Project Aiko
Would you like to react to this message? Create an account in a few clicks or log in to continue.
Search
 
 

Display results as :
 


Rechercher Advanced Search

Latest topics
» introducing the cyberpunk software design pattern
Who want to see Sumomo From Chobits made? EmptyTue Aug 15, 2023 1:57 am by japlu

» LG book updates
Who want to see Sumomo From Chobits made? EmptyTue Mar 21, 2023 5:54 pm by japlu

» LivinGrimoire swift version added
Who want to see Sumomo From Chobits made? EmptySat Jul 23, 2022 9:55 pm by japlu

» the living grimoire is now available in python as well
Who want to see Sumomo From Chobits made? EmptyTue May 03, 2022 9:03 am by japlu

» livingrimoire auxiliary modules
Who want to see Sumomo From Chobits made? EmptyMon Apr 25, 2022 4:51 pm by japlu

» [Python] Thinkable class
Who want to see Sumomo From Chobits made? EmptyTue Apr 05, 2022 7:11 pm by japlu

» owly check this out
Who want to see Sumomo From Chobits made? EmptyTue Mar 22, 2022 5:03 pm by japlu

» the living grimoire AGI
Who want to see Sumomo From Chobits made? EmptyTue Mar 22, 2022 4:35 pm by japlu

» Cherry Blossom Technology Launch!!!
Who want to see Sumomo From Chobits made? EmptySat Mar 25, 2017 10:42 am by BlueSodaX

March 2024
MonTueWedThuFriSatSun
    123
45678910
11121314151617
18192021222324
25262728293031

Calendar Calendar


Who want to see Sumomo From Chobits made?

+17
ico
mikeccuk2005
Mel Addison
PJ
Discoman
SixStringSamurai
Balmung
Artilects rule!
kalaith
poto
Kal-el
WulfCry
captquazar
andrew motz
Reaper
Konrad13
Aiko
21 posters

Page 1 of 3 1, 2, 3  Next

Go down

Who want to see Sumomo From Chobits made? Empty Who want to see Sumomo From Chobits made?

Post  Aiko Sat Jan 10, 2009 3:07 am

Hi,

I have noticed that a lot of you are talking about wanting to make your own Sumomo from Chobit.
If you want see one in real live, just for the heck of it....

Buy me a Sumomo Doll from Japan, and donate about $500. And you will see the world first Sumomo walking around in 4-6 weeks.
$500 will buy me small motors, microvision, and micro controller and simple mainboard, I will loaded with BRAINS Lite version in it.

I think this will be my pet projects.... will make it open source.
Will send the sumomo to people in this forum, and have you guy play with it.
You guys can play with her, upgrade her.... don't really care....
Hopefully, this will encourage more people into robotic....


Very Happy


Last edited by Aiko on Mon Jan 12, 2009 4:09 am; edited 1 time in total
Aiko
Aiko
Admin

Posts : 335
Reputation : 2
Join date : 2008-08-22

http://projectaiko.com

Back to top Go down

Who want to see Sumomo From Chobits made? Empty Re: Who want to see Sumomo From Chobits made?

Post  Konrad13 Sat Jan 10, 2009 6:15 am

Sumomo is a little larger in size than an iPhone.

Also, I would like to see a Sumomo "phone," but I feel we should focus more on Aiko. Once she's up and walking around, then I'd love to see a Sumomo up and walking...

Konrad13

Posts : 70
Reputation : 0
Join date : 2008-12-31
Age : 34
Location : Missouri

Back to top Go down

Who want to see Sumomo From Chobits made? Empty Re: Who want to see Sumomo From Chobits made?

Post  Reaper Sat Jan 10, 2009 6:29 am

A Sumomo project would be great.

I say go for it if you can manage such a small package. I imagine there's not much to be done on Aiko in the meantime except A.I. work and it may help validate your Aiko work with the skeptics and bring in more funding. Besides, she's too cute to say no. Very Happy

edit: found this website:http://www.amazon.com/Chobits-Talking-Sumomo-Action-Figure/dp/B000GPX06A
Apparently she's 7 inches tall.
Reaper
Reaper

Posts : 12
Reputation : 0
Join date : 2008-12-11

Back to top Go down

Who want to see Sumomo From Chobits made? Empty Re: Who want to see Sumomo From Chobits made?

Post  Aiko Sat Jan 10, 2009 7:26 am

It's will be hard not seeing the actually doll first.
But from what I see from the picture...
We can make the head moves and arms...
(not elbow.... too small to fit a motor)

The leg can move.... so she can do a little excerise like in the anime.

Can follow your faces.... and only simple object...
Will understand about 15 sentenses... with 10 movements..



Le


Last edited by Aiko on Mon Jan 12, 2009 4:10 am; edited 2 times in total
Aiko
Aiko
Admin

Posts : 335
Reputation : 2
Join date : 2008-08-22

http://projectaiko.com

Back to top Go down

Who want to see Sumomo From Chobits made? Empty ur a genius

Post  andrew motz Sat Jan 10, 2009 8:14 am

7 inches that 17.78 centimeters
and if u can remeber in the series at the end sumomo says she is 18 cm tall

oh and im so angry that i cant find that video with the motherboard smaller than a mobile phone

andrew motz

Posts : 13
Reputation : 0
Join date : 2008-12-25
Age : 30
Location : arad(romania)

http://www.wildcatsreality.piczo.com

Back to top Go down

Who want to see Sumomo From Chobits made? Empty Re: Who want to see Sumomo From Chobits made?

Post  Aiko Sat Jan 10, 2009 8:18 am

"oh and im so angry that i cant find that video with the motherboard smaller than a mobile phone"

I think I know what motherboard you are talking about it's VIA..... it's suppose to be 3cm by 5cm
It was suppose to be release in 2009 or 2010.... if I remember it correctly...
Aiko
Aiko
Admin

Posts : 335
Reputation : 2
Join date : 2008-08-22

http://projectaiko.com

Back to top Go down

Who want to see Sumomo From Chobits made? Empty found em

Post  andrew motz Sat Jan 10, 2009 1:07 pm

yep found it
and look it comes in package very small chipset+very small CPU
https://www.youtube.com/watch?v=2l_ox-Na12Y

+i wasnt taking about VIA NANO i wqas talking about intel atom
https://www.youtube.com/watch?v=3MklPBVPSp0&feature=related
but its clear VIA NANO dose way better job


P.S. u were searchin for a other therm for aiko
why dont u call her CPU(Cute Personal Umanoid) lol!

andrew motz

Posts : 13
Reputation : 0
Join date : 2008-12-25
Age : 30
Location : arad(romania)

http://www.wildcatsreality.piczo.com

Back to top Go down

Who want to see Sumomo From Chobits made? Empty building Sumomo.

Post  captquazar Mon Jan 12, 2009 9:16 pm

Reaper it’s great you found that action figure. Am surprised it is still for sale. Chobits stopped being a big deal around 2005, but then the DVDs still sell well. As a basis to build the robot sumomo it could only be used to take measurements from. The head is rigid, not flexible, the eyes are painted on. The neck is too thin to support the weight of the head. (This defect is present in the anime version too) The body will have to be a bit thicker to hold the batteries and main board and motors. True in Chobits Sumomo was only 18cm tall. This body size just may not be practical to get all the stuff inside. Andrew it is fantastic you found that tiny main board! It might help with some of Aiko's subsystems too. A big design issue is how much can we duplicate the behavior and capability of the Anime Sumomo. She was able to jump into her owner’s hands, and dance. She could use her own hands to. She could recognize objects and people. That is very advanced artificial intelligence with physical abilities too. For all the vision and dedication and effort that went into the Pleo toy robot, the results are rather disappointing. It just does not do enough. As we cut features from Sumomo because it just can’t be done now for technical reasons; there comes a point when building the result is not worth the effort. I don’t yet know where or how many of those compromises have to be made. I am working on a list of specifications as detailed as I can make it for discussion. The initial goal would be to build a Sumomo every bit as capable as shown in the Chobits series. This would be a major achievement in AI and systems intergration. Extremely difficult with current technology.

captquazar

Posts : 17
Reputation : 0
Join date : 2008-12-19

Back to top Go down

Who want to see Sumomo From Chobits made? Empty Impossible is nothing sumomo like mobile robot can be done.

Post  WulfCry Tue Jan 13, 2009 8:01 pm

Sorry man I have to disagree with you on that impossible is nothing.
And I rather give Le Trung the benefits of the doubt.

The reason simple its easy to reject the similar factor of build,
with what we know to be already out there, To what should be.
If Steve Jobs of apple looked to the musicplayer market with the same
Eye the rest of the market did then the Ipod would not be possible.
The technology is out there so finding the pieces that really fit
well together needs insight from various people with a niche specialty
building small things that could do it.
Or in the case of Aiko a person like Le Trung with a vision that say yes I can.

Aiko is a big prototype that is build in less then 3 months.
Scourge the net and you find research papers describing applied ,
methods to certain automaton problems in moving , learning and such
Technically its not impossible then coping with factors what could limit
small based androids from performing effectively.

The cleo very impressive but lacks innovation deu to mass manufacturing
standard components to construct it and anything else.
Lets compare list of pro's and con's to consider the fact.
Knowing that some of the thing a sumomo minaiko could not do,
Such as jumping , climbing. Walk RUN and Sit can be in the range.
As for I will bet my money on what Le Trung can do.

P.s Andrew Motz UMANOID sounds fun its also a nice work in progress
Found the meaning of Uma on google and yahoo. Means horse in japanese
and Uma
Girl name, origin: Sanskrit
meaning: flax; luminous, tranquillity.
AF ancronym finder. Wink

WulfCry

Posts : 109
Reputation : 0
Join date : 2008-12-10
Location : Netherland , Rotterdam

Back to top Go down

Who want to see Sumomo From Chobits made? Empty Sumomo design parameters

Post  captquazar Tue Jan 13, 2009 8:31 pm

Sumomo design parameters

We will be discussing two versions. Version one will exist as an internet avatar. Version two will have a doll body as seen in the Chobits TV series.
(This is a chubric kind of doll 18 to 26 Cm in height with a large head in proportion to the rest of its body. The rest of the body has standard human proportions) The Avatar version being virtual, will have a subset of the doll’s capabilities. For both versions a web cam, microphone, and internet connection are required. For the doll, these would be built in. While connected via Wi-Fi or Cell internet; the doll will have all of the Avatar’s functions. The main purpose of both versions will be to act as a personal assistant for practical requests; and as comic companion for whimsical requests. Lets give due credit to the CLAMP Inc. ladies for knowing how to design a compelling character. Why does Sumomo have a big head? It is so we can see her facial expressions. Most of her cuteness comes from her bubbly hi energy personality manifested by her expressiveness and body language. Although least practical, it is her most important capability psychologicly. In other words the AI having interpersonal skills is vital to a successful implementation. Persocoms are ‘socializing companions’, not an appliance. Sumomo’s personality is what makes her fun and cool.

Features common to both versions:
1 Voice command as primary input and Speech as output.
2 Search capabilities (posed as questions) such as:
Contact information.
Calendar and appointment information. (PDA)
Current weather report.
Location / map queries.
Wiki questions (what is a xxxx)
News queries optimized per user preferences.
(User preferences are memorized via questionnaire)
Find PC files.
Via hypertext the AI can continue to provide additional details on a subject; but cannot interpret the results. (No AI can currently do that)
3 Send and receive email or instant messages, forward messages, filter same.
4 Take dictation and act as voice command interpreter for preferred word processor.
5 Telephony functions. Make and receive calls. Screen calls, e.g. “Hideki is currently not available, would you like me to record a message?”
6 Face recognition and face tracking.
7 Real time emotion analysis via either facial expression analysis or audio stream analysis, or both. (For the bot’s acting performance and rapport with the user.) The AI must be aware of the user’s emotional state to establish resonance with him/her. This is a psychological imperative to establishing a life like personality and avoiding the uncanny valley.
8 Answer questions about itself and about history with the user.
The character will have a KB of ‘life experiences’, stories, and jokes to tell. Remember recent events; esp. requests.
9 Be able to ‘act’ respond emotionally to provocative remarks or ‘goal’ denial / achievement. (Note: 7 & 10 are required)
10 AI will have performance goals, is happy when used, sad when ignored. Pleased when user is pleased, likes praise ect….

Doll version.
11) Be able to walk, maintain balance.
12) Have eye hand coordination.
13) Be able to pick up small objects less than 2 pounds.
14) Be able to recognize the 250 most common household items (fetch) Understand which objects are too large to move. Learn new objects defined by user. Or report location of object is currently unknown.
15) Be able to recognize and follow it’s owner. This includes being able to maintain eye contact. Recognize family vs. visitors.
16) ‘Acting’ ability will include body language and an expressive face.
17) Be able to avoid or defend itself from the family pet.
18) Report internal status such as battery level and quality of internet connection.

Obstacles to implementation.
Most of the necessary technology exists now in commercial versions or at least exists in the lab. About half dozen technologies needed are owned exclusively by different companies. Their cooperation would have to be secured. Some software / algorithms have never been integrated before. A few AI performance capabilities are only now being tested in the lab; are very leading edge. Of course the 900 pound gorilla is market potential vs. development cost and unit production costs. The internet avatar has the minimum development cost and least use of new technology. The doll has significant challenges. Areas of difficulty are itemized below for both versions.
1 The best speech to text, text to speech software is owned by Nuance, IBM, and Microsoft. Accuracy and speed are now satisfactory.

2 Google has the best search tech, and has integrated voice command already. Many software choices in this area.

3 4 and 5 have lots of software choices. Voice command would need to be integrated together with the other features. Seamless switching between features is a major design goal; regardless that many features are historically separate applications.

6 Face recognition and tracking do exist in a few commercial applications and many lab versions. Very leading edge, expensive to license. Facial expression analysis currently only exists in the lab that I know of.

7 Emotion analysis by audio stream or face expres​sion(or both) is still mostly in the lab. (Several companies now provide ‘tone of voice’ analysis for auto-attendant phone systems) And it has not yet been applied to; or integrated with emotion simulation via facial expression, tone of voice output and body language. Especially not all three together. I am aware that several software development tools are being developed for this purpose, but I am not aware a finished product exists yet.

8 An advanced chatbot AI will need to include contextual awareness of emotions, cause and effect; and have internalized goals. For the doll it will also require spatial situational awareness. The knowledge base domain will need to include these objects and event possibilities.

9 & 10 A successful product has to consider a host of psychological parameters to create an ‘actorbot’. It will need a palate of several dozen emotional states supported by dozens of facial expressions, body language gestures, and tone of voice used in the correct combinations and timing. This area will require some significant R&D including the interaction of goal behavior with emotional states.
Various ‘goal behavior’ simulations have been written for graduate projects, but never integrated into a larger AI that I know of.

11, 12 & 13 exist as isolated lab prototypes focused on specific ‘proof of concept’ design issues. No integration yet. Note significant processor power required. (Honda’s Asimov accepts objects placed in it’s unmoving hand, it has never picked up a moving object) Surprisingly eye hand coordination and the ability to pick up an object never seen before have turned out to be very difficult AI problems each. After years of work, software exists to do both in different labs. Getting the cooperation of all these different universities will not be easy.

14 & 15 Object recognition is a major area of AI research. There has been some recent success in major universities. Not sure if it can be applied outside the lab yet. Decades of work have gone into it. It won’t be cheap to license. Note the autos in the DARPA challenge contests carried hundreds of pounds of computer equipment to do the real time visual processing. Visual processing will require several dedicated CPUs. Will probably have to be inside the doll; WiFi or cell broadband connection probably not fast enough. This is a major R & D challenge itself.

16 All the issues of 7, 8, 9 and 10 will need to be solved as a single AI engine. Several disparate branches of AI research will have to be brought together as never before. Parts of the problems have been solved as graduate studies. A major effort here also.

17 Game theory can be applied here together with goal behavior.
Actually this part of the AI is not difficult. Lots of published algorithms available.

Well ladies and gentlemen, what mistakes have I made? What have I left out? What do you think the development priorities should be? The personality simulator needs hundreds of details I’m sure. This document needs lots of footnoting, who has what technology in what form? I am working from memory, sorry I do not have time to re-research all the sources. I have never done any lurking in the open source software forums, what good stuff can be applied to the above requirements?

captquazar

Posts : 17
Reputation : 0
Join date : 2008-12-19

Back to top Go down

Who want to see Sumomo From Chobits made? Empty Re: Who want to see Sumomo From Chobits made?

Post  Kal-el Tue Jan 13, 2009 9:24 pm

captquazar wrote:
Location / map queries.
Wiki questions (what is a xxxx)
News queries optimized per user preferences.
(User preferences are memorized via questionnaire)

The wiki search is just brilliant, any interactive robot has to have that capability in its software.

Besides what you already said these are the capabilities I think Sumomo should be packing.

-a integrated GPS in it so that you can use her as GPS when you travel.

-a integrated powerful small LED(light diode) like those in mobile phones so that you can use her as a flashlight in your room or outside.

-be able to tell the time so you can use her as watch.

-be able to play MP3 so that you can use her as an MP3 player or Ipod, with the help of a small but powerful speaker like those in mobile phones and also a output jack where you can connect you headphones if you don't want to listen to the speaker.

-be able to take pictures and films with her integrated camera so that you can use her as a camera and the option to record audio without video with a very sensitive microphone.

-be able to put her in motion detection mode and when her camera detects motion above a certain predetermined level she should take picture of it so that you can use her as security or spying or etc.

-be able to keep schedule so that when you ask her what do I have programmed for 8 o clock she answers a dinner with a chick so that you don't forget or put her to remind you without you asking etc just like a secretary.

-be able to tell the exact temperature in the room or outside,with the help of a temperature diode integrated in its motherboard.

It's very simple what her capabilities should be, look at all the electronic mobile devices people use everyday and integrate their capabilities in Sumomo and you have a best seller.

Kal-el

Posts : 97
Reputation : 0
Join date : 2008-08-31

Back to top Go down

Who want to see Sumomo From Chobits made? Empty Re: Who want to see Sumomo From Chobits made?

Post  Reaper Wed Jan 14, 2009 12:48 am

While all this stuff you guys are talking about would be neat to have, if it is even feasable, it is way outside the initial project parameters:
Aiko wrote:It's will be hard not seeing the actually doll first.
But from what I see from the picture...
We can make the head moves and arms...
(not elbow.... too small to fit a motor)

The leg can move.... so she can do a little excerise like in the anime.

Can follow your faces.... and only simple object...
Will understand about 15 sentenses... with 10 movements..
Le
I believe it will be challenging enough just to do that.
Reaper
Reaper

Posts : 12
Reputation : 0
Join date : 2008-12-11

Back to top Go down

Who want to see Sumomo From Chobits made? Empty Re: Who want to see Sumomo From Chobits made?

Post  Konrad13 Wed Jan 14, 2009 3:03 am

Something else that should be considered is the capability for add-on software/hardware, like thumb-drives and such. This is somewhat seen in the iPod hookup capbility, but should definitely be expanded. Another thing to consider is the ability to hook it up to a computer via USB cable(s), and download her memory day-by-day, and allowing it's capabilities to expand while hooked up to a computer. While unplugged, it is little more than a cute little robot GPS/phone/iPod, but get her jacked into your home computer, and her abilities can be increased by using the hardware and whatnot on the computer. An idea is for a near full "Aiko program" to be installed onto one's computer, and while connected, the program runs greatly increasing her repertoire of abilities.

Another cool idea is for the use of external portable hard-drives, like my modest 350 Gig one (though 1Tb ones will be available cheap and small before too long) to carry a simple os, but also the Aiko program (and act as a charging station for the battery, as well as a signal booster for phone/internet/GPS), thus allowing a "full" personality everywhere.

This will, no doubt, be expensive at first. I'd say with all of these features and a 350 Gig hard-drive (that could be packaged with it), a modest price of roughly $5000 would be in order. But, as time comes, within a mere two years, the price could be as low as $500! So, I suggest this model of including a software package (definitely get in contact with those labs and companies, and promise that if this takes off they'll get a percentage of the sales), 350 Gig- 1 Tb drive/charger/booster, as well as a carrying case (don't want to break her, do ya?).

What do you all think?

Konrad13

Posts : 70
Reputation : 0
Join date : 2008-12-31
Age : 34
Location : Missouri

Back to top Go down

Who want to see Sumomo From Chobits made? Empty Re: Who want to see Sumomo From Chobits made?

Post  Konrad13 Wed Jan 14, 2009 3:09 am

Also, if in the event I win the Powerball (lottery game which is now up to, I believe $140 million) I will finance this project, so long as I get my own little Sumomo and a custom Aikocom (see: https://projectaiko.forumotion.com/project-aiko-hardware-software-f1/new-terms-for-robot-or-special-term-when-refering-to-aiko-t68.htm). But yes, full funding.

Konrad13

Posts : 70
Reputation : 0
Join date : 2008-12-31
Age : 34
Location : Missouri

Back to top Go down

Who want to see Sumomo From Chobits made? Empty Re: Who want to see Sumomo From Chobits made?

Post  poto Wed Jan 14, 2009 4:10 am

I've been giving this a lot of thought, and I think this is what you should do.

Start off by making a software only version that includes some useful personal assistant features like:
task lists (with reminders)
exercise programs (yoga would be nice in the morning)
dictation
voice search function for google, wikipedia and the weather
and some cheerful conversation to help brighten my day or calm me down when I get mad at something

It would be acceptable if you didn't launch it with all those features, as long as they were added sometime later.

Make it a monthly subscription service and open an affiliate program for it. You can open an affiliate program at places like paydotcom.com fairly cheap. That would probably generate enough extra revenue to fund Aiko as well as supporting continued software development. You already have an amazing piece of software, IMHO it just needs a bit of tweaking, repackaging and polishing and you could make a good bit of money from it.

I would be willing to pay for a virtual Sumomo, if I could get some practical use out of it. I work on my computer all day, so something like that I might even be able to write off as a business expense Very Happy
poto
poto

Posts : 11
Reputation : 0
Join date : 2008-09-29

http://poto.me

Back to top Go down

Who want to see Sumomo From Chobits made? Empty Re: Who want to see Sumomo From Chobits made?

Post  Aiko Wed Jan 14, 2009 6:09 am

Hmm...

I have no intention making a virtual Sumomo.
I will make a physical Sumomo.
Since it's a copyright by Clamps, I will only make ONE as fans to fans base.
I will give to one of you to in charge.
Play with it, updgrade it.... pass a long with other member of the forum.

I have no intention of keeping it for myself... or making money from it.
I just want people to experience human interaction with mini robot in this case Sumomo.
The only time it will be send back to me is for major repair or update...


p.s. Having a real physical Sumomo is a millions time better than having a virtual Sumomo.
My inspiration come from building a robot with physical body... without a body.... no inspiration, mean No work is done.

I hope that make sense.
Very Happy
Aiko
Aiko
Admin

Posts : 335
Reputation : 2
Join date : 2008-08-22

http://projectaiko.com

Back to top Go down

Who want to see Sumomo From Chobits made? Empty Re: Who want to see Sumomo From Chobits made?

Post  poto Wed Jan 14, 2009 7:03 pm

I didn't mean make an exact copy of Sumomo from Chobits for a virtual Sumomo. I meant make something similar (that doesn't infringe on any copyrights) and market it as a virtual personal assistant/companion to generate revenue.

Making one-off fan models is very interesting, and I'll be looking forward to see whatever you create, but I was just trying to think of a way to better monetize what you've already created. Maybe that's just the marketer in me.

I fully understand your point about having a physical body to interact with. However, if you build a customer base with an interactive software product you could easily up-sell them to a hardware version later on. Rather than waiting for donations or investors, you could do everything with your own revenue.
poto
poto

Posts : 11
Reputation : 0
Join date : 2008-09-29

http://poto.me

Back to top Go down

Who want to see Sumomo From Chobits made? Empty R & D on Sumomo

Post  why-do-I Wed Jan 14, 2009 9:05 pm

On their web site the CLAMP ladies are more generous than most companies regarding derivative works / copyright. First they insist that you do not insult or degrade their work. It must be positive and complementary. Secondly you cannot do an exact trace or copy of their art. However you can duplicate (BY HAND) any character you want. Some of the fan art on the web is quite excellent and legal for use. As with all the toys, they are quite willing to license their stuff if it is of quality and builds their image.
Anyway Poto, I agree with you. The internet avatar tools are good and low cost. There are many chatbot engines to choose from. All the Personal Digital Assistant (including email, IM, chat, IP/phone, texting, searching, and scheduling) are all standard apps, dozens of software source code apps to choose from. It would take some work to integrate it all into a single app; but really no R & D at all. For both versions the Brains and AI chat software is on a local PC or even a shared internet server. This is to reduce unit cost and space limitations. Yes without an internet connection of some kind the doll becomes very dumb. Some default functions would need to be local, a limited voice command. You all seem to be overlooking the really really hard part. The ‘actorbot’ emotion simulator and personality. This is some of the most bleeding edge AI research. Very little of it exists as software. This is what would make Sumomo jaw dropping cool, a killer app. Not all the net PDA stuff. A Sumomo doll that can barely walk around and has NO personality would be a show stopper. It would be ‘So what, another clunky robot toy’. YOU MUST MANAGAGE USER EXPECTATIONS. So to speak, people don’t want half a glass of beer, they want a full mug. They want an electric car that can drive just as far as a gas car without stopping, not need recharging after 60 miles. An avatar or doll robot without interpersonal skills would not be much of an improvement over what is out their already. It must be able to relate emotionally with its owner. Again, the CLAMP ladies have clearly shown in the Chobits maga and Anime what the ideal Persocom companion would be. One of those Kotocos or Sumomos would make the designer millions of $$$$. However a comprehensive voice command PDA app would be a good start, get rid of all the clutter of so many separate applications. A single user interface that does it all. Make things simpler / easier to use. That is what people want.

why-do-I
Guest


Back to top Go down

Who want to see Sumomo From Chobits made? Empty Sumomo performance examples

Post  why-do-I Wed Jan 14, 2009 10:13 pm

Sumomo takes a message:

why-do-I
Guest


Back to top Go down

Who want to see Sumomo From Chobits made? Empty performance examples

Post  why-do-I Wed Jan 14, 2009 10:20 pm

My last post, what happened to all the youtube urls? Geejs I spent 45 min finding good examples of Sumomo.
Anyway just search chobits Sumomo. OH and 'if Sumomo ran Vista'. It had me in tears.

why-do-I
Guest


Back to top Go down

Who want to see Sumomo From Chobits made? Empty Really

Post  kalaith Wed Jan 28, 2009 2:40 am

So, with the doll, do you have a specific doll in mind?

or a specific material the doll needs to be made from?

kalaith

Posts : 1
Reputation : 0
Join date : 2009-01-28

Back to top Go down

Who want to see Sumomo From Chobits made? Empty Re: Who want to see Sumomo From Chobits made?

Post  Aiko Wed Jan 28, 2009 5:32 am

Aiko
Aiko
Admin

Posts : 335
Reputation : 2
Join date : 2008-08-22

http://projectaiko.com

Back to top Go down

Who want to see Sumomo From Chobits made? Empty Re: Who want to see Sumomo From Chobits made?

Post  WulfCry Thu Jan 29, 2009 3:34 pm

? Neutral ? Will it work can the multiple points of articulation be hacked to be functional in free movements what would they mean with display base?.

WulfCry

Posts : 109
Reputation : 0
Join date : 2008-12-10
Location : Netherland , Rotterdam

Back to top Go down

Who want to see Sumomo From Chobits made? Empty Re: Who want to see Sumomo From Chobits made?

Post  Artilects rule! Sun Feb 08, 2009 9:02 pm

captquazar wrote:Sumomo design parameters

We will be discussing two versions. Version one will exist as an internet avatar. Version two will have a doll body as seen in the Chobits TV series.
(This is a chubric kind of doll 18 to 26 Cm in height with a large head in proportion to the rest of its body. The rest of the body has standard human proportions) The Avatar version being virtual, will have a subset of the doll’s capabilities. For both versions a web cam, microphone, and internet connection are required. For the doll, these would be built in. While connected via Wi-Fi or Cell internet; the doll will have all of the Avatar’s functions. The main purpose of both versions will be to act as a personal assistant for practical requests; and as comic companion for whimsical requests. Lets give due credit to the CLAMP Inc. ladies for knowing how to design a compelling character. Why does Sumomo have a big head? It is so we can see her facial expressions. Most of her cuteness comes from her bubbly hi energy personality manifested by her expressiveness and body language. Although least practical, it is her most important capability psychologicly. In other words the AI having interpersonal skills is vital to a successful implementation. Persocoms are ‘socializing companions’, not an appliance. Sumomo’s personality is what makes her fun and cool.

Features common to both versions:
1 Voice command as primary input and Speech as output.
2 Search capabilities (posed as questions) such as:
Contact information.
Calendar and appointment information. (PDA)
Current weather report.
Location / map queries.
Wiki questions (what is a xxxx)
News queries optimized per user preferences.
(User preferences are memorized via questionnaire)
Find PC files.
Via hypertext the AI can continue to provide additional details on a subject; but cannot interpret the results. (No AI can currently do that)
3 Send and receive email or instant messages, forward messages, filter same.
4 Take dictation and act as voice command interpreter for preferred word processor.
5 Telephony functions. Make and receive calls. Screen calls, e.g. “Hideki is currently not available, would you like me to record a message?”
6 Face recognition and face tracking.
7 Real time emotion analysis via either facial expression analysis or audio stream analysis, or both. (For the bot’s acting performance and rapport with the user.) The AI must be aware of the user’s emotional state to establish resonance with him/her. This is a psychological imperative to establishing a life like personality and avoiding the uncanny valley.
8 Answer questions about itself and about history with the user.
The character will have a KB of ‘life experiences’, stories, and jokes to tell. Remember recent events; esp. requests.
9 Be able to ‘act’ respond emotionally to provocative remarks or ‘goal’ denial / achievement. (Note: 7 & 10 are required)
10 AI will have performance goals, is happy when used, sad when ignored. Pleased when user is pleased, likes praise ect….

Doll version.
11) Be able to walk, maintain balance.
12) Have eye hand coordination.
13) Be able to pick up small objects less than 2 pounds.
14) Be able to recognize the 250 most common household items (fetch) Understand which objects are too large to move. Learn new objects defined by user. Or report location of object is currently unknown.
15) Be able to recognize and follow it’s owner. This includes being able to maintain eye contact. Recognize family vs. visitors.
16) ‘Acting’ ability will include body language and an expressive face.
17) Be able to avoid or defend itself from the family pet.
18) Report internal status such as battery level and quality of internet connection.

Obstacles to implementation.
Most of the necessary technology exists now in commercial versions or at least exists in the lab. About half dozen technologies needed are owned exclusively by different companies. Their cooperation would have to be secured. Some software / algorithms have never been integrated before. A few AI performance capabilities are only now being tested in the lab; are very leading edge. Of course the 900 pound gorilla is market potential vs. development cost and unit production costs. The internet avatar has the minimum development cost and least use of new technology. The doll has significant challenges. Areas of difficulty are itemized below for both versions.
1 The best speech to text, text to speech software is owned by Nuance, IBM, and Microsoft. Accuracy and speed are now satisfactory.

2 Google has the best search tech, and has integrated voice command already. Many software choices in this area.

3 4 and 5 have lots of software choices. Voice command would need to be integrated together with the other features. Seamless switching between features is a major design goal; regardless that many features are historically separate applications.

6 Face recognition and tracking do exist in a few commercial applications and many lab versions. Very leading edge, expensive to license. Facial expression analysis currently only exists in the lab that I know of.

7 Emotion analysis by audio stream or face expres​sion(or both) is still mostly in the lab. (Several companies now provide ‘tone of voice’ analysis for auto-attendant phone systems) And it has not yet been applied to; or integrated with emotion simulation via facial expression, tone of voice output and body language. Especially not all three together. I am aware that several software development tools are being developed for this purpose, but I am not aware a finished product exists yet.

8 An advanced chatbot AI will need to include contextual awareness of emotions, cause and effect; and have internalized goals. For the doll it will also require spatial situational awareness. The knowledge base domain will need to include these objects and event possibilities.

9 & 10 A successful product has to consider a host of psychological parameters to create an ‘actorbot’. It will need a palate of several dozen emotional states supported by dozens of facial expressions, body language gestures, and tone of voice used in the correct combinations and timing. This area will require some significant R&D including the interaction of goal behavior with emotional states.
Various ‘goal behavior’ simulations have been written for graduate projects, but never integrated into a larger AI that I know of.

11, 12 & 13 exist as isolated lab prototypes focused on specific ‘proof of concept’ design issues. No integration yet. Note significant processor power required. (Honda’s Asimov accepts objects placed in it’s unmoving hand, it has never picked up a moving object) Surprisingly eye hand coordination and the ability to pick up an object never seen before have turned out to be very difficult AI problems each. After years of work, software exists to do both in different labs. Getting the cooperation of all these different universities will not be easy.

14 & 15 Object recognition is a major area of AI research. There has been some recent success in major universities. Not sure if it can be applied outside the lab yet. Decades of work have gone into it. It won’t be cheap to license. Note the autos in the DARPA challenge contests carried hundreds of pounds of computer equipment to do the real time visual processing. Visual processing will require several dedicated CPUs. Will probably have to be inside the doll; WiFi or cell broadband connection probably not fast enough. This is a major R & D challenge itself.

16 All the issues of 7, 8, 9 and 10 will need to be solved as a single AI engine. Several disparate branches of AI research will have to be brought together as never before. Parts of the problems have been solved as graduate studies. A major effort here also.

17 Game theory can be applied here together with goal behavior.
Actually this part of the AI is not difficult. Lots of published algorithms available.

Well ladies and gentlemen, what mistakes have I made? What have I left out? What do you think the development priorities should be? The personality simulator needs hundreds of details I’m sure. This document needs lots of footnoting, who has what technology in what form? I am working from memory, sorry I do not have time to re-research all the sources. I have never done any lurking in the open source software forums, what good stuff can be applied to the above requirements?
By the way, as an animated interactive avatar, I think I remember a Sumomo out already, but that doesn't mean another can't be made.

As for the big head:
It's just coincidence facial expressions can be seen better.
She is Kawaii (cute) which has a long tradition of big heads, stemming from shadow puppets and marionettes.

Hey, is it possible to hinge the ear to reveal a USB hub/port (to recharge her and connect to the Internet) like the series?


Last edited by Artilects rule! on Sun Feb 08, 2009 9:25 pm; edited 1 time in total

Artilects rule!

Posts : 202
Reputation : 0
Join date : 2009-02-06
Location : United States

Back to top Go down

Who want to see Sumomo From Chobits made? Empty Re: Who want to see Sumomo From Chobits made?

Post  Artilects rule! Sun Feb 08, 2009 9:17 pm

Aiko wrote:It's will be hard not seeing the actually doll first.
But from what I see from the picture...
We can make the head moves and arms...
(not elbow.... too small to fit a motor)

The leg can move.... so she can do a little excerise like in the anime.

Can follow your faces.... and only simple object...
Will understand about 15 sentenses... with 10 movements..



Le
I know this isn't really the character, but I thought you might be interested.
E.M.A. Eternal.Maiden.Actualization
Dance mode needs improvement.

Artilects rule!

Posts : 202
Reputation : 0
Join date : 2009-02-06
Location : United States

Back to top Go down

Who want to see Sumomo From Chobits made? Empty Re: Who want to see Sumomo From Chobits made?

Post  Sponsored content


Sponsored content


Back to top Go down

Page 1 of 3 1, 2, 3  Next

Back to top

- Similar topics

 
Permissions in this forum:
You cannot reply to topics in this forum