bobdina
08-14-2009, 07:40 AM
Two robotic test-bed vehicles showed off their ability to move at unprecedented speed while avoiding moving obstacles during an Aug. 10 demonstration at Patuxent River Naval Air Station, Md., U.S. Army and General Dynamics Robotic Systems (GDRS) officials said.
Accompanied by a Stryker control vehicle, the T2s hit 50 kilometers per hour while doing a perimeter check of a mock village, then rolled through the "town," autonomously detecting and avoiding moving mannequins that simulated pedestrians.
Testers in the Stryker drove one of the robotic vehicles by remote control, while the other drove autonomously.
"What we showed here was not just the ability to detect static obstacles but actually moving obstacles and predict the path of where the vehicle thinks the obstacles are going to go," said Ed Mottern, business development manager with GDRS's Safe Operations Program.
The algorithms that enable this have reached usability only in the past year, Army and GDRS officials said.
The Army Research Laboratory developed six algorithms, two of which were used in the T2s, said Jeff Jaster, deputy associate director for Autonomous Systems Intelligent Ground Systems at the Army's Tank-Automotive Research, Development and Engineering Center in Warren, Mich.
"We came to see that the algorithms complemented each other. When the other one was not performing as well - one was seeing better at a distance and one was seeing better up close - we brought them together and they now have a fused output," Mottern said.
The T2 vehicles also used a next-generation LADAR [a laser and radar scanning technology] to scan the surrounding area and so-called "image perception modules" consisting of color cameras, IR sensors and low-light cameras mounted on the front and back.
"The LADAR we are using is a preprototype which does have some limitations, but it offers you 360-degree scanning and you can get greater distances. We are feeding information back into FCS (Future Combat Systems) to help them improve their sensors and build the next-generation ANS hardware," Mottern said.
The algorithms were specifically adjusted to work with the new LADAR.
"This is an improvement over the other ones [LADARs] where you are typically limited to about a 90- to 100-degree field of view," Mottern said.
Much work is still needed before autonomous robots can fully navigate a complex urban environment by themselves, Jester said.
"We want to add more vehicles and more details, such as street signs," he said.
http://www.defensenews.com/story.php...97&c=AME&s=LAN
Accompanied by a Stryker control vehicle, the T2s hit 50 kilometers per hour while doing a perimeter check of a mock village, then rolled through the "town," autonomously detecting and avoiding moving mannequins that simulated pedestrians.
Testers in the Stryker drove one of the robotic vehicles by remote control, while the other drove autonomously.
"What we showed here was not just the ability to detect static obstacles but actually moving obstacles and predict the path of where the vehicle thinks the obstacles are going to go," said Ed Mottern, business development manager with GDRS's Safe Operations Program.
The algorithms that enable this have reached usability only in the past year, Army and GDRS officials said.
The Army Research Laboratory developed six algorithms, two of which were used in the T2s, said Jeff Jaster, deputy associate director for Autonomous Systems Intelligent Ground Systems at the Army's Tank-Automotive Research, Development and Engineering Center in Warren, Mich.
"We came to see that the algorithms complemented each other. When the other one was not performing as well - one was seeing better at a distance and one was seeing better up close - we brought them together and they now have a fused output," Mottern said.
The T2 vehicles also used a next-generation LADAR [a laser and radar scanning technology] to scan the surrounding area and so-called "image perception modules" consisting of color cameras, IR sensors and low-light cameras mounted on the front and back.
"The LADAR we are using is a preprototype which does have some limitations, but it offers you 360-degree scanning and you can get greater distances. We are feeding information back into FCS (Future Combat Systems) to help them improve their sensors and build the next-generation ANS hardware," Mottern said.
The algorithms were specifically adjusted to work with the new LADAR.
"This is an improvement over the other ones [LADARs] where you are typically limited to about a 90- to 100-degree field of view," Mottern said.
Much work is still needed before autonomous robots can fully navigate a complex urban environment by themselves, Jester said.
"We want to add more vehicles and more details, such as street signs," he said.
http://www.defensenews.com/story.php...97&c=AME&s=LAN