Many of our projects are gearing up into production mode, including the acoustics and wiring projects, as well as the testing team.  We had our first in-water test of the semester two weeks ago, and it was a good opportunity to let the freshmen see exactly what Barreleye can do.  We have another test day after tomorrow, so we are working hard on getting everything together on time.  We had a slight scare at the last test when the end-cap got pushed out of the hull about an inch because of the new camera, and the wireless buoy finally started leaking.  We’ve decided to do everything right this time around and not use any of the last-minute hacks that happened right before the competition.

Another part of our strategy is more cooperation with the other teams in the Ware Lab.  We share the lab with almost a dozen other teams, and we have the opportunity to share skills and techniques with most of them.  We are unique, however, as being the only autonomous vehicle team left in the lab, and one of only two submarine teams, the other being the Human Powered Submarine team.  We are also continuing our relationship with The Mathworks, and have also gained several sponsors since the beginning of the semester.

The progress we have made is directly in line with our priorities for this year, and at this rate we will have a very competitive entry!

Anyhow, as it turns out, we placed 18th out of 30! Given our horrible score on the static judging (placed 28 out of 30), it’s a bit better than we were expecting. There were five teams that pre-qualified and still scored below us, which is strange, given that most of the performance scores were the same.

We definitely have a good shot at the top five next year, given that we’ve worked on this for a whole summer so far. Our new team members are excited and ready to take over the competition!

For your reference, here’s the entire final standings:

1. Cornell
2. UVIC
3. URI
4. UCF
5. UT @ Dallas
6. USC
7. FAU
8. U Maryland
9. Kyushu Institute of Technology
10. ETS
11. Amador Valley HS
12. iBotics
13. US Naval Academy
14. Delhi College of Engineering
15. San Diego City Robotics
16. NC State U
17. Saint George’s School
18. VT
19. U Toronto MDA
20. UF
21. ARVP
22. MIT
23. SPSU
24. Embry Riddle
25. KAIST
26. UWF
27. U Ottawa
28. Pacific Nautilus
29. U Colorado @ Denver
30. McGill U

We have spent the last few weeks unpacking and organizing the team, as well as recruiting fresh talent at local campus events. It looks like we will have a much larger team this year, and we are still welcoming new members.

We are already well on our way in preparing for a new year of AUV design, and we hope to field a top-five entry next year. I hope that you will join us again for what is sure to be another life-changing experience for all of us!

The last two days of competition were decidedly less stressful, since we completed our final run of the competition at8:00 Saturday morning and had little to look forward to for Saturday and Sunday.  Since we had made friends with the NC State team on Friday by street-racing them on the way back from Transdec, we got to know them a bit better and one of us eventually challenged the other to an AUV race.  We finished one last “practice” run to capture some more video (and almost hit the flare!), then we trekked over to the Kona Kai resort pool (via downtown San Diego, that’s another long story) to see who had the fastest AUV in town.

We all put our AUVs in the water and started driving them around, at which point we immediately had network problems.  While we waited on fresh batteries for our wireless buoy, the NCSU team had plenty of time to reconfigure all five of their thrusters to thrust forward instead of down (we only have 2 forward thrusters!)  Even though we were far outmatched in power (2 thrusters at 14V vs 5 at 24V (that’s 5 times more!)), our advantage in weight and hydrodynamics made up for the difference.  Unfortunately, we soon discovered that our heading control loop is not optimized for high speed, and we have to operate at 30% of maximum thrust to go in a straight line.  Undeterred, I attempted to steer Barreleye manually, and found that our thrusters are unbalanced and cause the AUV to turn sharply left when driven at full speed.  Seawolf (NCSU’s AUV) even managed to beat us while dragging one of their team members!

The next day we didn’t do much besides talk to people about our AUV and watch some of the finals runs.  It was still a lot of fun to hang out with everyone and talk shop.  After the competition was over everyone headed to the awards ceremony/party, which was held on the beach at the Kona Kai resort.  The awards were presented, and everyone had a blast (BananaPhone, anyone?).  As we were leaving, we heard that NCSU was planning to race Seawolf against a team member from the Maryland team.  We had to get going across the country the next morning, so we weren’t able to stick around and find out what happened.  If you know, please leave a comment!

I’m writing this from West Memphis, Arkansas (next to Memphis, TN), because it’s the first place I’ve had good internet access.  On the way here from San Diego we stopped by a lot of fun places like Las Vegas (saw the fountains at the Bellagio), the Hoover Dam (and bridge!), and the Grand Canyon (which is really big, btw).  Tomorrow night Barreleye will be once again safe and sound at the lab in Blacksburg.  Once I get there there you will hear a lot more from us.

On the surface Barreleye seems to have lost both races this year, but in the end we all had a lot of fun and made some new friends while we were at it.  The trip has been so much more than we expected, and we have learned more things about AUVs and the American Southwest than we thought possible.  We have a lot of new ideas and a whole summer of experience behind us now, so I think we’ll be the team to beat next year!  We can’t wait!

A quick trip to Arby’s later, we headed to the pool at the Kona Kai Resort (the official competition lodging). The pool was to be open from 8pm to 5am, so we wouldn’t have to leave early. We threw (gently dropped, really) Barreleye into the pool and started running the vision code.

We soon got the camera to detect the buoy, and after a quick change of the wrong batteries (and then the right ones) we were fine tuning it (and by we I mean Stephen). Stephen persevered and fine-tuned the controller numbers until it could find and hit the flare reliably. We finally have vision!

Enthused and dead tired, we retuned to the hotel around 3:30AM (PST). Since we had been testing, we threw the batteries on the charger to make sure we would have power for the 7:30 run. We tried to get a few hours of sleep, and were soon awakened by the two team members who weren’t up all night.

We drove straight over to Transdec and unloaded straight onto the launching platform. After installing the batteries and fixing one loose connection, we went straight into the water to begin the run. The first run looked good, and Barreleye very slowly dove straight through the gate. We excitedly ran around the pool to see if she would catch sight of the buoy. We were a bit dissapointed when she didn’t, but the diver dragged her back and we started again.

On the next few runs it didn’t dive, so I quickly recalibrated the depth sensor and we tried one final go. Yet again she powered straight and level through the gate, and yet again we were dissapointed (but definitely not surprised) to see that she missed the buoy.

All in all, we are very glad to have semi-working vision processing after one day’s work (thanks Simulink!), and we are looking forward to watching the finals tomorrow!

We decided that running completely untested code wasn’t the best idea, so we switched back to the validation gate program we practiced with.  When the diver started the run, Barreleye decided to go up instead of down, and didn’t dive.   I yet again fell back to the manual dead-reckoning system from pre-qualifying, and guessed that the heading to the gate.  Starting the run over, Barreleye headed off to the left of the gate, which was to be expected.  I adjusted the heading settings while Ben tried to capture some video, and we tried again.  This time the AUV acted like a drunk and twitched off towards the center of the pool.  This was a problem we had never encountered before and we had no idea what was happening.   After some loud troubleshooting, we discovered that Ben had decided to compress the video as it was being captured, and the compression algorithm was taking up all of our processing power.  We only had two minutes left in the run, so we decided to run it without recording any video.  This time she finally made her way slowly straight through the gate, and with only nine seconds to spare. We were elated that we had scored points, and headed back to the booth with much more relaxed attitudes.

After our non-failure during the first qualification run, we are working hard on finding the “flare” with the camera, and we are hoping to score more points in qualifying tomorrow!

Connectivity issues aside, we have been very busy the last three days here. We got off to a slow start on the first practice day (Wednesday), because we had a long to-do list when we got there, and it didn’t get much shorter. We first tried to fine-tune the ballast to make sure Barreleye could dive, then we tried to get the camera to work. Even though we thought the camera had been working, Ben discovered a flaw in the method we were using to interface it with the Simulink code.

While we worked on the camera problem, we decided to test the program for going through the validation gate. We prepped Barreleye and threw her into the testing pool. Soon after everything was set up at the pool we discovered that she wasn’t connecting to the network and we couldn’t run anything! Thinking it could be caused by an automatic routine maintenance routine, we left it for a while and had dinner.

We came back to find no improvement, so we borrowed a monitor from the neighboring team and hooked it up to the AUV. We were shocked to discover that it wasn’t a software problem at all and the computer wouldn’t even make it past POST (power on self test, or the first bootup screen)! Not knowing what to do about this unexpected and time-wasting problem, we took out the e-rack and hooked it up to external power. It worked! For some unknown (at the time) reason, the problem had magically fixed itself.

By that time we had used up all our time for the day, do we packed up and returned to the hotel. Although we hadn’t even put Barreleye in the big Transdec pool, we were optimistic that we could make it through the gate and decided to make it our top priority. Even so, we decided that we were too tired to get any more work done and went straight to bed.

This morning we all got up and headed back to Transdec for day 2 of practice. Today was the last day to pre-qualify, so we knew that we had to make it work well enough to get through the gate. The first thing we did after we unloaded the van was put Barreleye in the testing pool and test our ballast and our straight-line navigation code. After a little tuning it seemed to work fine, so we went to sign up for a practice time slot in the big pool.

Imagine our dissapointment when we discovered that we would have to wait five hours for our practice session! Since we didn’t want to touch the code that we had working, we just put the batteries on the chargers and had some breakfast burritos. We eventually realized that we needed to work on something useful, so we all set to the next items on our to-do lists. Everyone worked on various things for a while, and we talked to a lot of people about Barreleye and underwater robots in general.

At some point during all this, we encountered the same computer problem we had yesterday, so I set about debugging it. I eventually found that if I turned the motor controllers off before I turned the computer on, it would work. This approach requires that we turn on the motor controllers seperately via (somewhat unreliable) software, but it works.

We finally got called up for our run in the big pool, so we hastily threw in the batteries and ran the wagon all the way around to the launching dock (our booth is on the opposite side of the facility). After briefing the diver on how to handle Barreleye, they hoisted her over the rail and into the water We had tested our network on the competition side of the pool, but the practice side is on the other side of the bridge in the middle, which caused a lot more interference. We weren’t able to start the run, so we had to run back around and grab our router. By the time we were set up my laptop had crashed twice and we were 15 minutes into the half hour session. Since nothing else could possibly go wrong, we asked the diver to point Barreleye at the validation gate and hit the GO button.

Barreleye smoothly ventured straight into the green Transdec waters, only to veer to the right and miss the gate completely. We had the diver retrieve the errant bot and set about trying again. I made a last second decision to ditch the “smart” code and aim for it manually. The first two times I tried running the old program the AUV would turn sharply to the right, at which point Ben corrected a rather stupid erreor on my part. After replacing the values we were testing with in Blacksburg with new values for the Transdec pool, we got Barreleye headed off in more or less the right direction. We fine tuned the heading, and made one last run with Dave (the manager/official) watching. We watched Barreleye slowly but surely make her way in the general direction of the gate, and after some worriment on our part and some autonomous course corrections, she finally made it through!

We later discovered that we were the 17th team to pre-qualify out of 23 teams to make it through the gate. Elated and educated, we returned to out booth and set about making the program more robust. After a few more hours of coding we threw the AUV in the testing pool again and started playing a strange version of pong with it. We had several of us at one end of the pool and we made Barreleye drive to the rest of the team at the other end, and then back again. At one point, She had just hit the bottom of the pool when another team’s bot passed straight over her in a near miss. Several minutes later, yet another sub interfered with the compass and we nearly rammed it twice.

We decided that was enough for one day, so now we are back at the hotel working on the camera and preparing for qualifying and static judging tomorrow.

Be sure to check out the video updates on the competition website starting tomorrow afternoon (EST)!

There will be a live webcast of the competition finals (Sunday), so I will post the link when I get it.  Be sure to look for it and tune in Sunday afternoon.  Also, daily video will be posted each morning after the day of competition at the competition website.

I will post trip photos when I can, and look for photos of our testing today!