Slashdot Mirror
Robosapien: Latest Toy Robot From Mark Tilden
Onnimikki writes "Mark Tilden has been building really cool BEAM robots for a long time. Now, he's come up with RoboSapien, a toy that no self-respecting geek can go without. Videos of the RoboSapien at the 2004 New York City Toy Fair have been made available by Solarbotics. Mark offers some really good explanations about what makes them work."
at best buy, 100$ pricetag
every day http://en.wikipedia.org/wiki/Special:Random
The philosophy of BEAM robotics
Bigger picture:
http://www.androidworld.com/www_toy.jpg
Video:
http://www.iirobotics.com/downloads/robozip.zip
Hate to do this to this poor server.. But there's a zip file with two videos here:
http://www.iirobotics.com/webpages/hotstuff.php
Have fun!
.
The trailer/advertisement for the I, Robot movie being made right now. Looks more like an ad for an actual robot, rather than a movie.
Join moola.com, play games to earn money.
BestBuy is taking preorders for RoboSapien at $99.99 shipped free.
Toysrus.com has it for $89.99 but no free shipping.
The Solarbotics server is under a bit of stress, so here's a torrent for all four video files, 42.7MB total.
Correct, Aibo is approximately 8X the cost, but it adds:
color CCD,
3-axis gyroscope,
300+mhz 32bit RISC processor,
MemoryStick slot for data storage/user-written code (I have a 16mb card in one of my Aibos, 32mb in the other)
wireless lan card
OS based on Unix (Aperios)
stereo microphones
14 DOF
etc. etc. etc.
In short, this is a neat toy. Aibo is a neat toy, too, but can serve as a robust hardware platform for serious robotics/AI research.
Well, strictly speaking everything in the real world is analog, of course. But, in the way we're using the terms here, analog means made from discrete analog components and feedback circuits with fixed values, which are distinctly unlike the human's (and other animals') unique ability to vary the analog operations in such widely varying and relatively precise ways.
Digital, as relevant here (like an Aibo), means able to be approximated by binary values and transformed by logical operations using digital circuits that drive digital-analog converters such as servos and motors with "digital" imputs and controls. This sort of thing lends itself very easily to programming that can be changed and modified easily, sensors added to the system with little impact or re-design needed, etc.
My point was that analog discrete devices, like the ones used in this toy, tend to be only cheap enough to warrant a system price of $80 when they are the plain old-fashioned fixed values, which means the circuit made of these that controls the behavior is not variable (its behavior depends on these fixed values). It does one thing, and has a few circuits that it can shunt in an out to do several canned things. But making it do a new thing, even a slight variation is hard and expensive, and adding a new input from a new sensor, something trivial in most digital control systems (like an Aibo), is nigh impossible.
So, again, the only way this sort of analog-circuit control system robot toy will help bring down the cost of other, digital processor-based robots, is if we find a way to make cheap discrete components with variable parameter values controllable by digital logic, and even then the savings would be pretty small. You still need the ASIC with the microcontroller in it. Maybe your servos and motors could be a bit cheaper -- maybe.
everything in moderation
Let me hold your hand as we stroll through it together, then:
This is not a robot (in the opinion of the original poster and me) because it is neither autonomous nor does it have the capacity to be made autonomous-ish by adding sensors and a brain (microcontroller). It can only perform canned macro-functions and sequences of these canned macro-functions. Micro-scale control of its functions is not available.
A programmable assembly-line robot is, however, a robot not because it's autonomous in itself (assuming you're talking about the arm / mechanical part), but because it could be autonomous if you grouped it with it's controller, which is a reasonable thing to do despite the fact that they tend to be seperated by some distance in practice, they key is they need not be. In most assembly lines, the robots are programmed to do repetitive tasks with minimal or no variation of behavior based on sensor inputs. But they could easily -- you just have to re-program the microcontroller (brain) and add sensors. So they are indeed robots -- micro-scale control of their behavior is available, and with inputs, a microcontroller, and some clever code, you can make an autonomous robot.
everything in moderation
Good, strong joke - but c'mon man, you're making my eyes bleed over here.
People, it's would have. As in "We would've done it that way, had we known better. We would have written it like so, but we insisted on doing it incorrectly - for some incomprehensible reason." /Grammar-nazi-within-me out.
I knew Mark as an undergrad- he mostly developed the insectoids that led to BEAM while employed by UW ;) Only he could get a patent on using CMOS inverters in analog mode when thousands did so before him. Mark proseletyzed his bugs around and was offered a six figure salary and his own lab at LANL, an offer he couldn't refuse. Nothing like DARPA dollars! Of course they had to send him away a few weeks to buy him a PHD somewhere as he doesn't have a legitimate one, a requirement for that level of pay at LANL. Green card? No problem - when the US gov wants you to build robot soldier hordes for them... last I heard he's doing a quarter mil a year from LANL even without wowee's and BEAM revenues. He is however a super nice guy, very enthusiastic even though a bit full of himself now, considering he fancies himself god's gift to AI, rather than just a smart guy that made cool bots out of walkman parts and turned it into a religion.
The player system.
If you have a robot which supports some form of connectivity (IR, wireless, tethered.. protocol isn't all that important), you can make player connect to your robot. Player is a TCP server which then allows you to write your robotics code in whatever language you see fit, provided it has the ability to connect via TCP. It abstracts away hardware in much the way a driver does, and provides a uniform way to access sensors and effectors.
It's a nice system.
-
Yes, Honda has ASIMO, or Advanced Step in Innovative Mobility. I remember first seeing ASIMO walk around, looking a little creepy, since it walked with a relatively "human" style. It also "...turns sideways, climbs up and down stairs, and turns corners." And it's starting to look more and more human with each new prototype.
Quidquid latine dictum sit, altum sonatur.
Part of the genius of Tilden's nevous network (different from neural network) technology is that it makes use of the analog noise. The back-EMF (noise) from the dc motors is used to directly inform the nervous neuron about physical interactions with the environment.
What are ordinarily considered problems to be engineered out of analog designs are considered as opportunities for exploitation by BEAM roboticists