Junk Bot Journal Group3: Yannick Djoumbi, Krystal Reinwand, Derek Wedley Day 1 After reading the project description, the first concept was to move the light sensors from both face down, to one centered down, and one centered straight ahead. We were originally thinking to use a third motor, some way to move the front piece to grab the can. We were unsure how the motor would move, but the idea was to pinch the can to hold it. After trying to come up with how to build it and code it to do that, we scrapped the idea because it was going to be too complicated. We then thought of an easier way. Instead of gripping the can, we put two long Lego pieces straight in the front to hold the can still once it came into its path. Derek started working on the code. He had the light sensor aimed at the ground programmed to recognize the black line and back up. We didn’t have any issues with that to start. However, we still had multiple issues getting the second light sensor to recognize the can. Day 2 On day two, we still had the issue of recognizing the can. We tried multiple times switching the code. We tested the threshold of the can to the light switch and when we set up the arena, it would move past a can, knock it over without recognizing it, and wouldn’t play a tone. At one point, after switching parts of the code, the tribot stopped recognizing the line, or any code after that. When we tested it, it would randomly search until it hit the black line, back up one motor, and just spin around in reverse. We had to fix the code to tell it to turn both motors on. Eventually we resolved that issue, and by the end of the second day, our bot was both recognizing the can, and randomly searching within the black line. The tribot would slowly search around the arena, and when it recognized a can, it would charge straight at it to push it away. The design was further adjusted to add extra Lego pieces as a bar at the bottom to help support the can as it was being pushed away. Day 3 and 4 We started timing the bot and its ability to find multiple cans within three minutes. It would find one or two cans easily, and then would take the remainder of the time to find another. We adjusted the random timing, because the bot would always stop short of a can, and/or wouldn’t turn long enough to find it. With the two long straight Lego pieces pointed outward, it would knock the can over if it didn’t come upon it in a straight line. Our team took off the two long pieces and adjusted the design again so it wouldn’t keep knocking every can over. It became a V-shaped design with two Lego pieces. After altering the random time and extending it, and multiple rounds of testing, the tribot would find, recognize by playing a tone, and correctly push the cans out of the arena. On Day 4, Derek recorded a 2 minute 30 second video of the tribot correctly finding and pushing away all the cans.