DARPA Grand Challenge (2005)
The second driverless car competition of the DARPA Grand Challenge was a 212 km (132 mi) off-road course that began at 6:40am on October 8, 2005, near the California/Nevada state line. All but one of the 23 finalists in the 2005 race surpassed the 11.78 km (7.32 mi) distance completed by the best vehicle in the 2004 race. Five vehicles successfully completed the course:
|Vehicle||Team Name||Team Home||Time Taken
|Stanley||Stanford Racing Team||Stanford University, Palo Alto, California||6:54||First place|
|Sandstorm||Red Team||Carnegie Mellon University, Pittsburgh, Pennsylvania||7:05||Second place|
|H1ghlander||Red Team Too||7:14||Third place|
|Kat-5||Team Gray||The Gray Insurance Company, Metairie, Louisiana||7:30||Fourth place|
|TerraMax||Team TerraMax||Oshkosh Truck Corporation, Oshkosh, Wisconsin||12:51||Over 10-hour limit, fifth place|
Vehicles in the 2005 race passed through three narrow tunnels and navigated more than 100 sharp left and right turns. The race concluded through Beer Bottle Pass, a winding mountain pass with sheer drop-offs on both sides. Although the 2004 course had required more elevation gain and some very sharp switchbacks (Daggett Ridge) had been required near the beginning of the route, it had had far fewer curves and generally wider roads than the 2005 course.
The natural rivalry between the teams from Stanford and Carnegie Mellon (Sebastian Thrun, head of the Stanford team was previously a faculty member at Carnegie Mellon and colleague of Red Whittaker, head of the CMU team) was played out during the race. Mechanical problems plagued H1ghlander before it was passed by Stanley. Gray Team’s entry was a miracle in itself, as the team from the suburbs of New Orleans was caught in Hurricane Katrina a few short months before the race. The fifth finisher, Terramax, a 30,000 pound entry from Oshkosh Truck, finished on the second day. The huge truck spent the night idling on the course, and was particularly nimble in carefully picking its way down the narrow roads of Beer Bottle Pass.
National Qualification Event (NQE)
Of the original 195 applicants, initially 40 teams were selected to participate in the National Qualification Event (NQE). Three teams were added on Aug. 23, 2005 to the semi-finalist 40 teams who were selected from site visits, sending 43 teams to the NQE.
The National Qualification Event was held at California Speedway in Fontana from September 27 through October 5. The results of the NQE were used to cut the 43 teams down to 23 for the race on October 8.
Vehicle performances at the NQE were judged by (1) elapsed time to complete the course; (2) number of obstacles successfully passed without contact; (3) number of gates successfully passed. DARPA did not reveal the relative importance of these three factors. DARPA's final ranking of the vehicles, for purposes of pole position in the Grand Challenge Event (GCE), may have been partly subjective.
The results of the 2005 DARPA NQEs are shown below sorted top to bottom by runs completed and gates passed, compiled from DARPA's published NQE results. This ordering does not correspond to DARPA's ranking of team performance. (For example, this ordering does not reflect speed as an element of performance.) Teams highlighted in green were the teams that DARPA selected to participate in the Grand Challenge desert race.
|Team name||Run 1||Run 2||Run 3||Run 4||Finishes||Total gates||Desert race participant|
|Red Team Too||10||47||3||9||50||4||10||50||4||10||50||5||4||197||Yes|
|VA Tech Rocky||x||9||0||13||48||4||15||49||5||15||49||5||3||155||Yes|
|Va Tech GC Team||x||0||0||x||5||0||x||16||0||17||44||3||1||65||Yes|
|Indy Robot Racing||x||6||0||x||12||1||x||14||0||x||23||0||0||55||No|
|Indiana Robotic Nav||x||0||0||x||0||0||x||0||0||x||2||0||0||2||No|
There were four NQE runs. The above four major columns are sorted from worst to best runs of each team. "Time" signifies the time in minutes for a completed run and "x" indicates an incomplete run. "Gates" indicates the number of gates along the track that were passed (there were 50 total per run). "Obstacles" indicates the number of obstacles on the track that were passed (there were 5 total per run). "Finishes" indicates the total number of runs that the team successfully completed (there were 4 runs total). "Total Gates" indicates the total number of gates that the team successfully passed. The teams are sorted from top to bottom according to runs completed and then by total gates passed. Teams indicated as a "desert race participant" are those teams invited by DARPA to participate in the Grand Challenge desert race. Source data
For the first and second run, only 4 obstacles were present whereas for the third and fourth runs 5 obstacles were present. The NQE results presented in the table above show each team's runs sorted left to right from their worst run to their best run in order to better illustrate relative ranking (i.e. not in order of the actual run sequence). The ranking of who was accepted into the Grand Challenge desert race seems to be driven by number of gates passed. Beyond that, it is not known what criteria DARPA used to qualify the participants as acceptance was seemingly driven by decision of the Chief Judge.
On October 6, the selected teams transported their robots to the starting location. On October 7 the teams had a day to fix any portions of the robots broken in the course of transportation prior to the actual race.
A Google Talk video about the DARPA 2005 race is available at YouTube
The route to be followed by the robots was supplied to the teams two hours before the start as a computer file with GPS coordinates, one every 72 m (237 feet) of the route, with more frequent waypoints in difficult patches. Some teams used topographic maps and aerial imagery to manually map out and program precise path and speed settings. Once the race had started, the robots were not allowed to contact humans in any way.
Each robot started at a different time and was "paused" for different amounts of time during the race; DARPA compensated for the staggered start times and subtracted the pause time from each robot's total to derive its final official time. The $2 million prize was awarded on Sunday, October 9, 2005.
The 2005 competitors were much more successful than those of 2004; only one failed to pass the 11.84 km (7.36 mi) mark set by the best-performing 2004 entry, Sandstorm. By the end, 18 robots had been disabled and five robots finished the course. On the first day, Stanley from Stanford University, and H1ghlander and Sandstorm from Carnegie Mellon University, finished within minutes of each other, with Stanley crossing the finish line first. Kat-5 from Gray Team started much later, but finished in a comparable time. The race paused overnight with one competitor, TerraMax, left on the course at mile 83; TerraMax had the stage to itself on Sunday as it belatedly rumbled home.
The winner of the 2005 DARPA Grand Challenge was Stanley, with a course time of 6 hours 53 minutes and 8 seconds (6:53:08) with average speed of 30.7 km/h (19.1 mph). CMU's Sandstorm followed with 7:04:50 at 29.9 km/h (18.6 mph) and H1ghlander at 7:14:00 at 29.3 km/h (18.2 mph). Gray Team's Kat-5 came through at 7:30:16 with average speed of 28.2 km/h (17.5 mph). Oshkosh Truck's Terramax finished at 12:51 and would not have been eligible for the prize because it exceeded the ten-hour limit.
|#||Vehicle||Team Name||Team Home||Time Taken
|1.||Stanley||Stanford Racing Team||Stanford University, Palo Alto, California||6:54||First place|
|2.||Sandstorm||Red Team||Carnegie Mellon University, Pittsburgh, Pennsylvania||7:05||Second place|
|3.||H1ghlander||Red Team Too||7:14||Third place|
|4.||Kat-5||Team Gray||The Gray Insurance Company, Metairie, Louisiana||7:30||Fourth place|
|5.||TerraMax||Team TerraMax||Oshkosh Truck Corporation, Oshkosh, Wisconsin||12:51||Over 10-hour limit, fifth place|
|6.||DEXTER||Team ENSCO||ENSCO, Springfield, Virginia||DNF||Out of race at 81 miles; tire blowout after going off-course due to bent frame|
|7.||Spirit||Axion Racing||Westlake Village, California||DNF||Out of race at 66 miles; got stuck in sand, possibly after mechanical failure in suspension|
|8.||Cliff||Virginia Tech Grand Challenge Team||Virginia Tech, Blacksburg, Virginia||DNF||Out of race at 44 miles due to vehicle motor mechanical problems.|
|9.||Rocky||Virginia Tech Team Rocky||DNF||Out of race at 39 miles; a steep hill caused the oil in its generator to slosh around. The generator's low-oil sensor went off, cutting power to computers.|
|10.||ION||Desert Buckeyes||Ohio State University, Columbus, Ohio||DNF||Out of race at 29 miles|
|11.||DAD||Team DAD||Digital Auto Drive/Velodyne Acoustics, Morgan Hill, California||DNF||Out of race at 26 miles; LIDAR scanner failed, loose wire due to vibration.|
|12.||Desert Rat||Insight Racing||North Carolina State University, Raleigh, North Carolina||DNF||Out of race at 26 miles|
|13.||Xboxx||Mojavaton||Grand Junction, Colorado||DNF||Out of race at 23.5 miles|
|14.||Golem 2||The Golem Group/UCLA||Los Angeles, California||DNF||Out of race at 22 miles; software bug crashed main computer causing 60 mph rampage|
|15.||CajunBot||Team Cajunbot||University of Louisiana, Lafayette, Louisiana||DNF||Out of race at 17 miles; The motor on the brake actuator burned out when vehicle was paused for about fifty minutes.|
|16.||RASCAL||SciAutonics/Auburn Engineering||Thousand Oaks, California||DNF||Out of race at 16 miles; software problems|
|17.||Desert Tortoise||Intelligent Vehicle Safety Technologies||Littleton, Colorado||DNF||Out of race at 14 miles, instability in steering controller caused robot to drive offroad for a power pole.|
|18.||NaviGATOR||Team CIMAR||University of Florida, Gainesville, Florida||DNF||Out of race at 14 miles; faulty reporting of GPS accuracy caused robot to drive into a bush, where a bug in "blocked path" logic left vehicle unable to recover.|
|19.||Prospect Eleven||Princeton University||Princeton University, Princeton, New Jersey||DNF||Out of race at 10 miles. A code bug slowed down steering and throttle control, sending the robot looping in circles and narrowly missing a clump of reporters.|
|20.||Spider||Team Cornell||Cornell University, Ithaca, New York||DNF||Out of race at 9 miles; when pause applied, rolled up against a guard rail; when unpaused, could not back away from rail since it had no functional backing software|
|21.||Alice||Team Caltech||California Institute of Technology, Pasadena, California||DNF||Out of race at 8 miles; after GPS reacquisition, veered over barrier and towards media|
|22.||JackBot||MonsterMoto||Cedar Park, Texas||DNF||Out of race at 7 miles|
|23.||The Meteor||Mitre Meteorites||MITRE, McLean, Virginia||DNF||Out of race at 1 mile; dust interfered with sensors causing false positive obstacle detection|
DNF = Did Not Finish
|Wikimedia Commons has media related to DARPA Grand Challenge 2005.|
- Analysis of CART algorithm used by the winning Stanford team
- CART source code used to distinguish the road from surrounding desert
TV & video coverage
- NOVA: The Great Robot Race
- Pre-event videos of the competitors
- WMV Video showing the world through Stanley's laser-beam eyes.
- Google Video: Lecture by Sebastian Thrun on winning the 2005 GC
- CNN.com: Robots fail to complete Grand Challenge
- Journal of Field Robotics, Special Issue on DARPA Grand Challenge, Part 1
- Journal of Field Robotics, Special Issue on DARPA Grand Challenge, Part 2
- News from the qualifying events and team blogs
- Wired Magazine article on the DARPA Grand Challenge and Stanley.
- Popular Mechanics article on the DARPA Grand Challenge.
- Popular Science article on the DARPA Grand Challenge.
- Scientific American article on the DARPA Grand Challenge.
- NOVA: The Great Robot Race