DARPA Grand Challenge Most Interesting -- Team DAD
DARPA announced the list of 23 robots which are finalists to compete in the Grand Challenge. These are completely autonomous vehicles, which will attempt to drive 150 miles across rough southwest terrain both on-road and off-road.
There are some interesting robots out there, but I think the most interesting is Team DAD (Digital Auto Drive). Last year, they were one of five teams to finish the qualifying course. They came in third at the actual event, getting six miles down the road.
The winners, Red Team from Carnegie Mellon, got seven miles and some change. The interesting part is that Red Team got totally hung up on a road berm, eventually spinning the front wheels so much that they caught on fire.
Team DAD, in contrast, was only a little stuck -- the DARPA chase vehicle had paused the robot, at which point it got hung up on a small rock. When it was restarted, it couldn't get over the rock, apparently because the speed limit for the segment was 5 mph, and so it didn't really hit the gas to get over the rock.
When the DARPA personel got into the truck and hit the gas, it easily surged past the obstacle. So it is entirely possible that the robot could've gotten much further.
The cool part is, rather than grouse about their bad luck, the team later took the vehicle along the marked course, past the six mile point, and acknowledged that their vehicle would've gotten hung up eventually, anyway. That sort of honesty is admirable and rare.
But that's not really the amazing part. The amazing part is that the only sensor on the robot (barring GPS) was a stereo camera system feeding into a set of DSPs. This is in a field of competitors who deployed a dizzying array of LIDARs, sonic sensors, and massive computing power.
Team DAD's setup was so compact that the truck it used as a platform could still be driven manually with no problem. They drove it both to the qualifier event and to the final Grand Challenge event this year. Again, this was in a field of competitors who extensively modified and trailered their vehicles, including the 32,000 pound TerraMax and the extensively modified Team Red robots.
Team DAD really seems to fulfill the Grand Challenge's desire to showcase radically different technologies. This year they seem to be using some strange spinning laser scanner (their website doesn't really talk much about it), but it still seems to be a minimalist solution.
I doubt that they'll win the competition -- both Stanford and Team Red are faster -- but they certainly seem to be applying some fresh thinking to this old problem.
There are some interesting robots out there, but I think the most interesting is Team DAD (Digital Auto Drive). Last year, they were one of five teams to finish the qualifying course. They came in third at the actual event, getting six miles down the road.
The winners, Red Team from Carnegie Mellon, got seven miles and some change. The interesting part is that Red Team got totally hung up on a road berm, eventually spinning the front wheels so much that they caught on fire.
Team DAD, in contrast, was only a little stuck -- the DARPA chase vehicle had paused the robot, at which point it got hung up on a small rock. When it was restarted, it couldn't get over the rock, apparently because the speed limit for the segment was 5 mph, and so it didn't really hit the gas to get over the rock.
When the DARPA personel got into the truck and hit the gas, it easily surged past the obstacle. So it is entirely possible that the robot could've gotten much further.
The cool part is, rather than grouse about their bad luck, the team later took the vehicle along the marked course, past the six mile point, and acknowledged that their vehicle would've gotten hung up eventually, anyway. That sort of honesty is admirable and rare.
But that's not really the amazing part. The amazing part is that the only sensor on the robot (barring GPS) was a stereo camera system feeding into a set of DSPs. This is in a field of competitors who deployed a dizzying array of LIDARs, sonic sensors, and massive computing power.
Team DAD's setup was so compact that the truck it used as a platform could still be driven manually with no problem. They drove it both to the qualifier event and to the final Grand Challenge event this year. Again, this was in a field of competitors who extensively modified and trailered their vehicles, including the 32,000 pound TerraMax and the extensively modified Team Red robots.
Team DAD really seems to fulfill the Grand Challenge's desire to showcase radically different technologies. This year they seem to be using some strange spinning laser scanner (their website doesn't really talk much about it), but it still seems to be a minimalist solution.
I doubt that they'll win the competition -- both Stanford and Team Red are faster -- but they certainly seem to be applying some fresh thinking to this old problem.
posted by K at
5:30 PM
|
0 comments
