Advancing Robotics: The Urban Challenge Effect

Abstract

The Defense Advanced Research Project Agency’s (DARPA) Urban Challenge, a contest that pitted robots against each other for 6 h without human intervention, was aimed at advancing the state-of-the-art in research for perception, planning, sensors, and situational awareness with the goal of military application. The event was a success, gaining national media attention for DARPA and robotics; but how well did DARPA succeed in advancing research for the military? To answer this question, a number of post-event activities conducted by Lockheed Martin Advanced Technology Laboratories, a participating member of the Ben Franklin Racing Team, will be presented and examined. In addition, a comparison of requirements between the Urban Challenge contest and current military applications will be conducted. I. Introduction I T is rare to be both a part of history and to write it.That is what we attempt to accomplish at the DefenseAdvanced Research Project Agency (DARPA) Urban Challenge. This paper pulls ourselves back for a moment and tries to view the field from afar, seeing from whence we have come and anticipating where we will go. DARPA’s Grand Challenges have become a large part of robotic history. At the same time, there is a congressional mandate that sets a goal for the future that a third of all ground vehicles in the military be robotic. How far have we come and how far do we have yet to go? Answering these questions is the focus of this material. The Urban Challenge (Fig. 1) was DARPA’s third Grand Challenge—a contest that pitted robotic vehicles against each other in a simulated urban environment. The final event included 11 robotic vehicles and their DARPA chase cars, along with 50 manned vehicles that were used to simulate various traffic situations. All robotics vehicles were required to be completely autonomous and team members were not allowed on the course while the vehicles were running. The contest was both a race, with a maximum time limit of 6h, and a “driver’s test” for the autonomous vehicles that evaluated their skill in executing maneuvers that a typical human driver would encounter on populated city streets. For brevity, we will not list all the requirements for the Urban Challenge. In summary, the Urban Challenge required vehicles to detect roads, lanes, and other vehicles; perform left, right, and U turns; pull in and out of parking spaces; merge into traffic; change lanes in traffic; queue up in line at traffic stops; understand and follow intersection precedence;andavoidcollisionsatalltimes[1,2].Ingeneral,DARPArequiredallvehiclestoobeyCaliforniadriving laws and be safe at all times while performing missions over a 60 mile course in under 6h [3]. This set of requirements was vastly different from previous Grand Challenges, and arguably more challenging, owing to the focus on interaction with other vehicles. Previous installments in the Grand Challenge were primarily