Saturday, August 10, 2013

The stunning conclusion

Ladies and gents, we have finally made it back from competition. With a stable internet connection and free time now on our side, we would like to provide an in-depth update on what happened. So let's rewind everything back to 8 am on the morning of August 7th, 2013.

You could sum up our mindset at this time as one of panic and desperation. The preceding night was nothing short of awful; at least one of everything we owned broke along the way. Things that were basically unbreakable broke, and stuff that used to work suddenly didn't. It took the entire night just to break even and get back to where we were before our disasters. So at 8 am we knew that we still had a lot of tuning to do, but we were at a spot where we had a chance again.

We knew that the competition flights began at 10 am, and also that we would have 4 flights and one pass. In our position we determined that our best chance was to leave four people at the warehouse to continue testing until the vehicle at least had a chance at mission success. Meanwhile the other eleven people would go to the competition site to offer our pass and stall. We figured we were willing to miss up to two of our competition flights in exchange for testing time, if that's what it came down to.

And so we went forward with our sub-optimal and scary plan. Our brave four stayed behind and tried to tune the un-tuneable, while our scouts went ahead to report back on the situation.

When 10 o'clock came around the judges officially assigned flight times, and we learned that we would be flying sixth of six teams. This was taken to be a great thing, because we still weren't quite where we needed to be. A less great thing is that every other team seemed to be in a similar situation to us, because pretty much everyone passed on the first round. The trouble with this is that passing doesn't take very much time, so the first round of competition flights went really fast, putting us in a bit more of a pinch. After about another 40 minutes the flight-testers had to make a decision to either miss our first flight or go and start field testing. The consensus was that we would need all four of our test flights to tune the vehicle enough to have a shot at winning. So the testing folks headed over and showed up with the vehicle just in time to make the first competition flight.

Ye olde pink foam fortress-- The only known natural predator to our vehicle

Our vehicle getting ready to face its fears... or rather our fears

When our turn came up we set down the beast, held our breaths, and watched as it took off to do its job. And then, about ten seconds into the flight, it dropped out of the sky for what appeared to be no reason. 


If it had been three seconds later the vehicle would have been on the floor

This is a problem we had experienced the night before that didn't come up consistently, so we were really worried when it happened again because we still didn't know what was causing it. We had about an hour before our next flight, so the structures guys went to fixing things from the fall, and the software and circuits folks went to sniffing out the problem. The hour turned out to not be enough, though, and so we had to put our vehicle back on the line and hope that the problem would disappear like it had the night before. 

Again, vehicle went down, breaths were held, and the go button was pressed. The vehicle got into the air, and things were shaky, but it went forward towards the window. Our controller wasn't perfectly tuned, so everyone was tense as it struggled forward, and we cheered when it made it past the window. Then we all just about cried when the motors cut off and it fell again. The problem was more persistent than before and it looked like we wouldn't even get a chance to show off our software's new capabilities because we couldn't even explore the course.

WHY DON'T YOU WORK??

Our team went off again to solve the problem, now more determined because it was clearly not going to go away. And this time they figured it out. The issue was with our radio-controlled kill-switch; when the kill-switch is engaged it sends a signal to the vehicle to immediately kill power to the motors. Our problem was that there was more background noise at competition than we experienced on campus or at the warehouse, so the vehicle was receiving signals to kill motors that weren't actually being sent. We fixed the problem by increasing the number of consecutive messages that needed to be received before it would kill power.

The fix made us more optimistic for our third run. We set it down, turned it on, and were ecstatic to see it fly through the window and actually start exploring. It was still extremely nerve-wracking, because the controller was awful loose and we were wavering very close to walls. But the vehicle seemed to be doing well enough. 


This picture was exciting because the vehicle is inside the course


In this one the vehicle's not just inside, it's about to head into a new room

We still did have some problems, though. By nature, the vehicle is programmed to be afraid of walls and obstacles. This is imperative because it makes it such that the vehicle will plan paths that avoid any solid objects, which would kill it. In this test flight, though, the vehicle was tuned to be a tad bit too afraid of walls, so it had a very hard time going through door-ways. 

From the groundstation we could see that the vehicle really wanted to explore the course, but it was having a hard time. After a little while, though, the vehicle got through one of the doorways it was stuck in front of, and explored the room. It just so happened that this was the flash drive room, so everyone was really excited about how this could play out. Especially, because we were watching the map that it was generating and saw that it recognized the table. From here the vehicle is supposed to explore the table, look for the flash drive, and then retrieve it and leave. However, there was an error in the vehicle's exploration software, and for some reason it did not fully explore the table, so it did not find a flash drive. Even though it didn't quite do what it was supposed to, the vehicle thought it had done its job well and tried to move on to other parts of the course to continue its hunt for the flash drive. In the end its erratic behavior got the best of it, though, and it crashed into a wall and died.


"It looks like a dead cat" -- Joseph Potter

Based on what we saw on that run the software folks knew that there were some changes that had to be made in order to make our last attempt more successful. There was also the immediate problem of us having totaled our last vehicle. I mean the vehicle was completely demolished. Fortunately, we had planned for this, which is why we came to competition with three flight-ready vehicles. Due to time constraints, we hadn't tuned our second vehicle prior to competition, so we were in a bit of a bind there. Flight testing was not allowed at competition, except for the 15 minutes immediately before an attempt, so that would be our only chance to get things flight-ready. Until then, our dedicated programmers worked on figuring out why the vehicle didn't explore the table, and fixing it.

"Is it broken?"      "Demolished, actually"


 Two looks of intense frustration

After our third flight there was a break for lunch, which was really good for us, because it gave us even more time to repair things and try to get our vehicle flight-ready. Everyone was really busy at this point, because everything was broken. We had just demolished a vehicle, so we had to get the backup up to par, and the software folks had to make changes that would hopefully fix the weird behavior in the last flight. 

Time went by, we made our repairs, and finally it was time to set our vehicle down for the last flight. There was an extremely stressful 15 minutes of tuning, before we all held our breaths and pressed go. From there, we watched the vehicle venture forward for the last time. 

Oh you know, just tuning before our flight, what could go wrong?

Things started out the same as before-- shaky and terrifying. We watched the vehicle fly through the window, but all got very nervous when it made its way towards the smallest rooms in the test course. Somehow, the vehicle decided that different parts of the course were more interesting than the ones it chose for the third run. During its travels it ended up in the smallest room in the entire test course; a death-trap of a room that was roughly closet sized. The vehicle made its way in without a problem, but had a hard time navigating back out, and ended up crashing. Our hearts sank as we heard the sound of it crashing into those cursed foam walls, and we knew that it was over. 

And so ends our competition this year. Despite our hard work, we ended up not accomplishing our goal of completing the mission. However, we are still extremely proud of all of our work, and the strides we made from last year. Following competition was the awards banquet, where it was announced that our team won Best System Design, Best Presentation, and first place in the North American Venue. There is, however, an Asian venue as well, and it turns out that one of the teams there successfully completed the mission. That accomplishment marks the end of mission 6 of the IARC, which means that this was our last chance at the ultimate prize for this particular mission. However, the judges are currently finalizing the details of mission 7, and MAAV has been looking for competitions outside of IARC to compete in. So, even though we were unable to take home the grand prize, we have all learned a lot and are coming out of this as better people. There was a lot to learn from our four years of involvement with IARC mission 6, and we look forward to future competitions and to tackling more challenging tasks.

At this time we would like to sincerely thank everyone who has followed along with us and who has believed in us. It's the support of our peers, friends, and family that helps us to move forward. We would also like to thank our sponsors for all of their support. Without the folks at Northrop Grumman, Lockheed Martin, the University's college of engineering, and the departments of aerospace, computer science, and electrical engineering, none of this would be possible. Thank you all for following along with us this competition season, we look forward to doing it all again next year. 

The proud summer 2013 members of MAAV






No comments:

Post a Comment