The key points to note in these runs are the sustained speed of Terrahawk and, in NQE 4, we exercised the active suspension to raise and lower the vehicle. It was programmed to raise briefly after coming out of E-stop. This can be seen in the NQE 4 start video.
By the 4th run, DARPA was getting very efficient with the NQE runs and were putting out two vehicles on the course at the same time. Then they used E-stop on the second vehicle on the course to ensure adequate separation. For this reason, the DARPA controllers did an E-stop to TerraHawk just prior to our entering the tunnel, since we were barreling along and closing on the team ahead of us on the course. Coming out of this E-stop, Terrahawk did another suspension standup, as seen in the NQE 4 tunnel video.
The lack of certainty is simply a reaction to DARPA's demonstrated flux when it comes to notifying teams of practice field time assignments, and "line up to be escorted to the starting chute" times. For instance, an entire practice field time was lost yesterday (typical to expect no more than 1 practice opportunity every day and a half or so) because the DARPA official came to get us 2.5 hours EARLY and expected the team to be ready. Since the team was in the middle of fixes/troubleshooting and was not ready, DARPA told us that that meant we were forfeiting our practice opportunity. Nice...
We have reasonable confidence tomorrow's run will be in the afternoon because they reverse the line-up for each set of trials. We were 4th to last, then 4th up, then 4th to last, etc.
If you want to see TerraHawk's run, tomorrow 4 Oct is your last opportunity. As a spectator, you can just come to the Speedway and after going through the main gate on Cherry ave, go directly to the infield and there are signs pointing to the stands.
Lagging teams (those unable to complete the course) have been given a special RDDF (file with GPS waypoints) that they will run at tomorrow's qualification attempts. Special dispensation is being offered in the form of an easier course. DARPA has demonstrated, in many ways, willingness to give teams every opportunity to demonstrate what they have been designed to do in order to enhance the competitive base of the entrants.
Vehicles and a team representative must be present at the Closing Ceremony on 5 October; where all vehicles will be in place for observation.
These videos show that, in spite of lingering and emerging integration problems that we are working hard to solve, the TerraHawk vehicle, as a prototype, has the power, agility and precision control required of autnomous ground vehicles designed for DAPRA's goals. We are happy to have the opportunity to show this work in the NQE venue.
Good luck to all the teams who are still struggling to make the 20-team cut.
Then, with 40 minutes notice, we had to decide to take our first practice slot, to be followed immediately by a full NQE run. Although the known statistics of the unknown unknowns would suggest that we should do some practice runs and find remaining bugs before going for the NQE, Todd decided to go for the opportunity. It was a good decision.
After three good practice runs (two of them figure eights that ensured that we had no remaining hidden navigation errors), we took the vehicle to the NQE start queue and waited for more than an hour while the teams in front of us finished their runs. We were the last vehicle run of the day.
TerraHawk blasted out of the start chute with more speed than most bots and proceeded at full speed toward the first obstacle. We dramatically mowed down the first obstacle, a narrow gate made out of corrugated steel, making a lot of noise and dust. Very good (well, we lost some points, but it was worth it). We made it through the next obstacle and took the hill “decisively” according to the announcer.
The vehicle unfortunately exhibited increasing glitches as it traversed the high speed section, and we suggested shutting it down before it got to the end of that section. (One of those unknown unknowns that we know are lurking in there) We have lots of forensic log data and we will be tracking down that bug next.
All the networked computers are Linux-based. Getting the software and the networking for all these computers to work together has been a continuing adventure. In comparison, scratch-building a custom vehicle from aluminum billets and steel box beam stock was the easy part.
But now we are bending the code to our will, and we plan to do NQE practice runs on Thursday 29 September. Please see the 27 Sept (NQE first day) pictures we have taken of various vehicles including TerraHawk, in the NQE garage area, at our team member Aaron's site, http://www.startreker.com/.
It has been a while since we have updated the site. We will do a better job to keep you guys informed of our progress at the Speedway.
Meanwhile, we have collected many gigabytes of integrated sensor suite data in the desert, that we are crunching now to improve the sensor performance.
But really, he wanted to know what the heck it was that we were towing. We gave him the DARPA Grand Challenge Overview, which was slightly difficult without our slides and laser pointer.
Turns out, he writes software when he is not cruising in the Sheriff's car, and so the pivotal question he asked was, "You're not running Windows on that, are you?" We were able to answer that TerraHawk runs Linux. This greatly relieved the potential seriousness of the situation, avoiding the obligatory high speed chase and final standoff, but most important, he did not have to write a fix-it ticket for conversion from Windows to Linux. We were extremely relieved.
Our advice to other DARPA GC teams who may be running Windows: check your brake lights, license stickers, drive under the speed limit and hope you don't get pulled over -- else you could be hauled into court with big OS conversion task on your hands.
