Slashdot Mirror

robot throwing

Pay attention to the motion of the camera (and what the robot does just prior to the camera moving!)

Ohhh Boy, embedded video on the web page. That site will be smoked QUICK!

Here is a link that shows what appears at a glance to be a superior robot model from Japan doing all kinds of neat tricks. It has 22 Degrees of Freedom rather than the 17 DOF in the robot from Wow Robot. This page also has video, but they are all links at the bottom of the page.

I agree that one of these would be great fun to play with!

Hey people,

I was browsing slashdot today and found out there was a storie about me. Somebody apparently posted those movies I made of myself playing soccer naked. Very funny. If I ever find out who did this, i'll punch you soo hard!!!

Kiss my shiny metal ass,
Robovie-M

Can your three-year-old do this?

This is a really interesting opportunity for some high-tech to be applied to a real non-military situation. Robin Murphy came here to UCSD and talked a few months ago, and she actually brought (and we got to drive) the tracked one with the flippers in front.

Those things are not easy to drive. One of the most difficult things is getting a perspective on where the robot is in relation to it's surroundings (very rough rubble). This is an ongoing research area for many robotics teams, and one we have been working on also.

The submitter mentioned something about autonomous robots, I think they don't fully understand the difficulty of the problem which robotics researchers are working on. Navigating uneven building wreckage autonomously is an incredibly difficult problem, in general. Especially under the conditions of the WTC rubble. There may be some small parts of the process which can be automated, but I doubt it would be useful in this situation anyway. They were using the robots as probes to discover what was inside areas where it was dangerous for people to be there, so a human is already "in the loop". The real use of these systems is for remote visualization (i.e. show me what's in there) in hard to reach areas.

They didn't specify what types of cameras are being used, but this is a mostly visual problem from my understanding. Most robots have standard rectilinear camera views that are forward facing, unfortunately operation of these platforms is difficult becuause of the restricted field of view and inability to see on the left, right and behind the robot. Multiple cameras helps, but adds significant complexity and disjoint views. A technology which really makes this easier is an Omni-directional Video sensor (which has a 360 deg. field of view around the sensor). These are ideal for "immersed" applications like this, and they literally give the operator a view of the entire space around the bot (except for directly overhead) and allow you to determine the robot's orientation relative to obstacles easily. The same data can also be unwarped and used to create a perspective or panoramic view of the area in real time. A pair of these and stereo software (which also has been done in our lab, [shameless plug over]) can provide a full depth-map of the area. The ODVS has the difficulty of limited resolution (same CCD, larger fov) but this can be supplemented by a Pan/Tilt/Zoom rectilinear camera.

Really the interesting part of research in this direction is the remote operation and visualizations that help the perator navigate through the area to achieve it's goal. This is what my thesis is on, actually.

More info: UCSD CVRR Lab The Page of Omnidirectional Vision and our source of ODVS. Also check Vstone (in Japanese, may need to run that last one through babelfish or something).

Mobile Robots are cool. We even have one that pulls cables for us in the drop-ceiling of our lab... we're slowly working on a web-page for that new one.. I have a cool video for it already but it's HUGE (100M or so). Anyway, I'll shut up.