I am happy to see interest in Squeak-AIBO port. This summer I have acquired a few AIBOs (ERS 110) for the purpose of preparing for the legged robots competition where AIBO is the official platform. Although AIBO is promoted as an open platform it is not the case at all. There is development software (SDK and compiler) for the AIBO available to qualified ROBOCUP competitors but: a) It is $3500 apiece and you have to get one for each CPU (min 4) b) You have to sign a tough non-disclosure agreement c) You cannot use the development platform for any other purpose other than the competition. Obviously releasing a Squeak for AIBO under these terms is totally out of question. Even porting to a platform like that would not be allowed under Squeak license, as I understand it because you can't release it to the public under the terms of the NDA. After all this I was disillusioned and put the project on halt.
BTW again my interpretation of the Squeak license would not allow for porting to proprietary hardware that a company wants to keep the details a trade secret and therefore would not want to disclose the ported code as it would give away a lot about the hardware. Is this a correct interpretation?
There is not much doc about Aperios (some on its predecessors) but it is an interesting OS esp. for such purposes. It would not be trivial to port Squeak on it, but would be well worth it. I have a cool system that I have partly ported to Squeak. It is a merger of fuzzy logic and object oriented approaches that gives you a lot of power in controlling things like AIBO and it would do a better job of representing emotions and certain movements than what they have currently. It is not ready to be released yet and without such platforms it is difficult to demonstrate the power of that system but Luciano Notarfrancesco had a quick glimpse of it and you can ask him what he is thinking about it.
Hardware wise AIBO is more than enough to run a headless Squeak with my extensions quite comfortably. We were able to cram a lot to a Cassiopeia E100 (16M CE handheld) for a demo that had about 100+ full color images occupying more than 8+M of memory and 2MBs of sound (For the most part, decompression on the fly did not prove practical for CE for us). In AIBO, the most space that we may want to occupy will not be the code logic, but things like sounds.
Anyway, if Sony continues with this policy I might need to change the computer and software in AIBO and put my own embedded system with Squeak in it.
I hear Alan has some sort of contact with Sony and he could be in an influencing position. Is this the case Alan? Maybe you could help us with PS2 and AIBO development platforms.
To friends in Japan: if you are working in Sony or have friends in Sony may have a better chance of raising your voice to keep to their word and release information and development tools about OPEN-R and Aperios, or just remove those OPEN-R stickers from them. Also, I will be in Tokyo on 2nd or 3rd of February briefly, if there is a chance of meeting you can email me privately.
Cheers
Tansel
Yoshiki Ohshima wrote:
Hi Soren and Luciano,
# Thank you for sending the mystery fish movie the last # time!
Soren wrote:
It seems that Sony is supporting at least one open source project with hardware and development kits. See http://wulfstation.sourceforge.net/ . The goal is to make a Beowulf supercomputer out of a cluster of networked Playstations. So maybe we will not have to pay. Certainly our project is as worthy as theirs.
Interesting.
One thing I can say is wulfAIBO is less worthy than wulfstation.
# theirs?
**(Surely the hardware accelaration part will take much much longer:-)
That is also what Luciano thinks, but I disagree. Imagine porting Balloon3D to a machine which required OpenGL commands to be issued to the hardware. It would not be difficult, would it? Surely the PSX2 comes with a C library for accessing the rendering hardware -- not OpenGL, of course.
What I meant was it will take longer to write the 3D support code than to write minimal interface support such as BitBlt and pointing device. I don't say 3D part is easy, but I agree with you that what to do is not uncertain.
# But my experience taught me that making it work and tuning the # performance is quite different.
## I guess we might want to implement it so that rendering ## and interpreting are carried out in parallel.
Luciano wrote:
AIBO! Having Squeak running on a dog! Great idea!!! I love it!! Do you have an AIBO? A friend of mine (Squeaker too) has one, and I might be able to play a little with it. But I don't have a development kit for AIBO, actually I'm not sure such a thing exist. Both PSX2 and AIBO have MIPS in them, right?
I don't have one, but a research group I sometimes visit own one. AIBO has (built-in) "motion editor" and you can program its motion for some extent, but I don't think there is a development kit available.
They say the chip on AIBO is R4200. The OS of AIBO is an "object-oriented OS" Aperios, which was formerly called Apertos, which was formerly called Muse. The OS was designed by Yokote-san based on his Concurrent Smalltalk work.
The chip on PlayStation2 is *completely* new. I'm not sure that even it has 32 bit arithmetic and 32 bit load/store ops.
-- Yoshiki
squeak-dev@lists.squeakfoundation.org