Over the years, the automotive industry has seen the emergence of increasingly autonomous, “smart” technology enabling automotive electronics features such as adaptive cruise control, lane departure warning, and smart airbag deployment. Next on the agenda will be cars that can see, offer advice to drivers, and even react autonomously to changing road, traffic, and weather conditions.
Much of the required technology to put autonomous vehicles in motion already exists. Current challenges have more to do with product definition, system integration, and building a smart-vehicle infrastructure as well as the reliability, tolerance, and cost requirements of the automobile industry. But as both the automotive and semiconductor industries have demonstrated in the past, the chief barrier to progress is lack of imagination.
Not your dad's truck
Digital Auto Drive (DAD), a California-based research and development organization specializing in vision-recognition-based 3D imaging and navigational technologies, and 22 other competitors in the DARPA (Defense Advanced Research Projects Agency) Grand Challenge competition suffer from no such lack of inspiration. DAD has created a driverless Toyota Tundra truck, based on Texas Instruments digital signal processor (DSP) technology, that can see the road ahead of it, make immediate navigation decisions, and drive itself with no human intervention.
Team DAD and the others will race their unmanned vehicles in the 2005 Grand Challenge on October 8 in Primm, Nevada. Qualified teams are required to send a fully autonomous land vehicle approximately 150 miles across the Mojave Desert, in an effort to help the Department of Defense discover new technologies that will enable unmanned military vehicles to be added to the U.S. armed forces.
Part of the impetus for the DARPA Grand Challenge came from a congressional mandate that one-third of all military vehicles must be autonomous by 2015. The desert route that the DARPA Challenge’s driverless vehicles navigated in 2004 was approximately 110 miles from Barstow, Calif. to Primm. A transparency overlaid on the 2004 Grand Challenge route showed almost an identical route from Baghdad to Fallujah, Iraq. The word “autonomous” can have many meanings but in this case it means that vehicles must be fully self-navigatingno remote control or other outside influence on the behavior of the vehicle is permitted.
While the military market is the immediate target for unmanned vehicle technology like that found in the Grand Challenge's competitors, the consumer automotive market will also benefit. Many applications used in the Challenge vehicles, such as obstacle detection, could even impact collision avoidance systems technologies currently under development by automotive manufacturers.
2004 in the rear-view mirror
The first Grand Challenge race took place on March 13, 2004. Of the 15 teams that made it to the 2004 starting line, only seven vehicles made it out of the starting gate area and only four made it five miles or farther.
Team DAD’s Grand Challenge experience began at the qualification round in 2004, where they completed eight test runs, more than any other team, and were awarded a 4th pole position. The Team DAD truck traveled a total of six miles before hanging up on a rock after being paused to let a tow truck past. Team DAD passed one other team along the waythe first recorded passing between two autonomous vehiclesand traveled the third farthest of all the competitors.
Design engineers' challenge
Several complex technologies must come together seamlessly to enable a full-size pickup truck to drive itself across the rugged desert. The Team DAD entry consisted of two main components: Real time stereo 3D imaging and vehicle servo control. This approach to autonomous navigation leveraged state-of-the-art TI DSP technology to enable a high-speed stereo vision system and advanced servo control. DAD also benefited from TI’s experience in servo control, robotics, and automotive to orchestrate their autonomous vehicle.
Team DAD deployed a 3D imaging system that provided environment sensing and obstacle avoidance. This system employed two CCD digital cameras that rendered a comprehensive 3D depth map, which in turn was used to identify required deviations from the given GPS-derived path in order to avoid obstacles.
Team DAD’s approach to autonomous design was, and still is, deceivingly simpleautonomous operation is achieved through a simple throwing of three switches, one each for steering, throttle, and brakes. Riders can be in the vehicle when the vehicle is in autonomous mode. This streamlined design enabled DAD’s engineers to rerun the course after the race while sitting in the truck.
Having a truck that was street legal with full passenger comforts, even in autonomous mode, enabled DAD’s engineers to do all of their pre-event testing while riding inside the vehicle. In fact, the team drove the truck to the Grand Challenge.