Shape of things to comeFirst round of tests for entrants in DARPA's Urban Challenge

Published 15 June 2007

In DARPA’s Grand Challenge, driverless cars competed in traversing a 130-mile course in the Mojave Desert; in Urban Challenge, driverless cars will compete in navigating through urban traffic

As you drive on the highway or even on city streets, you often have that uncomfrotable sensation that you are surrounded by driverless cars. There is such a thing as a driverless car, and during the past year we have reported on two DARPA-sponsored Grand Challenge competitions in which driverless cars competed against each other traversing a tortuous, 130-mile track in the Mojave Desert. The winner of the Grand Challenge was Stanley, an autonomous vehicle designed by Stanford University researchers led by Sebastian Thrun.

The same Stanford University team has now converted a Volkswagen Passat into an autonomous vehicle called Junior, which is the team’s entry in the third DARPA-sponsored Grand Challenge — the Urban Challenge competition which will be held on 3 November 2007. In Urban Challenge an autonomous car must navigate city streets, obey traffic laws, avoid obstructions, and drive well among other cars in traffic.

Developing reliable driverless cars has two important benefits. The system and intelligence put in these autonomous cars may be deployed in regular cars to improve these cars’ safety. For battlefield and homeland security applications, the advance of the autonomous car would have the same benfits as advances in UAVs and robotics have: More and more difficult and dangerous missions could be carried out effectively without risking the lives of soldiers or first responders.

Technology Review’s Kate Green reports that Stanley had a number of GPS sensors and lasers, a camera, and other equipment to help it make its way through the Mojave desert course. Junior is based on the same fundamental technology, says Thrun, but with some crucial improvements. Junior uses the same kind of laser perception as Stanley, but with longer range. The new car has a total of eight light detection and ranging (LIDAR) systems which emit beams of light and detect reflections to determine the distance of other objects. One system is mounted on the front of Junior’s roof and has a range of about 100 meters—many times that of Stanley. Another LIDAR system is aimed at the ground and constantly keeps track of the road and reflective lane markers. A third system constantly takes a 360-degree image of its surroundings. All this data is processed by two Intel quad-core machines running at 2.3 gigahertz, and the relevant information is relayed to the driving systems, which guide the car.

Junior also has a precise location system which includes GPS and other sensors that measure the revolution of the wheels and the direction the car is moving in. Together, these sensors allow Junior to pinpoint its location to within thirty centimeters.

Thrun points to a major difference between Stanley and Junior: Junior has much more “intelligence.” This is necessary because urban traffic is much more complex than traveling in the desert. Junior’s intelligence comes in the form of about 500 different probabilistic algorithms which process all the environmental information collected by the sensors and make the decision which is deemed most likely to be the best. As importantly, these decisions are made in less than 300 milliseconds, which is sufficient for slowing down or changing lanes if a car in another lane tries to merge into Junior’s.

Yesterday, Junior successfully completed all the tasks DARPA assigned to it in its preliminary tests. There will be another round of tests in October, and if Junior passes, it will make it to the final on 3 November.