I am usually wide awake by 6 but today, I think the entire team was tired as we left at 7 AM. George Cahen's favorite the night before - the "Olive Garden" - was the fuel that carried us into the day.

Day 3 102707 (2)

Team Jefferson would not be running until the afternoon - in Test Area C: Focus on intersection and route re-planning. Paul and Dave Rubarsky took the robot to our offsite testing area to work on route re-planning.

Professor Cahen and I were charged with intel for Test Area C. Test Area C consits of a curvy diameter with a single spoke through its center. By taking three revolutions about the spoke, vehicles were confronted with six increasingly devilish intersections.

Day 3b 102707 (9)

Pete B and Mike Britt-Crane went over to Test Area A - moving traffic and merging.

Meanwhile a team that included Jarvis, Krishna and newest member Jeff Ebert, stayed at the hotel and worked all day on debugging a lane following routine.

George and I took notes on the Area C situation, sketched the plan, and tried to infer what was happening out of sight. Spectatorship was pretty poor here, we could peer at the closer intersection through a chain length fence.

Intersections became more complex, with the sixth intesection including a "double-precedence" problem: wait for the first vehicle at the intersection, but not for the one that pulls up as you allow cross-traffic to pass. Many teams missed waiting for at least one vehicle.

Watching these bots go live and start to navigate traffic is inspiring. It makes the observer feel connected to the future ... when many cars on our roads will not be wasting the time of a human operator. Who will see these cars? These roads? My children? My grandchildren?

It was also interesting to watch the obstacle vehicles -- which had real drivers. Race officials call these "people-driven vehicles". The cars are gutted out Ford Tauruses with roll cages.

Day 3 102707 (19)

The hired drivers are off-road racers or stunt drivers. There is kind of an interesting choreography they do to position themselves for the challenges. After the bot passes an intersection, they swiftly pull out of lurk driveways, and circle the intersection like pirahnas, stopping and gliding backwards into position. Running the trial for them is like running choreography in a play.

Team Terra Max seemed to do quite well.

Day 3b 102707 (10)

At the end of the intersection challenge, out of sight - but apparently - the vehicle is confronted with a full obstruction, and forced to re-plan a route to a destination. Terra Max nailed this, encountering first an obstacle, then the railroad arm on plan B, and finally a clear path to hit the point and finish.

Around 10 AM, we needed to get to the offsite team, including Paul and Dave and brief them on the trial. They finished up some de-bugging on route re-planning, took the brief, made some final tweaks... and then we did what people always do when confronted with this situation: we drove our robot to the robot base.

Day 3c 102707 (0)

Team Jefferson was second on deck when we arrived. We watched Desert Challenge winner Stanley run first. The Stanford team handled the intersection tests, and engaged route re-planning after the encountering the obstacle, but then the car wandered aimlessly through the course, apparently unable in its own estimation to reach that final waypoint ... until the 40 minute time limit expired.

Day 3c 102707 (2)

We had more wait time than we thought as Team Cybernet used its full 40 minutes. We were briefed, had our e-stop checked, and when we were called, we pulled right up and loaded our mission file easily in the required 5 minutes.

Tommy Jr. handled the intersection tests relatively adeptly, but he did fail to detect a left vehicle twice, as many other robots had done.

Day 3c 102707 (13)

After the passage on the last intersection, he initiatied route re-planning upon encountering the obstruction. Like Terra Max, Tommy tried approaching the railway arm as the next shortest route.

Unfortunately, Tommy did not detect the railway arm: he drove directly into it. The arm bounced off the hood, hit the instrumentation rack, and landed on the windshield.

Day 3c 102707 (16)

The vehicle was stopped and the team recovered it and hurried it back to the pits. The Discovery Team was in our face with cameras, and there was some frustration on the team that the robot did not receive a pause signal.

Day 3c 102707 (17)

The ops team sprang to action with the glass. We negotiated for a dealer in Ontario to hold a windshield for us after hours, and an installer in Victorville to meet us at the base. It wasn't trivial, but the installer delayed his poker game to help us out.

Day 3 102807 (92)

Tonight the team is exhausted but analyzing data for the fault. It does not appear that the railway arm was hidden in a blind spot, but rather, it appears that the route re-planning routine may have compromised laser data interpretation.

Meanwhile, we haven't had a complete briefing from the Area A recon, but we did see some video. It looks fiendish. Fifteen or 20 cars driving in a continuous loop. We heard that Sting Racing slammed into a K-barrier trying to merge.

We had a slight setback today, but we have to look forward to our encounter with Area A tomorrow afternoon with conviction and determination.