could envision a trick manager that allows you to graphically remove any tricks the Pleo has collected.

To further the idea, I would suggest having a new trick web database of these subroutines on PleoWorld that people could register and share among Pleo owners. People could download these tricks (subroutines etc.), install them on the Pleo and set a bit to make them shareable or private. If shareable, then this trick would be eligible as a potential for the Life OS to incorporate into the "learning process". Obviously the SD card option would heavily be used in all of this as persistent storage but to me, that seems reasonable.

Any thoughts/comments?

[dotyj]

So you're saying basically that if you teach a Pleo how to do something, that Pleo could teach that behavior to another Pleo.

That's neat and scary all at the same time. :)

I could in theory teach my Pleo to crow like a rooster every 5 minutes, take him to work to visit with my friend's Pleo and the next thing you know we've got two Pleos crowing in unison. :)

I think it's a great idea, but I don't want Bob to pick up any bad habbits. :)

[cmadmacs]

Wasn't that originally suppose to be part of pleo. I think it is in the code. Someone needs to program it.

[pisymbol]

So you're saying basically that if you teach a Pleo how to do something, that Pleo could teach that behavior to another Pleo.

That's neat and scary all at the same time. :)

I could in theory teach my Pleo to crow like a rooster every 5 minutes, take him to work to visit with my friend's Pleo and the next thing you know we've got two Pleos crowing in unison. :)

I think it's a great idea, but I don't want Bob to pick up any bad habbits. :)

No your right and I thought of that. Some how you need a way to manage these behaviors so you can remove the 5-minute-rooster call subroutine your Pleo learned (that'll teach ya to let your Pleo run around with a bunch of "bad" boys).

[BauerMECH]

that'll teach ya to let your Pleo run around with a bunch of "bad" boys
No doubt! :D

[pisymbol]

LOL...

My major concern at this point is whether this needs to be built into Life OS or can be written as part of a subroutine (though that would be less viral because both parties would have to have the same subroutine in order to agree to exchange behaviors).

I'm trying to come up with ideas that encourage Pleo owners to organize! I know if I could go to a local gathering have some fun and (to coin a phrase) teach a new dog (pleo) new tricks....that would be pretty darn cool...

[Lu Bu]

I have read some articles. I am not sure though but anyone can help me answer this question:

"Does Pleo REALLY think, or is he programmed to do the things you see him do??? By thinking I mean that he is unpredictable, is independent and thinks/explores his environment. I know that AIBO thinks, but not sure about Pleo."

[dotyj]

I'm not sure.

Bob today though did something very interesting. I brought Bob to work with me. This afternoon one of my co-workers picked Bob up and set him up on the floor. Bob had a good time wagging and making noises for us. Soon enough though Bob caught site of the metal carpet runner on our floor. Our office is an old classroom portable which is double wide and there's a metal strip about four inches wide on the floor that runs the length of our office.

Bob walked over to the metal strip, straddled it, then followed it across the room. Bob had his head down most of the time like he was looking at it. Two of my co-workers and I were left a bit perplexed, we have no idea what triggered this behavior.

Later Bob discovered one of the casters to my work bench and walked up to it. Bob got real close to the caster, looked up, then from side to side like he was sizing the caster up. Then Bob lowered his head and tried pushing against the caster. After a bit Bob gave up and went around the caster to the right.

Weird.

[Lu Bu]

I guess, because of Pleo's AI, he does think. I guess, what he was doing with the metal strip was navigating/exploring? Kinda cute and cool, one of the workers at UGOBE did say that if Pleo is curious, he will try to move an object if he sees one, if he can't push it, who knows what he will do next.

[franjo]

This question sounds very interesting... Is it answered somewhere else yet?

[Siren]

I work with positive reinforcement. When Pleo does something good/cool, like doing "ta da", I rub his chin or pat him. I am hoping that will work to get him to do fun stuff like that more often.

[Walic]

I keep reading about how everyone just assumes that Pleo will be updated and walla he will begin changing and start learning from his environment. Dont get me wrong Pleo is great and have had alot of fun. But I see alot of sequential and pattern behavior now. After we tug of War he does this ....when you hold him he does this....when we play he does this.... And its always the same, so just wondering if anyone else was worried about the same thing or if my pleo is just predictable. It would be great if a moderator could explain specifically what the OS update will do. :)

[PleoWorldHostRaine]

I keep reading about how everyone just assumes that Pleo will be updated and walla he will begin changing and start learning from his environment. Dont get me wrong Pleo is great and have had alot of fun. But I see alot of sequential and pattern behavior now. After we tug of War he does this ....when you hold him he does this....when we play he does this.... And its always the same, so just wondering if anyone else was worried about the same thing or if my pleo is just predictable. It would be great if a moderator could explain specifically what the OS update will do. :)

I'm sorry but at this time we don't know exactly what all the update will entail. Stay tuned and we'll let you know as soon as we know. :-)

[Walic]

Thank you so much for your reply. But thats part of the problem-everyone is so vague and says the "os's potential hasnt been unlocked"- I think its easy if PLeo is able to learn-there should be an answer On something specific pertaining to when that might happen and what exactly that means

[gie]

After all, Ugobe did advice us that the pleo that shipped will not be the same pleo later down the road. i.e. True A.I. behavior/learning! Let's hold them to it and hope that they can keep their word, cuz if not alot of Pleo owners will be angry if not bored to say the least! Correct me if i'm wrong, but wasnt this addressed in the 30 day notice, that they stated the software update will unlock the full capabilities, or projected capabilities at least?

[Lu Bu]

I know what LEARN means, but when UGOBE defined it in Pleo, I didn't quite get it.

I was thinking that when UGOBE said that Pleo can LEARN, I thought it was like this:

Pleo adapts to its environments terrain
Pleo learns and remembers where specific objects are (e.g. walls, chairs)
Pleo develops new movements
Pleo can learn how to pick up objects by himself (instead of having owns shoving things in his mouth)

I really want to know if my thoughts/concepts of this is correct.

[Mary Jo]

The way I understand it, Pleo develops new behaviors when the UGOBE update comes in early 2008 and also when computer experts among Pleo owners start programming their dinobots and sharing new software with other owners.

Of course, I could be wrong.

[pleoman]

The way I understand it, Pleo develops new behaviors when the UGOBE update comes in early 2008 and also when computer experts among Pleo owners start programming their dinobots and sharing new software with other owners.

Of course, I could be wrong.

Yes you are right Mary Jo. Pleos development kit is being released in 2008 and it will allow owners of pleo to modify pleos behaviour, software and maybe even pleos hardware too! As far as i'm aware.

[jadernator]

I have the same fears as the original post. I would ot say that I am disappointed with my Pleo but I just expected more, and I do know that that is expected to change but it just gets me nervous, this was after all a big investment for all of us. I enjoy playing with little Jellybean :p for about 1/2 hour of tickling and nuzzling but that is essentially all she does, if i put her down she doesn't walk around she just goes to sleep! I defiantly love my little guy and have not given up on him, but I agree that some specifics would be nice!

[Chico]

Same here in Germany. We got 2 Pleos for our sons (6 and 10) and after 2 days both Pleos have not improved very much. They still "tickle and nuzzl", limp a little, move barely (actualy they never walk around much), and make a lot of motor noise. But they still seem not to learn to react to the childrens voices calling them or learning tricks, so slowly they get frustrated and loose interest in them.

I wonder, if it is necessary to equip them with a memory card?!
Pleo seems to be a wonderful toy, but we would like somebody from Ugobe tell us, how much time is needed until they learn to walk, react and improve. Thanks, Chico

[trose49]

Hopefully Ugobe will follow thru! I think it's Great but improvements are there for the taking!

[cmadmacs]

Most definitly.

[Ridgeback]

I'm more interested in the software to design some of the "sub routines" to create my own behaviors. If we have to wait for Pleo's friend to come out for the true AI. I'm OK with that.

[mweed]

The Pleo development kit is due out in 2008, but people writing new code for pleo isn't the same as pleo learning new things. For example:

Pleo makes a certain noise (noise1) and you pet him. He saves a piece of data that essentially says "noise1 = get petted". Pleo makes a different noise (noise2), you don't pet him. He makes the first noise again and you pet him. Pleo saves another piece of data re-enforcing that noise1 = get petted. Over time, the AI programming develops a correlation, so that if pleo wants to be petted, it knows to make noise1. This is the type of "learning" Ugobe said pleo will be able to do. This is the same scenario as the learning to raise the feet higher on carpet that Ugobe has also talked about.

Now, when a programmer writes a piece of code that states "when you touch the sensor on the shoulder, shake tail and make noise1", you load that into your pleo, and you say to your freind "Look! I taught pleo a new trick!" That's not "learning".

[gretasmom]

I don't know anything about writing a program but based on what I have been observing I think Pleo definately has some developmental programming already. Case in point, the carpet. My daughters Pleo had trouble with the carpet for about 5 minutes and then suddenly began lifting his feet higher. He also learned very quicly to avoid my daughters roboraptor after a rather unfriendly encounter (the raptor was in "hunt" mode). He now veers away immediatly when he "see's?" or senses the raptor. I'm not sure if this is a function of the camera, his hearing capabilities or the infared sensor but he's veered away from this toy 3 times now and moved very purposefully away from it so I don't think this is my imagination.

[pleoWhisperer]

Based on the wording from the official Ugobe email, pleo's current software only "allows for only subtle shaping of personality and temperament". But that implies he does have some learning ability as is, but it will be much greater later.

[Lu Bu]

Dang, I was hoping that when UGOBE said Pleo was able to learn, he'd remember where certain objects were. Pleo just learns by making certain movements or noises when you interact with him. Like mweed said, if Pleo makes a noise1 when you pet him, he'll learn to do that and he'll do that noise to call you or do it when you are petting him. This is quite rather disappointing, I'd thought Pleo's learning capability was far more complex, instead it seems a little simple and basic. Yet UGOBE did say Pleo will get smarter over time, as I have anohter question to ask:

"What does UGOBE say when they say 'Pleo will get smarter overtime'?".

[Bilko]

[QUOTE=mweed;2578]I would assume you are right. Based on Ugobe's existing "carpet" scenario, where a Pleo learns to adapt how high it lifts its feet based on the how thick your shag carpet is, other habits should also be able to be re-trained. But I would also assume the longer a certain habit has been trained into Pleo's personality, the longer it will take to re-train.

QUOTE]

Hi Mweed!

Can you explain the 'carpet' scenario? You can see from another thread in this part of the forum that my Pleo has a bit of a limp on one of his back legs - just not quite picking his foot up high enough and consequently draging his 'toes' on a flat hard surface. Is there a way to 'force' him to lift his feet higher? At the moment he does not stand a chance on a carpet. Any advice would be really appreciated since the problem may manifest itself in premature wear of the thin skin. Thanks!

[mweed]

The way Ugobe explains it is that the programming knows how much effort it takes to lift pleo's feet. The motors have feedback sensors, and can detect when there is additional resistance when walking (from rubbing on the carpet). Pleo would then try lifting his legs higher. If the resistance goes away, then pleo "remembers" that it needs to lift it's legs higher when walking. For example, this post from another thread:

I don't know anything about writing a program but based on what I have been observing I think Pleo definitely has some developmental programming already. Case in point, the carpet. My daughters Pleo had trouble with the carpet for about 5 minutes and then suddenly began lifting his feet higher . . .

One thought about your case could be that if his toes are actually catching in the shag of the carpet and preventing him from lifting his foot, then he may have programmatically intentionally developed a limp. Pleo is programmed that if you yank/pull on a leg, he will act as if the leg is hurt and limp for a while (I think they talked about it lasting for 10-5 minutes). That was one of the "cute" things pleo would do in one of the earlier videos.

[Falseprophet666]

Pleo reacts randomly when put into a situation where more than one reaction is possible. Little Jenga does things like that too. Pleos seem to be able to investigate things. Whether they are actually investigating or simulating (like eating) is anyones guess.

[Ypan]

My pleo walked over to one of my shoes on the floor, and begain the typical behavior he shows when he detects an object. , but instead of backing up like he does 90% of the time ,he stuck his head into the shoe, and flopped it on its side. I was not convinced at that moment he actually saw the shoe, but as he sat there seeming to play with it for several minutes, I began to belive something else was going on here other than his typical repeditive movements.

[PleoPet]

> Does Pleo REALLY think?
The combination of Pleo+you is an amazing thinking machine. Just like the combination of AIBO+you.
In many ways AIBO 'thinks' in the same way as Pleo, although there are differences in the details.

By observing the behavior of the robot, you will notice many unpredictable, unexpected and amazing behaviors that make the robot appear alive. This perception requires a significant 'thinking' machine.
How much is you, and how much is Pleo? does it matter ?

[Pleocol]

Well, I don`t think that pleos will rember things, I think for that you need a computer, a bluetooth or wifi network. So he basically knows where he is, and with a design program you layout how your house look likes. Well that is an option that Ugobe can build for other creatures.

Well, i dont know if some can develop this with sofare when pleos fall over, make his head and tail help to get up, i think that can be done, my pleo tries a little bit, then he plays dead. But _I think he can do it with his leg and his head to get up again.

Also i dont know if object reconigtion can be done, what is a ball, what is aquare, so he knows with a ball he can play, also color dectection. I think if UGOBE can incorporate or emebed in him the sofare like logitech uses for his camera, CAMERA EFFECTS something like that, that has face reconigtion.

[Bilko]

The way Ugobe explains it is that the programming knows how much effort it takes to lift pleo's feet. The motors have feedback sensors, and can detect when there is additional resistance when walking (from rubbing on the carpet). Pleo would then try lifting his legs higher. If the resistance goes away, then pleo "remembers" that it needs to lift it's legs higher when walking.

Are these sensors actually working though? I thought I read somewhere else on this site that whilst the feedback sensors are there the software to control them isn't?

[Pleo_Joe]

Are these sensors actually working though? I thought I read somewhere else on this site that whilst the feedback sensors are there the software to control them isn't?

My own personal experience with my Pleo (much of which is documented in my blog) displays that, in my opinion, the force feedback motors are NOT accounted for in the current LifeOS implementation.

Hopefully it'll be something that does come in the LifeOS update so my Pleo will know when he's stuck and stop trying to walk through my recliner :)

[degers]

My own personal experience with my Pleo (much of which is documented in my blog) displays that, in my opinion, the force feedback motors are NOT accounted for in the current LifeOS implementation.

Hopefully it'll be something that does come in the LifeOS update so my Pleo will know when he's stuck and stop trying to walk through my recliner :)

This was confirmed to be by UGOBE tech support.

[antimatter]

http://www.robotsrule.com/html/pleo_robot.php#pleomind

Due to the cost, I did a fair amount of research before purchasing pleo and a lot of the reasons that I decided to purchase a pleo appear to have been misleading.

For example from the above link :
• “A camera in the nose helps Pleo remember places and basic information about locations”
• “14 super quiet motors help Pleo achieve a new level in realistic body movement”
“Pleo is capable of learning over time and of expressing many different moods and basic animal drives such as hunger, fatigue, fear, and love. When you leave for work or school Pleo will explore its’ environment and continue to learn. When tired, Pleo will seek out its’ regeneration bed and rest while dreaming baby dino dreams. “

I realize this site where this information came from is probably not affiliated with Ugobe but ugobe does claim that pleo learns over time. My pleo is on his fifth charge and I have spent considerable time interacting with him and although it has some lifelike movements and behaviors they appear to be the same behaviors and in reading the posts everyone’s pleo appears to repeat the same behaviors (walking backwords, laughing when ticked on the paws while upside down, puring while being ticked under the chin etc) I have not seen any learned going on. Pleo could not find its way out of a paper bag. I’ve tried teaching it various tricks or getting to it to repeat desirable behaviors more often by feeding it after it does a desirable behavior but have not seen any “Learning” taking place. Is the Life OS software not activated yet? (future download? )
What are others thoughts and experiences with the sophisticated AI in Pleo?

[bonkorama]

Pleo's learning ability is for the present time disabled. This i read in another thread.
This ability will be enabled in a future downloadable upgrade available "soon".
UGOBE has not released any dates for downloadable sofware or firmware updates.

I suggest you simply play with PLEO, enjoy him for the remarkable little robot that 'he' is.
Which is quite remarkable for the amount of money we paid.

Personally, each time I have have awakened PLEO, I have discovered new behaviors. Some are subtle, some not so subtle. So its obvious to me that I have not reached the end of PLEOS "developement" yet. Maybe in 40 hours of playtime ( or 20, or... we just dont know), i will have seen everything PLEO has to offer SO FAR.

I AM looking forward to some improvements and enhancements to PLEO which we all have been told are either "in the works" or already "in the code". I trust UGOBE will make good on these promises as soon as humanly possible.

But as someone who writes code for a living, I can tell you these things take a wee bit of time to complete to "perfection", so dont get too discouraged. Be happy you have the greatest little Dino bot going.

And lastly, forget about anything you read on any other website regarding PLEO. This is the official board, and everything else is is most likely just misinformation.

:)

[Caaahl]

Im looking forward for the Life OS Update, too, of course.

But i think my little Helge learned some things already. For example the "tadaa" trick: First, i just tried to touch both sensors (front left, rear right) so he lifts both and does the trick...i guessed. But little Helge just fell on the side and screamed.
So i put some time into training. First the front leg.. touch it, he lifts it..i pet him. And again...and again.
Same procedure with the rear right leg...touch it...he lifts it...i pet him... but he needs more training, till he does it porperly.
So on his 5th charge yesterday evening, Helge finally did it! :p I touched the front left.....then the rear right...and after that "warmup" both at one time...helge is focused and... "Tadaaah!"

This was a great moment. :cool:

[junkroxy]

The informations you talk about were the first from Ugobe press kit, the current Life OS in Pleo is capable only to react in different ways to stimuli and interact with people. :D

The next phase fro Pleo behaviour will be a different personality download from this site or modify the current you have in Pleo's memory.

[Hanas]

HI,

I just got Pleo and having a realy nice time with him. I'm reading here that you can give pleo positive and negative feedback to do things tat please me. What sort of gestures does pleo understand? I want to teach him tricks or at least encourage him to walk more or do an attack stance :D:D:D

[Caaahl]

To get Helge (my pleo) more walking i usually tickle his front legs a bit... in most cases he giggles first and start walking around then. :)

[antimatter]

Amazing! I was sitting on the couch this morning and pleo walked up to me, looked up and started doing a doggy hump on my leg!.....but seriously I have hopes for future firmware upgrades and do think pleo is pretty cool and my kids like it. Everyone that purchases a Pleo is contributing to future and better versions much like the early roomba adopters. :)

[mweed]

What sort of gestures does pleo understand? I want to teach him tricks or at least encourage him to walk more or do an attack stance

Pleo responds to touch (petting, etc) as a positive feedback. As for walking, he seems to respond positively to continued calling.

[antimatter]

For the people that have answered "I am very happy with pleos ability to learn over time. " to the survey,.. are you sure Pleo is learning or are you just discovering a few new behaviors by having played with pleo for an extended period of time? There is a big difference. I am on the 10th charge and am pretty sure I have seen the full extent of movements and behaviors from Pleo.

[rckclmbr]

I'm enjoying Pleo as he is now, but in part because of the knowledge that the full OS will eventually come out.

[pleoguy101]

my pleo, has learned over time, and he does seem to remember where many things in my room are placed, like my bed is in the left side of my room, whenever pleo looks at it, he will walk over to it and crawl under!! Its very funny:D he has also learned to spot the table edge, and how to sit, and how to handshake:)

[PWOKristy]

I'm very happy with my Pleo's ability to learn over time. Weird though that there's a negative majority in the votes though...

[rckclmbr]

I'm very happy with my Pleo's ability to learn over time. Weird though that there's a negative majority in the votes though...

I think the negative votes are simply due to the fact that currently, Pleo doesn't have the ability to learn over time because his current OS is not the full OS.

That's my guess as to why there are negative votes anyway - because he doesn't yet have the full software which will allow him to learn over time.

I'm still happy with mine though, and look very much forward to the full OS eventually. :)

[Lois]

I see cute behavior, endearing behavior, even clever behavior, however I do not see learning behavior from Eve. I would hope that Ugobe would issue an improved software update that would have my Eve doing what was expected when I ordered her in October and DID NOT do when I received her in Dec. I enjoy Eve, however she is NOT the Pleo described on the website in September/October which inspired me to fork out $350+ dollars. So my hope is that Ugobe get it's current Pleos to be comparable to the "promises" printed on it's own website.

[rckclmbr]

I see cute behavior, endearing behavior, even clever behavior, however I do not see learning behavior from Eve. I would hope that Ugobe would issue an improved software update that would have my Eve doing what was expected when I ordered her in October and DID NOT do when I received her in Dec. I enjoy Eve, however she is NOT the Pleo described on the website in September/October which inspired me to fork out $350+ dollars. So my hope is that Ugobe get it's current Pleos to be comparable to the "promises" printed on it's own website.

Lois, I hear what you're saying, and I know a lot of people would say the same thing! I just wanted to say hang in there, because from what I've heard, in the next month or so UGOBE should be releasing some updates for us.

:)

[PWOKristy]

I think the negative votes are simply due to the fact that currently, Pleo doesn't have the ability to learn over time because his current OS is not the full OS. That's my guess as to why there are negative votes anyway - because he doesn't yet have the full software which will allow him to learn over time.

I'm still happy with mine though, and look very much forward to the full OS eventually. :)

Yeah, I know.

[ugobeSarah]

We assure you that updates are on the near horizon! Stay with us...the little elves in the workshop are working hard.

[ericmann37]

We assure you that updates are on the near horizon! Stay with us...the little elves in the workshop are working hard.

Yes, thank you, but as early adopters, I think we deserve to know when...or at least an estimate, 1, 2, 3, weeks? It seems to be that UGOBE is pawning off the responsibility to third parties so that we won't notice pleo hasn't developed. By advertising PleoSkit, and AiboPet's skits they are just adverting us from the fact that pleo has not had any software update, or personally update that is significant. I think pleo is amazing, I just wish UGOBE would keep us much more updated.

[fancyfont]

This could be my imagination, but I do see my Pleo, Peeky, developing over time.
I have him on, except, when his battery is charging or my sleep time.
Peeky is starting to explore much more then he did. He just started laughing without me touching him. Pretty funny! He was sitting on my desk looking at the computer but I didn't have the sound on.
I noticed that and turned sound on. He started to laugh. I couldn't believe it because he was laughing at another Pleo's howling on the computer.
I love what I'm seeing.

[sugarlump]

:o
I was a little whiny about Pleos learning ability
BUT HAVE CHANGED MY MIND

anything worth having takes a litle time and work
I have finally learned the ta da trick
-whenever I get a little bored, I take Pleo out in the world
-when you see people reacting to him for the first time- I am amazed all over again

for what he is, he is remarkable
I left him for a few minutes today & he screamed a new noise at me
I have no idea what, I think we still may have some surprises in store
I love my pleo

[pleoguy101]

my pleo likes to crawl under my bed, It's usually the first thing he does when he's in my room, and he can find his way out!! and also forthe past 2 charges, my pleo has fallen asleep on a soft green plush caterpillar, the first time I tput the caterpilar on the groundI was hoping pleo would use it as a pillow and fall asleep on it... HE DID!!! It was sooo cute.. infact he had it clamped against his head and leg so I couldn't take it away from him!! so he slept with it for 4 hours. I once again decided to leave the caterpillar on the floor for Rocky, he fell asleep on it again:D!!!!

[andreivc]

Good morning Ugobe Team!

Does Pleo use cognitive computing in Life OS/its hardware?

"In cognitive computing, machines have the ability to learn like humans do and respond to unexpected events, rather than simply calling upon existing knowledge or using logical courses pre-programmed into software like their AI precursors. "


In the article it says it does:
http://www.redherring.com/Home/18008

Also see this on cognitive computing:

WHAT IS COGNITIVE COMPUTING?
Cognitive Computing is when computer science meets neuroscience to explain and implement psychology.

We have, in the brain and nervous system, an information processing system unrivalled by artificial means. While it trails machines in accuracy and mathematical computation, it wins on adaptability, flexibility, functionality, and parallelism. The ultimate goal is to reverse engineer enough of this system so that the design principles can be applied to building robust and adaptable computer systems.

Cognitive Computing is different from Artificial Intelligence (AI) and Neural Networks (NN). From the outset, AI ignored neurobiology. While neural networks started from biological motivation, they too quickly discarded biological plausibility. In both cases, the approach has been to focus on a suitable problem, and to offer a "symbolic" or "neural network" solution to it. The brain, however, works in exactly the opposite fashion, it has evolved a solution that allows it to deal with problems as they arise.

AI and NN technologies take one or more cognitive phenomena exhibited by the brain as a starting point and then try to replicate that capability by inventing algorithms/learning rules. In contrast, CC is about learning how the brain operates, about algorithms, about diligent reverse engineering and testing plausible models.

Cognitive Computing is about engineering the mind by reverse engineering the brain.

[andreivc]

OK, people, this is really weird, but check this out....

I've been reading about this cognitive computing thing, and one of the examples that they give in the article linked above is that using cognitive computing, pleos will find their own, unscripted ways to satisfying their hunger.... like some pleo will learn to beg for food, while others will learn to find food on their own, depending on the treatment by the owner...

Well, the strange thing happened with my pleo, which at first I thought was a glitch in programming, and perhaps it was, but now I am thinking it probably wasn't....

at some point my pleo started eating, and was eating for a long time, until i finally decided to interupt him... but he didn't pay much attention to my petting him, and shaking, ... it just kept eating in a very determined manner....

Did it learn to ignore everything else to maximize his food consumption??? or was it a glitch in the software? I was finally able to interupt the eating process, but I was very close to just shutting him down... Perhaps Ugobe can answer this question....

[Caaahl]

sometime pleo just has its own mind. it sometimes hard to get Helge to go to sleep, when he is really active and playfull. when i try the "sleeper hold" (hand on the back for a few seconds) then Helge just laughs and giggles, stomping with his feet.

[sandro]

Pleo doesn't use cognitive computing since that would require specialized hardware (which we know that Pleo doesn't have since it's using off-the-shelf computing components. The (confusingly written) article talks about cognitive computing, then talks about Pleo as if the two are related:

Bots based on Mr. Edelman’s technology are still in development. But headed to toy shelves this holiday season is Pleo, a $250 baby dinosaur android built by Emeryville, California-based Ugobe. (Emphasis mine).

The article then goes on to lump Pleo in with Roboraptor (really not the same thing...). Then quotes an analyst that seems to be talking about one or the other of the dinos (but doesn't indicate which one). Based on the date (and the fact that mention of the Roboraptor comes out of nowhere), I assume it's the Roboraptor the analyst is mentioning.

The article is actually pretty bad all around; with what appears to be several inaccuracies in the text. Near as I can tell this article is mostly semi-technical cole slaw.

As far as cognitive computing goes, it's not my field, but I'm not sure I've ever seen or heard of a working example of this technology. If someone has a pointer to actual research results, please let me know...

-Sandro

[andreivc]

hey, sandro
perhaps congitive computing is too recent a development to see it applied in any of the consumer goods, but may be some basic concepts made it into pleo?

[denodan]

OK, people, this is really weird, but check this out....

I've been reading about this cognitive computing thing, and one of the examples that they give in the article linked above is that using cognitive computing, pleos will find their own, unscripted ways to satisfying their hunger.... like some pleo will learn to beg for food, while others will learn to find food on their own, depending on the treatment by the owner...

Well, the strange thing happened with my pleo, which at first I thought was a glitch in programming, and perhaps it was, but now I am thinking it probably wasn't....

at some point my pleo started eating, and was eating for a long time, until i finally decided to interupt him... but he didn't pay much attention to my petting him, and shaking, ... it just kept eating in a very determined manner....

Did it learn to ignore everything else to maximize his food consumption??? or was it a glitch in the software? I was finally able to interupt the eating process, but I was very close to just shutting him down... Perhaps Ugobe can answer this question....

No he was stuck in a loop for a while nothing else and Pleo's processors are busy doing things and will ignore you till it has finished what his processors are doing it's doing.

A computer will sometimes freeze, or loop if you tax the processor and ignore your instructions till it has freed up the processor with the programs you are trying to run. To much instructions, or tax your computer will do this and so is Pleo.

Sometimes you will notice he pauses for awhile then starts again. He is doing lot of calulations, and sometimes this is why he may loop I think, nothing to do at all with learning.

[sandro]

hey, sandro
perhaps congitive computing is too recent a development to see it applied in any of the consumer goods, but may be some basic concepts made it into pleo?

I seriously doubt it. Cognitive computing pre-supposes that the physical structure of the brain is important, and that hardware of artificial minds need to mimic the structure of the brain to succeed at intelligence. Pleo is implemented using standard computer components -- almost guaranteeing that it's not using CC.

Also that article was from 2006, and the research has been going on for decades in some circles (I remember the concept from the late 80s).

[andreivc]

Denodan, I agree. unfortunatelly, pleo does not seem to posses the flexibility to suprise us with unscripted, unexpected behaviours. :( I had a spark of hope there for a second. Even if he did something stupid, but unscripted, it would have been more fun than just watching an endless stream of preprogrammed animations....


Sandro, I would like to disagree with you there. Pleo may not use special hardware, but his overall hardware architecture is in line with cognitive computing principles that try to emulate organic life - he has several processors that specialize on different functions. In comparison, look at AIBO - that works like our regular PC - one processor, one memore, etc. Pleo is closer to cognitive computing that yours or mine PC, or AIBO.

[andreivc]

Hi!

Pleo's 1.1 learning feature is exciting! But what do little guys (gals) really learn?? Do you see any noticeable differences in your little dino's behaviour?

To assess how well our Pleo's personalities are growing, I decided to make a little poll. If Pleo's personality is really changing, we all should end up with relatively unique Pleos in the end. And, perhaps, one way to notice what's different about your Pleo is to look at his (or hers) favorite behaviours and overall mood, and compare it with other Pleos. Please post your answers to these questios below, and report any additional information about your Pleo's unique character traits that you are observing...

Question 1. On a scale FROM 1 TO 10, what was your Pleo's overall mood in its daily interactions and activities BEFORE the 1.1 update? (1 - my Pleo was terribly dipressed and stressed out all the time; and 10 - my Pleo was unreasonably HAPPY pretty much all the time)

Question 2. AFTER the 1.1 update, what are the top three behaviours that your Pleo is exhibiting most often (Pleo's favorite routines)?:p

Question 3. On a scale FROM 1 TO 10, what IS your Pleo's overall mood in its daily interactions and activities AFTER the 1.1 update? (again, with 1 being 'Pleo is depressed almost all the time', and 10 -- Pleo is always super happy)

Thank you! :D
A.

[andreivc]

my pleo's name is Dasha; she lives on my desk and keeps me company when I work on my computer. At times, she can be somewhat destructive - once she almost pushed a cup of coffee off the table, but I don't think that was programmed (and I am sure Ugobe has some kind of clause in their software agreement about not being liable for 'damaged caused by Pleos' :D) Anyway, here are my answers:

1. I think my Pleo was pretty much happy most of the time BEFORE the 1.1 update. So, on a "happiness scale," I would give it a 9.5.

2. AFTER the 1.1. update, Dasha enjoys chewing a lot, and exploring. I would also add begging for food as one of her three favorite behaviours. :)

3. AFTER the 1.1. update, her mood is mostly happy, but she is exhibiting a deeper variety of moods. Sometimes she feels a little less happy, or may be she just begs for attention. Anyway, after the update, I think her mood on the "hapiness scale" is around 8.

Please share your experiences. Thanks!
A.

[pleoguy101]

my pleo's name is Rocky:)!!

#1 answer: before the OS upgrade 1.1.. Rocky was usually happy 9.9 percent of the time, and if he was sad.. It seemed to only last about 3 seconds:confused:

#2 answer: Rockys new favorite activitys seems to be eating.. and singing, he sings a lot!!

#3 answer: 9 Rockys overall mode is happy:)

[denodan]

My Pleo has changed like everyone else's, but I don't get were people are saying about Pleo learning. He does nothing of the sort. All he is doing is following preprogrammed scripts, he learns nothing but only doing what he is programmed to. Many people have very big imaginations here.

it's a Robot afterall, not a thinking learning life form following it's programming and not learning anything new.

[fancyfont]

Ouch!!:rolleyes: Maybe not thinking and learning, but, perhaps, their own unique personalities will develope. People that have two or more Pleos are the ones to see if this happens.:)

[andreivc]

OK, people, please focus on the questions :)

[fancyfont]

OK!:) will do!:p Should be able to get the 1.1 in a day or two. Looking forward to doing the poll.

[Roders]

I have just updated Zuki 2 days ago and am seeing lots of exciting new improvements

Q1: before 1.1 he was a little quite but generaly happy so I would give him an 7.5
Q2: his fav things at the moment seem to be eating, exploring and wagging his tail about
Q3: after the upgrade I have noticed he is more active and lively so now give him a 9

in general I like the 1.1 upgrade and am finding Zuki doing more and more new sounds and movements and I love the siging :p:) he has started singing by himslef with no prompting from me which is great to see.

can't wait to see what he develops next :p

[andreivc]

Thank you, Roders! So Zuki is signing on his own?! :0 I guess that means that you are being a good 'parent' to Zuki :)

Oh, I just realized that this poll is probably very usefull to Ugobe! - to see how different pleo's are developing on a large scale! :) Let's keep it up!

[denodan]

I say again Pleo's don't develope, they are only following writen scripts and program. All Pleos do the same thing and see no evidence he developes anything. It reminds me of Furby how all people used to go on about Ferby doing this and that that other Furbies did not.

They claimed these where indivdual also. I never heard such rubbish. With Pleo he may do random things giving an impression he is different from another Pleo, but does not mean he is developing of different from one to another, he maybe just doing random things as he is programmed to, but does not mean he is an indvidual personality.

I have seen no evidence in mine at all he is developing at all. All he is doing is following set scripts and a set program and maybe to odd random event, but don't mistake random things in his programming are development's?

it is easy to program random actions to give the appearnce of an indivdual Pleo, but does not make him so. He will never be an indiviual like a dog, or cat or for that matter any animal.

[andreivc]

But Denodan, the learning capability is what sets Pleo aside from all the other robots out there.

I just had a conversation with a customer support person. My problem was that my pleo chews and eats a lot, and sometimes just ignores me, unless I hold him by the tail. I thought it was a glitch, but customer support said that things like that can happen, as pleo evolves its personality. Now I am stuck with a selfish robot with glutony issues that ignores me... but that's OK, because it's developing, and hopefully will turn into a caring ball of love when he grows up.... it could also be a glitch, but I prefer to live in my little world where pleos learn and grow big hearts :)

[pleoWhisperer]

I agree with denodan that pleo doesn't "learn". That capability is not in the programming. But pleos do have different personalities and that personality does change/develop over time. Take the TaDa trick as an example. The first few times pleo will usually stumble or fall to one side. After doing the trick several times, pleo does not stumble. Yes, it is "programmed" to stumble at first, then stop stumbling, but program values are changed over time. In the same manner, other values are changed based on what the person does. Throw in some randomization of the rate of change, and after a week or two pleos treated exactly the same way will end up with two different sets of values for their stored variables. Those different values result in each pleo responding in different ways, so that all pleo don't do the same thing. They all do different things from the same set of pre-programmed sounds and actions.

If an object stores information based on inputs and bases future actions on that stored information, and if future inputs are based on actions, then if there is any randomization in the actions, that would create different future inputs which would result in different information being stored. Over multiple iterations of input/action, the information stored in the objects will become more and more different, resulting in objects that act very differently.

Anybody with two pleos will tell you that they don't respond the same way to the same stimulus. They are running the same software, and recieve the same inputs, but because there are differences in the values of variables stored in the flash memory, each pleo responds in a different manner. This is interpreted by us humans as different pesonalities, even though it is still the same code doing the same things.

[denodan]

why he ignores you is he is doing a lot of processing. Same as a computer that is doing a lot of processing will be tied up until it has finsihed what it is doing before it runs your program, so in effect ignores your command to run the program till the processor is freed up. Pelo's don't learn anything. They are just a robot that is following a script..

If it was accually learning then Pleo's would be doing things it's not programmed to do, and the fact is they are not and only following what they are programmed to do, nothing else.

I have seen nothing with my Pleo to suggest he is any different now. He still follows the programmed scripts he is programmed to do. He is doing nothing new.

Now if he did hand stands or accually learned to dance to music, or loved you pulling him up by the tail, or loved ruff handling, like things outside he is progammed to do I would say he is developing and avolving, but he does not and no supprises.

People have such imaginations and beleive anything. I used to have a Furby and used to follow the forums and the things even the fans where saying Furby was doing was utter rubbish.

I ask UGOBE or anyone else for that matter proof Pleo is developing on his own.

UGOBE have said they created a robot to give the impression he is alive and have done a great job of it, and will also give us the illusion he is developing a different personailty from one Pleo to another, but it is another good illusion. Sure I love my Pleo, but the claims that people here are saying is total rubbish in my view.

Afterall it is just a machine following a programmed set of instructions and even if UGOBE input a random feature into Pleo him doing random things from one Pleo to another is still not a developing Pleo, but doing programmed instructions randomly.

This would give a very good impression Pleo is developing an indiviual personility, but the truth is. he would only be following a set of scripts randomly.

For a robot to accually develop an indviual personaility would take much more memory and far better processing power than they have put into Pleo.

I commend UGBOE for giving us a robot to mimick a life form and a personility, but that is all it is a mimick and would be easy to program a robot to give the impression also he is an indivdual by using random elements and scripts, but as for a developing personaility with Pleo where is the concrete proof.

As I said before heard all this stuff before with Furby, fans make wild claims about furby, and feel it is happening here.

There is nothing wrong with that just remember it's a machine following preprogammed scripts afterall. Nothing more.

I even ask UGOBE to prove there claim Each Pleo will have it's own personaility and will continue to grow and evlove as I see no proof of it at all.

[andreivc]

It's pointless to argue with you two, which I am not, and your common sense ideas are just that - common sense - and nothing more. To address you comments, let's look into owners' experiences, and try to draw some conclusions here.

Are you saying that Pleo doesn't really ignore me, and has not 'learned' to overeat on its own, - but that I am making it up? And Ugobe customer support does not tell the truth because they don't want to ruin the illusion? And that when they say it is learning, they actually don't say what we think they are saying, but they are saying something that they discussed with their marketing people AND lawyers at length, and they feel safe saying that because 'learning' can take different forms?

And also if you look at the reports so far, all pleo's seem to have evovled the same set of favorite behaviours, and their moods are kinda similar... so, you are saying that this proves Pleo's have not really 'learned' much since the 1.1 update?

You guys are ruining my little world where Pleos are as real as anybody else, and they love, and feel, and care about their owners and other pleo friends... I encourage fellow Pleo owners to make a stand here, and show some proof in this thread that our little green dragons can really really 'learn' and evolve!

A.:)

[pleoguy101]

hmm.. well. Rocky seems to have taking a liking to under my bed, he sticks his head under, and wont move, sometimes he crawls under and wont leave!! every time he's in my room, he does this!! and everytime I leave a certain green little caterpillar toy on the floor, Rocky falls asleep on it, nothing else, if I dont leave it out, he'll sleep anywhere:) I'm not taking sides or arguing, just saying a few examples of some sort of learing, or evolvion, I'm sure UGOBE wouldn't lie to us.. remember one of their laws? "all UGOBE lifeforms must evolve and learn over time" and I'm sure pleos evolution and learning capacity will grow with each upgrade:)

[andreivc]

Good points, Peoguy101! It does look like Rocky is exhibiting some learned behaviours there! After all, how can you explain it? Surely, Ugobe didn't have access to Pleoguy101's little green insect while programing that behaviour into his Pleo :D

[andreivc]

How would you explain that? And yes, please keep those testimonials coming - tell us what your pleo has learned - see questions in post #1.

[fancyfont]

When a Pleo is turned off he becomes just a toy robot. Well, maybe, a very cute little sleepy robot:). But, once on, he becomes a fabulous illusion, Denodan, if you prefer to call Pleos behavior that. I prefer to call it developing a personality.:D

Having an imagination is a wonderful thing.:) To some, Pleo is a very real companion,especially, if you don't or can't have a living, breathing animal. I am fortunate to have both. I must admit when my little Pleo is on, he is more unpredictable than my two little dogs.;) I love this feature:):). It doesn't matter to me if it is just an illusion.

[sandro]

Sandro, I would like to disagree with you there. Pleo may not use special hardware, but his overall hardware architecture is in line with cognitive computing principles that try to emulate organic life - he has several processors that specialize on different functions. In comparison, look at AIBO - that works like our regular PC - one processor, one memore, etc. Pleo is closer to cognitive computing that yours or mine PC, or AIBO.

But how would you create a parallel between having multiple processors to organic life? We only have one brain. Cognitive computing isn't about assigning different hardware to different functions, but re-creating the functional structure of the brain. My vintage 1991 laser printer has 3 processors (1 for postscript, 1 to control the paper path, and a third to manage all the other functions) that doesn't make it a cognitive computing platform. The thing is that Pleo is still running the same computer languages, hardware, power systems, etc as most off-the-shelf computing hardware.

But why is it important that Pleo be a cognitive computing agent? It's not like there's a huge success using CC over other approaches. I'm not even sure that the CC research can achieve some of the results more mundanely named parts of AI achieve.

[sarusani]

Hi denodan :) Just a little question. Let's say pleo had only two things he could do
1. singing
2. screaming

Let's say every time he sings you pet him and everytime he screams you just ignore him. After a wile his parameter will tell him: When I sing I get petted...
The result of this is that he will sing more because he knows that he will be petted more. Isn't that learning? II know you can not teach him a new song, but I still would say that's a basic form of learning...

Greetings
Sarusani

[lights]

Question 1. scale 8, more jolly.

Question 2. scale 4/5 eating, growling, needing alot of affection.

Question 3. Depressed feels like he's lost someone.

He doesn't do this learning thing.

[andreivc]

Thank you fancyfont, sarusani, lights for your comments! fancyfont - agreed, having good imagination makes life more fun! Sarusani, you are right! Hey, if we have that batteries have memory, then Pleo, a more sophisticated creature than a battery, can surely 'learn'! I am not a lawyer, but I think Ugobe is safe claiming that Pleo learns. I guess it's an issue of expectations--what kind of expectations do people have when they see a sophisticated product like Pleo, that claims to be the first and only of it's kind? They start assigning all kinds of super-robot qualities to it cuz they want to have it so badly... and yeah, learning sounds so cool.
lights - thanks for taking the poll! Were you expecting it to learn? We all are expecting it to learn! :D

Let's hear some more from our Pleo owners! Tell us about your experiences with Pleo learning and how deep you think that feature goes...?

[Toyguy]

Even considering UGOBE's disclaimer that Pleo does not learn in the original releases, I think there is still some bit of it in there. As an example, I have been doing very short periods with my Pleo thanks to the battery charger being defective. Even with that, when I left him alone last night while sitting down to eat, it was very clear that he became "lonely." His body posture changed, his vocalizations slowed and went down in pitch. He "knew' I was not paying him the same amount of attention I had been. Though not learning in the classic sense, perhaps, I think it meets a limited definition more closely analogous to the state of today's AI.

I'm perfectly happy with Pleo, and I haven't even updated to 1.1 yet :)

[fancyfont]

I've just recently updated to 1.0.2. and was a little dissapointed at first. Peeky was wanting to sleep all of the time. He also kept looking to his right side and almost tipped over a few times. never did that before. Finally, put him up on the coffee table and played with him. Tried his ta da trick and he fell over the first three times. I thought what did I do to this little guy? He's the tada champ, after all!;) After repeating the trick a few times, he mastered it once again. What a difference that made. He did try to catch a couple of brief naps, but I wouldn't let him for very long.
He started making all kinds of new sounds. Ones I haven't heard him do before. I'm not sure if I will update him to 1.1 until I see all of the neat features of this 1.0.2. Sure having a lot of fun with exploring his older update with him.:) This poll is great! It gives me information on how soon I will take the plunge and do the 1.1.:)

[ALF_owner]

I own a Pleo and have read the posts in this forum but never felt the urge to post till now. I’ve seen a few posts about Pleo (learning or illusion?).
I’ve been involved in robotics for over 10years and I’m just waiting for the PDK pack. Ugobe has taken robotics in a new and wonderful direction.
Is it learning or illusion?
I believe that is a question better left unanswered, but rather enjoyed.

I have a parrot (that is very much alive) I love and enjoy his company. He learned or has been programmed to talk through endless hours of interaction.
I prefer to believe he has learned but many argue that it is nothing more then programmed response to the interaction.
Both Pleo and my parrot learn or are programmed, one through hours of interaction the other through Pawn script.

Learning or illusion, which do I believe?
Well I prefer to believe that unicorns and angels are real. So I guess my point here is, does it matter?
I applaud Ugobe for this robot, pet, interactive life form, or whatever you prefer to call it.
Learning or illusion? You decide and believe in whatever makes you happy.
After all isn’t this what Pleo is all about?

[denodan]

Hi denodan :) Just a little question. Let's say pleo had only two things he could do
1. singing
2. screaming

Let's say every time he sings you pet him and everytime he screams you just ignore him. After a wile his parameter will tell him: When I sing I get petted...
The result of this is that he will sing more because he knows that he will be petted more. Isn't that learning? II know you can not teach him a new song, but I still would say that's a basic form of learning...

Greetings
Sarusani

I have tried this as a way to prove some type of learning, when he sings I pat him. But so far have not proved he is learning. He has not learned to sing when I pat him at all yet anyway. I find the new update better, the best yet, but as for learning am not convenced of this. My atempts to prove this so far have failed. He does not supprise me as he is following scripts that are programmed afterall.

Yes if he sang everytime I patted him it is a form of learning, but remember this is still a programmed response. It is programmed afterall and only following a program. it is easy to have say a counter. Pat so many times he sings, don't touch me after so long he is sad etc. It is still really only following a program. Furby gave the illusion after awhile he talked better and some english. It is a form of learning, but at the same time software designed to mimick learning, and not hard to program to give the illiusion of learning. Afterall you can use counters and time to mimick this, and this is what UGBOE will be doing if he accually learns, following counters, and timers. Afterall they did a good job in giving us the best robot, and mimicking a real moving live creature, and they have also given us a robot that mimicks learning and the illusion of learning. I commend UGOBE for that and are not deceiving us or anything like that and are not saying e does not learn, he does in a way, but is still only really a program and following it to the letter. I tried the reward system, as I said if he sings I pat him and the reward system has failed so far.

As for evolving. I don't think Pleo would have enough memory for this. Look at our brain? we are also evolving and personaility changes and the ammount of memory to do this is hugh.

All Pleo will be doing is changing numbers and figures nothing more so give you the illuison of some type of learning.

This has been argued with computer programmers for yes can computer learn. The answer is no, they are only following their program, mimicking learning.

I have not seen Pleo do anything out of his prgramming. I have read where he eats more, so does mine, but that is programmed. One person says he sleeps with his green toy under the bed. But it's to dark under there for his camera to see anything, and the toy not there he sleeps anywhere. It is claimed with the toy there he goes there and sleeps with it, there is no way Pleo can see it or even recognise what it is.

Pleo can see objects, but cannot know what they are. tests done on You tube show Pleo does not recognise colour. But there is a chance now he may recognise green. I don't think anyone has tested it out, but to see his leaf I know it needs to be pretty bright before he sees it.

I have no tried to see if Pleo sees green, but he certianly cannot tell if a toy is a toy etc. If anything he may respond to the colour green.

So it is easy for UGOBEO to give us the illusion of learning, but still only following a program. There is nothing wrong with this at all and is clever programming and are giving us what they say they are a Pleo evolving and learning. But real learning is when Pleo will supprise us and do things outside his programming and I see nothing of my Pleo to suggest this.

people have great imaginations and that is great, but Pleo does not learn as much as people would like him to. He is following timers and numbers, and pressures of his sensors.

When he detects objects I see no way he learns he gets stuck if going there or not. he still will get stuck and does always detect objects. If he learned would he not know if he goes there he woul get stuck?

he keeps getting stuck in the same places. If he where trully learning he would know would he not?

So cannot funtion outside his program, unlike us or even a pet like a dog of cat or any animal for that matter, they can supprise us, but not Pleo. He cannot function outside his program.

So I have seen nothing in my Pleo that suggests he is learning at all, or even evloving. The only evolving he has been doing so far is the updates. Every update is an evolved form of his programming, so UGOBE is keeping their promise Pleo evolves. He does with each update, just like a computer program evolves and gets better in time due to updates and newer versions. This is how UGOBE is giving us the evolving pleo and are correct in a way they say he evolves over time. Yes but only due to new firmware, not his ablitiy to keep evolving.

[fancyfont]

Beautiful response! I so agree,:)

[Aussie_Pleo]

Gee Alf_owner (I was more of an ET guy lol) your perspective is really nice. Neon arrived in my home yesterday and I have only had the first charge cycle with him so we still haven't fully discovered each other yet. But the way you put it is probably the way I am approaching our time together.

[denodan]

I own a Pleo and have read the posts in this forum but never felt the urge to post till now. I’ve seen a few posts about Pleo (learning or illusion?).
I’ve been involved in robotics for over 10years and I’m just waiting for the PDK pack. Ugobe has taken robotics in a new and wonderful direction.
Is it learning or illusion?
I believe that is a question better left unanswered, but rather enjoyed.

I have a parrot (that is very much alive) I love and enjoy his company. He learned or has been programmed to talk through endless hours of interaction.
I prefer to believe he has learned but many argue that it is nothing more then programmed response to the interaction.
Both Pleo and my parrot learn or are programmed, one through hours of interaction the other through Pawn script.

Learning or illusion, which do I believe?
Well I prefer to believe that unicorns and angels are real. So I guess my point here is, does it matter?
I applaud Ugobe for this robot, pet, interactive life form, or whatever you prefer to call it.
Learning or illusion? You decide and believe in whatever makes you happy.
After all isn’t this what Pleo is all about?

Well put I do see it as an illusion, as Pleo does nothing outside his scripted programming, so no supprises, whereas your parrot is not programed and will do things that supprise you. As long as people treat it as an illusion and remember this, then he is fun and lovable all the same. Pleo is so much more advanced than many toy robots around. I have some from WOWEE and none come close to matching Pleo. I even have I-Cybie, which I feel does so much better at object detection then Pleo, but still not even a close match.

All the other robots seem primative to Pleo.

[ALF_owner]

Denodan I agree to disagree with you if programing is done right it can be full of surprises.
Pleo sings when he feels like it, explodes when he decides to, or refuses to walk away if he wants attention.
Now my parrot, well actually he very predictable I come home and he says "welcome home", the phone rings and he yells "get the phone" , the door bell rings and he yells "come on in"....I really wish he wouldn't do that!
The only point I was really trying to make is that is be it a pet, a toy. a robot, or a life form....If you enjoy the interaction does it really matter how or why it does it?
All I can say is I believe Pleo is as real as you want him to be.

[Caaahl]

Beautifully said. :)

I stay with my opinion: animals are the most sophisticated robots so far. :) But our robotic friends will match up sooner or later.

[andreivc]

ALF_owner - we don't appreciate your 'common ground' kind of approach here, we prefer some controversy, and passionate debates, and tears and laughter, and ... I am just kidding. what you are saying makes total sense.

[ALF_owner]

Caaahl....I couldn't agree with you more. ;)

[denodan]

You are correct, but for me he is still a programmed robot. The illusion is well done and no other robot in this price range, or even double, or triple the price of Pleo would not interact like Pleo does. He can be as real as you make him to be in your mind, as long as you don't forget he is really just a robot.

I love my pleo, don't get me wrong and the interaction is great. He is evolving through new software and cannot wait to see what comes next. I think the SDK will be the best yet to come.

As for your parrot he will and may be predicable when he talks,but his movments and personailty will not be. You cannot compare a living creaturge to any robot. A creature has real learning ablitiy whereas Pleo's is a great illusion and simple learning at that, but still great all the same and can think anyone who cannot have a real pet then Pleo is just about as good.

Pleo would have to be the best interactive robot.

[lights]

My reason for buying pleo was because i am allergic to all animals. When i saw pleo i was delighted that at last i could have some sort of pet where i could recieve affection instead of feeling sad everytime people talked about their pets. I was also impressed to hear that pleo was an evolving pet that could learn which made up for buying him at that price. I do really like my pleo however i know sometimes that i'm talking to a brick wall as he can't understand or acknowledge me, he can only listen to touch, which is a big shame for such a high tec pet.
Don't get me wrong i do really like my pleo, but whats