2011: The Year of the Personal Robot? | Hacker News Reader

2011: The Year of the Personal Robot?

scientificamerican.com

11 pointsRickasaurus10 comments

The PR2 robot seem to be overpowered and overengineered, making it cost too much.

I think the robot could do better by being easily producible by hobbyists and by using off the shelf component.

That being said, I am not an electrical engineer.

Agreed, no reason for this thing to have 8 core processors and insane amounts of RAM. It's not like this thing is sentient, there's only so much data to process.

I've done some work on image processing and read a substantial amount of literature, and nowhere have I seen anyone seriously trying to achieve sentience.

I have seen people using 30 node clusters where every node has roughly the processing power of this robot to track a ball with two cameras. Now, this was undergrad stuff and certainly could have been done better, but I'm sure the engineers who built this robot knew what it needed to do its job.

It's done as cheaply as anyone has done it up to now, which is important. Though it probably is still too expensive to be widely used, it sounds like a good step towards research to bring down costs to a reasonable level for consumer use.

While the PR2 is beyond the reach of a hobbyist, the company's meta operating system ROS (http://www.ros.org) can be installed on any personal computer for free. You can run the same navigation, vision recognition, and other algorithms the PR2 uses (among other robots) on most capable computers.

And with the new AVR bridge (http://www.ros.org/wiki/avr_bridge), you can even interface the "hobbyist" Arduino with the OS.

It sounds like it is a development platform, not a deployable product. Development environments for the Wii cost much more than one Wii does, not that this metaphor is perfect.

A common platform for robot development could really boost innovation in robotics. I smell disruption.

Sorry, there will be no disruption in personal robotics for many years. There have been articles proclaiming the year of the personal robot for at least a decade. The biggest problems in personal robotics are not computer hardware or software, but mechanical components and actuators. It still costs too much to build something that doesn't get boring after a week of working with it. Look at the Abio. It was beautiful but it got boring after a week because it couldn't do anything. Why not? Because it was too limited mechanically. If it was simply a matter of more computer power, there'd be people people hacking on more powerful processors and more memory, but there aren't. People want there $10K personal robot to be able to do something more useful than running a demo program. That requires power and robust and flexible mechanical system that can mimic what people can do. It may even require new materials (muscle wire, anyone?). Just my 2-cents.

It got boring because it wasn't part of a platform that could extend it - or be extended by it. It's like an operating system that only has a drawing program.

Edit: let's say the Aibo was an open sourced software plus blueprints for a 3D-printer. Anyone who wanted a "walking"-component for their robot could simply download it and build and extend it themselves. You could even run the entire thing in an emulator and then just press "print" when you wanted it in the physical world.

Now that _is_ disruption.

Yeah, we saw a lot of that happen with the Abio.

What about the Linux desktop :(

Reply on news.ycombinator.com