Ghost rider

A riderless motorcycle has been developed by a team of graduate students from the University of California at Berkeley’s electrical engineering program.

The team’s ‘Ghostrider’ is one of 40 semi-finalists selected for the 2005 Grand Challenge, a 150-mile off-road race for autonomous vehicles sponsored by the Defense Advanced Research Projects Administration (DARPA). The first vehicle to successfully complete the 150 mile course in the October 8 competition will win a $2 million prize.

The DARPA Grand Challenge is designed to further the development of autonomous vehicle technology for the Department of Defense. It requires self-navigating vehicles to travel on rugged desert roads using onboard sensors and navigation equipment to find the route and avoid obstacles.

To guide their vehicle, the University of California team, known as the “Blue Team,” chose Cognex In-Sight vision sensors over other technologies such as lidar, ultrasound and radar. The Blue Team is the only Grand Challenge competitor using machine vision as its primary guidance system.

“We have taken a different approach, both in the use of a two-wheeled vehicle for a narrow profile, and by relying solely on machine vision for guidance,” said Anthony Levandowski, leader of the Blue Team.

To recognise and avoid obstacles, the Ghostrider uses two In-Sight 5403’s, with the cameras mounted on a gyro-stabilised gimbal system to keep them level and looking forward, no matter the attitude of the bike. The robotic motorcycle also relies on one In-Sight 5400C, a colour vision system, to enable the vehicle to distinguish where the road is and to follow it.

The Ghostrider will compete head-to-head with the 39 other pre-qualifying teams in a final qualifying event at California Speedway in Fontana, California from September 27 – October 6, 2005. Twenty teams will be chosen from this last qualifying race to compete in the final off-road race on October 8, 2005.