Competition day today. We had no code to write so didn't pick up our robot till 4. We were nervous as hell. The course seemed like it would be alright. We were scheduled 3rd to last, so we got to see what and whom our competition would be. Everything ran as it was supposed to, although the robot did not go far in the mapping round and no significant data was gained. The robot did run into the trouble we had hoped it wouldn't, where the pvc pipe ran into the goal post, but luckily we managed to score all 3 balls we picked up in quite the suspensful finish. The course was almost too easy, and many of our tricky/tight space manuervings were not needed at all. Overall it was a success, although losing to a bunch of math majors that would yell at the top of their lungs in lab was a bit hard to take.
Mapping still seems to be ok, so we'll see. Improved some minor things like pincher movement. Made sure marking and such was working as it was supposed to. Hard to get practice runs in as we need a big course and there are a lot of robots that want time. We basically called it good at 4ish, Renaud also ended up getting orcspy and botclient to run off his computer, so that was nifty as well. Contest is tomorrow, hopefully our bot will live up to its potential.
Stuff gets impounded tomorrow so time to kick into overdrive. Worked on improving the wander code so it's a viable option if mapping turns out to not work. Also worked getting the robot to drive to a given spot based on the shaft encoders. This seemed to work alright, although we figured out we definitely need to code in a way to escape if the robot gets stuck, which it does every so often. Escape code now works, we began to work on mapping. One round is basic wandering where we have arrays of interesting spots and the next round goes to the most interesting spot left. We also worked on getting the backup procedures for going to the goals and balls working more effectively. This seemed to be pretty successful. Didn't finish till 4ish so mapping, while it seems to work, could have some issues still. Got some sleep before the last push tomorrow.
Mock Competition part duex. Our robot fared better than most, escaping from the starting area and scoring a point. Unfortunately it then went into the orc recconnect failure shutdown. Found out this is due to us creating bazillions of orcs. Our bot still came in third place, behind the grand leader who managed to collect 3 balls. So I think we're doing alright. We retired to lab to improve our wander/wall-following code, this now works quite well. We may still have time for mapping, have to see how tomorrow goes.
IR sensors were giving us issues today. Also, our code seems to quit every now and then with an orc recconnect failure. I think it's our fault but not sure how. Ryan worked on adding the IR break code to the wandering/ball gathering stuff we had before. Renaud added springs to our pincher, so it won't get deformed if our robot runs into something, tries to gather two balls, etc. Everything seemed to work ok, not great. We'll see how it goes tomorrow.
Renaud did some work with the long-range IR sensor, getting it to recognize 'breaks' in the wall, places where it'd be good to send the robot. Ryan had exams.
Worked on fixing things that had gone wrong during the mock competition. Wandering code especially needs improvement.
Mock Competition day. Our robot worked quite well, although the first time our camera was not aligned with our ball catcher and so it acted as if it was picking up balls when in fact it was missing to the side. Our second time managed to pick up 5 balls, although the colors had not been adjusted and so it couldn't find the yellow goal. It looks as if our wandering code is good and we can start working on advanced mapping code and such. Also some minor improvements on the mechanics are necessary but shouldn't be too hard. Olivier and Antoine? left for France today.
Created the gyroscope version of wandering. Put together wandering and catchball. Should be able to complete checkpoint2 although code problems keep showing up (shocking, right?).
Worked on writing a better wandering code. It responds to infared sensing, although it'd be nice to have gyroscope coordination instead of solely timing. Also created code to pick up the ball, which seems to work quite well.
Finally stuff is starting to come together. Got the imaging code to work well, needed to re-collect all the HSV limits, but once that happened things were good. Code was also able to write out the code of a barcode found. Will need to later modify for multiple barcodes. The chassis was mostly finsihed by the end of the day, including the dumpbin on top which looks as it should perform well. Still haven't finished checkpoint 2, hopefully tomorrow or wed. Also got code working to run the robot by pressing the keys in botclient. Couldn't get it to do it with the camera opened but still neat to have remote control. Shaft encoders also started reading out data.
Continued work on the chassis, the lower part, including wheel, computer, orcboard, battery and pincer mountings. Wrote some code to drive the robot using the onboard joystick to facilitate shaft encoder testing. Attempted to write code that would move the robot by using the arrow keys in botclient.
Worked on getting our RGB stuff to work. The changes by the staff caused all sorts of difficulties as a lot of our values were switched around. Started building a real chassis out of wood. Got most of the bugs resolved with the color by the end of the day, although botclient apparently reads BGR, which causes our apparent colors to be off. Wrote code to test the shaft encoders although didn't get to test.
Wrote some code to run around the field and try to find balls. Added the gripping claw to the front. The code was far from perfect, the bot usually missed the red ball and ended up running off someplace. The gripper seemed to work if the ball was placed directly in front of the bot's path. The battery slipped off the bot while being picked up and took the orcpad with it. The computer worked a coupla times more then our hard drive died on us. Antoine? started work on the chassis as the rest of us pieced back the code we had lost. That'll teach us to back up more often. Got the bot back working but the the program does not function as we wish it would. Also figured out how to stick the program on the orcpad menu, although that wasn't entirely successful and needs to be looked at again. Getting the wandering-finding code done is pretty necessary though.
The image processing is now working (at least for the red balls). We found a strange bug we don't understand which seems to come from inherited class in java which was the problem ... We solved the pb but we do not clearly understand. The 2 shaft encoders are now mounted, they work pretty well. We made some coding for tomorrow : checkpoint 2, the wandering part is all right, we have to combine it with image processing. We made a rotating long range infra red sensor to do occupancy grid mapping, it seems to work, but no use for the moment.
Design review. The image processing software begins to work : found some bug but still pbs with x position for the object we see and too many object found ... strange. For the hardware part, the shaft encoder built yesterday has been tested : it seems to work pretty well but is fragile. We will build the other one tommorrow.
Continued working on the recognition software. After some brainstorming and coding, we find our code does not work. Spend a lot of time trying to figure out bug, in the end unsuccessfully. Will try again tomorrow. Renaud and Antoine? worked on wiring the other shaft encoder.
Finished to build the device to hold the motor and the home made shaft encoder. Will have to test it and if it works build a second one. Begin to work on "ascenseur à came". For the software, we should be able to recognize all the features on the playing field, find their center, area, shape ... but not tested yet.
Worked on today's assignment, say trying to get the robot to seek for the ball, go to it and stop. After a lot of small problems, reach by the end of the day to do the job correctly. Antoine? and Renaud worked on designing a pincher with only one servo. Finally, they got something quite heqvy, but really well designed and which already works. Ryan and I worked also on trying to identify different balls in the field, and the barcodes. The program isn't perfect but with an improvement of the different thresholds values for the green, the red and the black, it seems to be able to work.
Worked on getting the robot to indentify the necessary objects of the course from a picture. Took sample pictures and figured out color ranges for a combination of HSV and RGB values. Worked on having the Robot identify the green boxes on a barcode and report back how many there were. Antoine? and Renaud managed to get the homemade optical shaft encoder work. It's a mess of wires, but works very well and should be very useful.
Continued working on the optical encoder. This is proving to be somewhat tricky. Also developed code to drive the robot toward the red balls. Nearly done with code that will identify seperate balls and also the number of balls in view.
Antoine? and Renaud brought in a optical shaft encoder they made out of parts from a computer mouse. We soldered the short-range infared sensor and looked at its characteristics. We worked on getting the optical encoder to register in OrcSpy and Olivier worked on some Java code to create a timer.
Started things off today. Soldered the orcboard and the gyroscope. Wrote a program to print "hello world" on the orcpad. Also wrote unsuccessful code to take pictures with the camera. While a lot of the errors were eventually found to be the result of inclusion of an old library file, there were still some unresolvved errors with our code. We decided to work and getting the motors to move instead and wrote some code (this time successfully) to run the wheels.