A brief recap of the Overbot's performance at the 2005 Grand Challenge

Preliminary hardware problems

Diodes in the E-stop relay box blew out during the opening ceremony. We'd had this happen before, and were frantically trying to fix the problem. The "10 amp dual diode" in our E-stop relay box which allows it to draw power from either the vehicle battery or the generator turned out to actually be two 5A, 50% duty cycle diodes. These were replaced with two 12A stud-mount diodes mounted to the E-stop box case.

The NQE course runs

NQE course run 1

Run 1 at the NQE went well until the vehicle stopped in an uobstructed area between the two speed sections. The cause was a timeout in the "usercontrol" program, which starts the watchdog. "usercontrol" was left running, and the SSH session started from the laptop was left open when the laptop was unplugged. One hour later, the session timed out and killed the real-time system.

Workaround: quit "usercontrol" after starting the real time system,

NQE course run 2

Engine quit after first speed section. Apparently a false engine kill relay trip. Checked and tightened relay connections for engine kill relay,

NQE course run 3

Vehicle stopped in a tight spot. GPS position was off by about 1M, and waypoint and obstacle constraints conflicted. This problem had been noticed in run 1, and a new version of "MAP" had been built to allow 2m of extra width in the waypoints. But that version was not put on the vehicle, because I (Nagle) forgot to type "make install". Installed sofware for later runs.

NQE course run 4

A transient network error occured, and the software faulted and restarted. But this happened after the first two turns, and the "turn number" check added to allow courses that cross themselves prevented a restart from that point. Made all errors recoverable without a fault and restart as a workaround for the final run.

NQE course run 5

On run 5, the fused GPS/INS position jumped about 15m for no visible reason. The log data indicates good GPS data before and after the jump, and the GPS position did not jump. Only the fused position jumped. GPS/INS initialization mode was not entered, so MAP did not clear the map, and despite retries, recovery was unsucessful.

The Overbot was eliminated from further competition at this point.

Conclusion

Testing was inadequate. We lost too much time in the last weeks on the Fusednav problem, the CPU overheat problem, and the E-stop diode problem.

With adequate testing, we should have been able to run the NQE course, although slowly.