Irked but Inspired...

[PaulL]

Ok, so yesterday, someone I work with showed me a video of 20 robots dancing. Not the kind of thing that makes me fall out of my chair, but does make me curious. I found the video he showed me, searched the web a bit, and found:

http://www.robotshop.ca/aldebaran-robotics-en.html

Now, it's not that there's this robot out there that uses a few ideas I had that bothers me, that's fine. What I'm irked about is the price: $17,589.60 USD. WHAT??? Yeah. 25 DOF, Ok, fine. Has a board that runs Linux. Ok, fine. Uses OpenCV for vision. Sure, Roboard can do that, too. SO WHAT IS SO SPECIAL ABOUT THIS ROBOT??? I have an answer, and this is what makes me mad: NOTHING!!! There is NO "MAGIC" in this 'bot. It runs a 500 Mhz CPU, 128 mb RAM, 2gb flash HDD. Servo torque specs are not particularly impressive. Run time of 90 minutes is probably overstated, but then I'm not going to fork out nearly $18k U.S. to find out for myself. And the cost for "maintenance" and lack of publicly available innards (or even any pics of what's inside), it's ridiculous. I get even MORE mad when I look into the business (Aldebaran Robotics) itself, but I digress.

How does this apply to Roboard? I am inspired, with a bit of anger that Roboard is more powerful, and yet there's this bot selling for nearly $18k with less power / capability than Roboard. WHAT IS WRONG WITH THIS PICTURE? Roboard needs software, the kind it truly deserves- robust, powerful, flexible, configurable, extensible software.

Companies like this should NOT exist, not with a product like Roboard out there. I am quite frustrated for not having made more progress on my 'bot software up to now, but I am going to be spending quite a bit more time getting MY bot software built, using my Roboard, for a whole lot less money. And another point, it's hands are 1 DOF, I very much intend to have 5 in each hand, that is NOT a forgotten aspect to my project. I need to get the CAD work done and fabricate the brackets on my CNC'd mill.

In the end, I guess you can say finding this overpriced bot is my newfound inspiration for developing software for Roboard.

Take Care,
Paul

[JavaRN]

I agree with you that Nao is definitely overpriced especially for what it offers in terms of computing power. My bioloid has much better computing power through the use of Roboard - developing software for roboard is an excellent experience, I've been doing this for the past six months (through various languages) and beleive me it is an excellent resource for robot hobbyists and at a reasonable price.

Just a side note to be fair . . .

In my opinion fit-pc2 is also an excellent computing unit for robots, but heavier and more expensive than roboard.

[Tyberius]

It's what you have to pay for an 'out of the box' humanoid with a pretty shell. Look at Manoi pricing compared to the standard Kondos, case and point.

The only thing that I have to chuckle at, is for that price you can buy a fitPC2, 2 Point Grey Research cameras, a large handful of EX-106+/RX-64, buy all the pre-made brackets from Robotis and pay someone to custom machine you the frames and parts they don't provide, and assemble one that would be FAR superior for less money.

From what I've heard, the Naos even use plastic gears =/

[Spiked3]

Yeah, I'd have to agree somewhat with what is said. The stuff is ridiculously overpriced.
My peeve; Laser range finders.
I found this one; http://www.roadnarrows.com/robotics/store/hokuyo.html for $1175. Given the costs of others I'm thinking maybe that is reasonable.
But then I see other things on their web site that look to be 3X a normal price, so it makes me wonder.

But the real problem is software - it just doesn't exist. Everything is custom written from scratch. Existing 'open' projects are immature junk, or at least I have yet to find anything useful in a couple of months looking. Yeah there are low level libraries, but why no high level libraries? I bought a RB110 thinking it would be a good brick for Microsoft Robotics Studio, the advertisement even says it would be. And maybe it will, after I write the code myself Come on, this thing has been out for over a year and no one has written MRDS services for it?
Robotics will never become popular until this changes. Now, we are like the nerds with altair computers back in the 70s, writing assembly language to make lights blink. Until something like IBM/DOS (and Windows) comes along robots will remain over priced novelties, not inexpensive practical commodities. I can't express how disappointed I am at Microsoft's lack of effort. MRDS is an abandoned platform with no resources ($$) available as best I can tell.

[limor]

Aldebaran have allegedly received something like $7m in 3-4 investment stages. The founder who worked in the financial industry previously and he managed to get an incredible agreement from the Robocup Association that guaranteed some return on the most sizable private investment ever in commercial humanoid robots outside of Japan and Korea ..

In order to understand the incredibleness of the deal, a few words about Robocup... Robocup is an academic non-profit organization chartered to promote robotics through competitions of various kinds. Thousands of researchers from around the globe gather at Robocup to compete and advance their area of research. The code running their robots is supposedly shared after the year's competition events thereby improving the overall performance of the robots year after year. The German humanoid robot leagues are composed of sometimes tens of students that take a year off just to help with preparing the robots for war (football) and running on uneven terrain.

Some of the most visually impressive achievements at Robocup were at the Aibo league (check out youtube). The small canines were reprogrammed from scratch and were able to actually run and play team football, passing the ball between players, following the leader, tracking motions of enemies etc. Unfortunately Sony axed the Aibo project in 2007 and so the Aibo league had to choose an alternative canine (i was actually there when a handful of suppliers presented their canine robot alternatives to the Aibo teams).

So how has that got to do with $17k NAO humanoid ?

To everyone's amazement, by 2008 Aldebaran was announced as the new Aibo replacement. The NAO humanoid (back then a prototype based on Robotis AX12 servos and a shell designed by a 3rd year design student who won a competition sponsored by Aldebaran), was elected to completely replace the Aibo league and Aldebaran were essentially guaranteed several hundred units to be sold to Aibo leagues by 2010. The cost to the first competing groups was highly discounted and they had to endure alpha/beta testing of the robots. However, several new teams were able to join the Robocup league because it was open field for everyone. With the Aibo league it was very hard for a new entrant to get in and compete with the hundreds of cumulative years of knowledge of other teams.

So you may ask, how come there's a humanoid robocup league and a NAO humanoid league ?

The answer is that the Humanoid league is composed of researchers that like to build their own humanoids from scratch. Many of them frequent our forums. HaViMo vision modules is the product of one of those teams. Aibo/NAO league on the other hand is composed of researchers that like to use a ready out-of-the-box platform that can be programmed in high level languages without the need to delve in electronics, control and screws. NAO allows computer science researchers to focus on AI strategy and computer vision while the platform is not as agile and versatile as the DIY robots of the humanoid leagues. But since they are all with the same platform, the challenge is how to make the most out of the platform rather than to try and re-invent new parts and mechanics to improve their agility.

[siempre.aprendiendo]

Awesome, Limor

I liked specially:

"In order to understand the incredibleness of the deal, a few words about Robocup... Robocup is an academic non-profit organization chartered to promote robotics through competitions of various kinds. Thousands of researchers from around the globe gather at Robocup to compete and advance their area of research. The code running their robots is supposedly shared after the year's competition events thereby improving the overall performance of the robots year after year. The German humanoid robot leagues are composed of sometimes tens of students that take a year off just to help with preparing the robots for war (football) and running on uneven terrain. "

[limor]

according to this article from yesterday:

NAO Academics has been available for only two years, but more than 700 models have already been sold to 200 of the most prestigious universities in the world. NAO's proven track record with academia and research has convinced Aldebaran Robotics that the time is ripe to introduce NAO to higher education.

[JavaRN]

I think that this should be used as a standard humanoid for robocup.http://www.youtube.com/watch?v=amiUhtSrHLI

Has anyone more info about it?

[bluecat]

the darwin is so beautifall, i googled it and thats what i found :

http://www.plasticpals.com/?p=23449

[PaulL]

Limor, thanks for all that info, just gives me more drive.

I still can't get over the cost... Ridiculous.