Sun Also Rises

Tue, 10/30/2007 - 13:22

Janie and I stayed out at the J.D. Pierce site with Paul and Mike Myers until about 1 AM on Monday morning. Paul returned to testing at 5 AM.

Sunset brought us back to the pit to find Dave Rubarsky right where we left him. He had "fixed some things" in lieu of sleep.

Today we donned our Sun Java apparel as the marketing folks from Sun were to be visiting the site. We looked forward to spending some quality time with the voluble Jean Eliot, and the peripatectic Dave Hofert (who was sandwiching his visit between trips to Colorado and China).

Dave Hofert had come in from Denver and he hit a meeting and phone call in the Darpa big tent as we waited for Tommy to run in Area B.

Area B is the "neighborhood challenge". While this test is mostly invisible to the observer, the starting area is a rabbit's warren of twists and turns that nettled many starters.

Tommy sailed out of the start, after making a perfect stop at the stop sign. He sailed around the traffic circle and towards a narrow barriered chute that would take him into base housing. The chute's diminishing perspective fooled many robots into perceiving a complete obstruction, and Tommy did not appear to be an exception.

The team reset the robot with the judge's permission, and then Tommy proceeded into the neighborhood. While the test becomes less visible at this point, Tommy ultimately made contact with a wall and damaged a sensor.

Certainly the 2007 Challenge has had a much greater degree of sensor trials than the previous race, and its had an impact on Tommy. We are consistenly finding waypoints placed off track, under curbs, inside or behind walls. This places a greater tax on less tested routines, such as curb feeling and lane detection: routines we would supress under better circumstances due to a the lack of testing. Much of our testing time, is being spent determining if we feel confident in deploying these routines under the difficult race conditions.

Back at the pit, John the Satanic Mechanic, with his assistants Pete Bonani, Mike Myers, and Mike Britt-Crane immediately set on Tommy Jr and began to disassmble the robot, repairing and replacing damaged foils and sensors.

Meanwhile we briefed the Sun team on the race, and the day's events, and Dave and Jean spent some time walking the site and in particular Test Area A.

The site is a mini trade-show with a few companies displaying their wares. This company asserts it can turn your car into a robot in four hours:

Though I expect the "four hour job" is more of a strap on actuation than true autonomy. True autonomy requires extraordinary machine intelligence that is the subject of Darpa's interest.

Dave and Jean very graciously took the team to dinner. We had a really nice time and covered a wide range of topics from the great advances being made in Java (check this out: http://java.com/en/everywhere/ ... did you know you can get java on a card? - nice stocking stuffer), to Jean's latest adventures at Sun, her move to San Francisco, and Dave's upcoming trip to Shanghai. Dave eagerly took tips from Jarvis, who hails from Shanghai, on what to do in his free time during his upcoming oriental urban adventure.

Festivity at dinner was muted by the challenges of the continuing race, and the absence of Paul, Pete, and Mike Britt-Crane's who, as you might imagine, continued to work at the test site, preparing for tomorrow.

Don and Judy Perrone also went out to the test site, and used their rental car, according to Judy, as field-ready robot fodder. Judy claimed she did not get bumped.

Tuesday morning finds us racing to Area C to recon any course changes before a 9:30 AM run.