Lawrence Tech Computer Science Students Participate in DARPA Urban Challenge Competition
By Brace Stout, MSCS student
Victorville, CA - It’s 7:00 AM on a cool morning in early November. Eleven vehicles sit at their starting positions; engines running, waiting for the starter’s signal. In the grandstand, spectators squint as the rising sun shines directly into their eyes. In the pit area, team members watch and wait with anticipation. As well they should; the winner of this race will go home with a cool two million dollars. Second place is worth one million. But even the biggest auto racing fans couldn’t tell you who the drivers are in this race. And there’s a good reason for that: The vehicles all drive themselves!
This is the “DARPA Grand Challenge - Urban Challenge” (often simply called “Urban Challenge”), where vehicles navigate autonomously through unknown urban terrain. This event builds on the successful completion of the “DARPA Grand Challenge” in 2005, where similar vehicles successfully navigated autonomously through 150 miles of desert terrain.
At the starter’s signal, there are no revving engines, no squealing tires, no smell of burning rubber in the air. A single vehicle eases away from its starting gate, makes a gentle turn, and makes its way onto the deserted and barricaded streets of the now-closed George Air Force Base. Vehicle starts are staggered to avoid the confusion (and almost certainly collisions) that would come with a simultaneous start. Once on the course, each vehicle is expected to visit in order a set of waypoints before returning to the starting point, obeying California driving laws, and safely sharing the course with other vehicles. Both autonomous and human-driven vehicles will be present on the course; human drivers monitor progress and safety of the autonomous vehicles.
Prior to race start, files describing the course in terms of latitude / longitude coordinates have been downloaded into the vehicles’ navigation computers. These files describe the course in terms of traffic lanes, allowable turns, and zones of free navigation, generally set up in parking lots. A ‘mission’ file describes speed limits (which never exceed 30 mph) and waypoints that must be visited in order.
There are two types of entries in this year’s race. “Track A” teams have each received up to $1 million in ‘technology development’ funds to help finance their entry. “Track B” teams received no such assistance. Team Cybernet is one such Track B team. Their “Cybervan” entry, also known as “Armadillo”, began life as a family vehicle of Heidi and Chuck Jacobus, owners of Cybernet Systems Corporation of Ann Arbor, Michigan. For the year prior to the final event, Armadillo was fitted with sensors, servos, actuators, computer systems and software that enable it to navigate city streets safely on its own.
Realizing that they could benefit from the autonomous robotics expertise at Lawrence Technological University, Team Cybernet approached Dr. CJ Chung of the LTU Computer Science Department in late 2006. What resulted was a year-long collaboration that culminated at the Urban Challenge “National Qualifying Event” (NQE), held for ten days prior to the “Final Event” on November 3.
As part of this collaboration, LTU students Marcus Randolph, MSCS, and Brace Stout, MSCS contributed software expertise in the areas of sensor data acquisition and data visualization. Marcus and Brace are not new to autonomous vehicles. They, along with others, represented LTU in both 2006 and 2007 at the annual “Intelligent Ground Vehicle Competition”, taking home the top design award in 2007 with their team’s H2Bot II robot.
In part due to their efforts, Armadillo performed admirably at NQE, passing many of the fundamental operational and safety tests that were run individually on the entries prior to the Final Event. However, unanticipated course configurations, sensor malfunctions, and calibration errors combined to prevent Armadillo from qualifying for the Final Event.
Even so, many praised Team Cybernet for their accomplishments, achieving more with $30,000 and an 11-year-old minivan than many teams having many times their budget. Team lead Chuck Jacobus was heard to say that Armadillo achieved the highest ratio of “autonomous miles per dollar spent” of any vehicle in the competition.