As a pioneer in robotics technology, Carnegie Mellon’s Robotic Institute is paving a path towards the future and relying on advanced ATI FireGL technology solutions to help get them there. Led by Professor William Whittaker, who has developed more the 60 autonomous robots, the Institute believes they are on the cusp of widespread adoption and use of robots.

Intrigued by DARPA’s (Defense Advanced Research Projects Agency) $1 Million Grand Challenge, Carnegie Mellon’s “Red Team” has developed Sandstorm, a modified Humvee set to compete in the 210 - mile race from Barstow, California towards Las Vegas. Twenty race teams will be given the course details just two hours before the start of the event. The unmanned off road race for autonomous robots requires extensive mapping and route planning to successfully complete the course in 10 hours, as teams will not be able to communicate with their vehicles once the race begins. The various different obstacles that Sandstorm may encounter include surfaced and unsurfaced roads, overpasses, water and other natural phenomenon, manmade creations such as fences, power lines and even other vehicles.

The team has worked on highly sophisticated projects in the past, but never a challenge with so many variables thrown into the mix. To conquer a course that is unknown until two hours before the race, the Red Team has relied on ATI FireGL graphics technology to develop extensive 3D terrain maps, forming a database for the robot to refer to come race day.

In 2003, the Red Team went to task, compiling data from aerial and space imagery, GPS co-ordinates, and topographical maps of the course area, and employing ATI FireGL technology to develop the most complex maps ever created of the Mojave Desert. As Prof. Whittaker explains, “By integrating these various data sources and utilizing ATI FireGL’s onboard graphics memory, we can now easily create and manipulate very detailed 3D maps of the course.”

In compiling these exhaustive maps, the Red Team is attempting to plan every scenario that Sandstorm may encounter along the way. Once the race begins, the Red Team’s vehicle will maneuver without outside assistance or human intervention, using laser scanners, radar sensors and stereoscopic cameras to continuously correct its steering and adjust its speed. Based on the terrain data assembled on ATI FireGL - based workstations and its real-time visualization systems, the Sandstorm will be free to decide on alternate routes as it avoids obstacles and other vehicles.

• Case study on Red Team
• Red Team press release