Last weekend, in an abandoned military base, an hour away from Los Angeles, DARPA (the Defense Advanced Research Projects Agency) pushed the limits of robotics by running the Urban Challenge. The challenge is quite simple: to have
autonomous cars driving 60 miles in an urban environment while adhering to the California driving rules.
If you thought robots were small toys for your kids or vacuums for your house, think again! This year, DARPA's Urban Challenge robots were (somewhat modified) manufactured cars (and one truck) that you and I drive. No human were controlling the cars at any time (except for emergency stop). Only computers were allowed as pilots. Contrary to last year DARPA's Grand Challenge, where there were no other moving vehicles on the course while the robots were racing, this year's event was all about other cars in motion.
Over 100 teams registered for the Urban Challenge. Among them, DARPA selected 35 semi-finalists that were invited last week for the National Qualification Event (NQE). During the week, all 35 teams had to demonstrate that their autonomous vehicle was safe to operate. The golden rule was: no matter what you do, do not hit anything. Out of the 35, 11 made the cut and were allowed to race last Saturday for the $2 million Grand Prize.
Watch this brief highlight video of NQE operations - made available by the DARPA.
What seems easy and obvious for human drivers on the road every day is a fantastic challenge for a robot as our eyes are by far superior to any camera or sensing devices we ever invented. The best analogy is that each team is trying to program a computer to drive, with the sensors that can only see like an old lady. In fact, compared to human, these robots are really driving like old ladies (safe and slow -- The average speed was 14 miles per hour --).
“Vehicles competing in the Urban Challenge will have to think like human drivers and continually make split-second decisions to avoid moving vehicles, including robotic vehicles without drivers, and operate safely on the course. The urban setting adds considerable complexity to the challenge faced by the robotic vehicles, and replicates the environments where many of today’s military missions are conducted.
-Dr. Norman Whitaker, Urban Challenge Program Manager
The cars were standard cars (Land Rover, Toyota, VW, GM, Ford, Porsche, Lotus,) outfitted with a large number of lasers, cameras, and all kind of sensors. The price tag for some of these vehicles was well above 500K making them some of the most expensive cars in the planet. Some team had even two cars (one as backup). In addition, the programming of the computers (yes most of these cars have their own data center running in the trunk) took a large number of engineers and graduate students. The budget for the richest team must be over 2 million dollars but like any race sponsors are picking up the tab.
1st Place - Tartan Racing, Pittsburgh, PA
For this event, DARPA set up a complex course and hired 55 stunt drivers (with specially equipped cars) to create the traffic all around the robots. Each robot had 3 missions to accomplish including traffic circles, merging, four way intersections (which means detecting the intersection, detecting other vehicles at the intersection, defining who has the right of way and then pass at the right time), parking (yes the robot had to find a parking spot and park itself) and passing cars. In total, each robot had to drive 60 miles in less
At the end of the race, three teams came minutes from each other (CMU, Stanford and Virginia Tech) with CMU winning the Grand Prize. But all 35 semi-finalists did an amazing job to bring autonomous driving even closer to reality. These robots are like the first planes ever built a century ago. I might still take a couple decades (the number varies depending whom you ask) but we will have autonomous driving car that will drive us everywhere.
By Regis Vincent, a valued Contributor.
For comments on this article,
email us at firstname.lastname@example.org