Our Log
Week One:
November 3, 2008
This was our first meeting with our group and our Wumpus Master, Evan. At this meeting received all of our materials, we were introduced to the Wumpus World Problem, we installed the needed programs on our computer, and we started on the design of the robot. We played around with different designs, trying to decide which was the most stable for movement. We also divided into smaller teams – the constructors, the coders, and the website designer. Alan and Mike were named primary constructors, Jordan and Amanda has joint charge of the website, and Amanda and Patrick were put in charge of designing our robot’s brain.
Week Two:
November 10, 2008
Today, we built the first prototype of our robot: a two wheel system with a glider in the top right corner and the bottom left corner, with wheels on the left and right sides. We had a gear attached to the motor with a chain that powered the wheel. This did not work very well – the chain did not fit exactly and we had very jumpy motion. We decided to do a line drive, with the wheels attached directly to the motor. We also built a face for our robot that resembled a clown and placed it on the front. The majority of our meeting was spent on building the robot, while the coders thought of a place to start. They were able to determine the basic moves and calculations that if would need to make, but did not write any down yet in code form. We closed the meeting with a fully built robot and decided that next week we would start on movement.
Week Three:
November 17, 2008
This week we were able to get the robot moving forwards and turning to the left and right. (Check out the video on the right - "Crabbot Walks!". Here we finally got Crabbot to take its first steps.) We decided that it would never need to move backwards, since it could easily turn around and move forwards. We figured out that full power on the motors was too fast for any practical movements, so we decided to use around 20% power for most movements. We started to determine the length of time for each of the movements (turn left, turn right, move forward) so that we could get the movements to be accurate over repeated tests.
November 19, 2008
On Wednesday, the Amanda and Patrick met to discuss the wumpus problem and write up an algorithm. They were able to come up with what they believed to be a fully functioning algorithm. Over the next week, the algorithm came to life in Interactive C.
November 22, 2008
Amanda brought the robot back to her dorm and worked on perfecting turning and forward motion. She was able to have it turn 90 degrees left and right with consistent accuracy. The robot could move in squares and back and forth along a line (check out the video at right "Crabbot can move in straight lines!"). This allowed the group to start working with the light sensors the next week.
Week Four:
November 24, 2008
This meeting was geared towards getting the light sensors working. We knew that we could use the light sensors to determine the boundaries of the square so we would know when we are moving from one to the other. We first thought to put one on the front and one on the back so that we would know when we crossed the line. It was then decided to put both on front so that we would be able to straighten ourselves out as we were traversing the board. If both light sensors hit white at the same time, we knew we were going straight. Otherwise, we would move only one wheel until we were lined up and then continue on our way. This proved more difficult than we thought, and it took the full meeting to get this to work. We built two holders at the front of our robot to hold the light sensors that had an odd resemblance to pinchers on a crab…which gave us the inspiration for the name of our robot – Crabbot.
Week Five:
December 1, 2008
In our final week, we knew we had to work fast to get our robot to work, so we spent this last week making adjustments and putting the finishing touches on the algorithm. We were able to start testing the robot on the board that it would be competing on, so we did full testing on the correction algorithm.
Alan, Amanda, and Patrick met again that night to work on the robot. They were decided to add “legs” to the sides of our robot, and it took the form of a true crab. They were able to get our robot to do figure-eights (check out the two videos on the right) on the board with a correction every time it crossed over the line. We also decided to increase the power on the motors as it was making corrections because it often got stuck as friction became too great. We tried to get the IR signals working as well. For some reason, our robot did not receive the signals, so we decided to create a 2-D array of integers to record all the IR signals on the board and our program would call a function that would retrieve these numbers. Still left to complete: finishing touches on algorithm.
December 2, 2008
The group of three met again to work on the problem at hand. The algorithm was complete, so we decided to give the robot a test run. We got several run time errors as we realized the code had bugs in it that the compiler did not pick up. We spent most of the time debugging. We also added small wheels to the front of the crab to decrease the friction along the table.
December 3, 2008
This was our last day to get it working, so Amanda and Alan met in Carnegie to fix up what was left. We found two more red “legs” and added it to our robot to replace the blue ones that we had. We were able to solve some simple mazes, but it had troubling returning home. We decided to keep on testing other mazes and work on the code at home. We tried the IR signals again but our robot could still not receive them.
December 4, 2008
Patrick and Amanda went to work on the finishing touches after Minds and Machines class on Thursday. We were able to fix our code so that it successfully returned home and we were beginning to feel more confident that we would be able to finish on time. We tested some more puzzles and we came into a small problem when it had to backtrack more than one square at a time. In all of tests, this only occurred when there was a wumpus on the board, so we thought it had to do with the presence of a wumpus. For a quick fix, we just treated the wumpus as a pit and determined that our problem was solved. Unfortunately it was not, and this was shown to us when we were at the Lego Round-Up later that night. Magically our IR receiver worked so we were able to use the remote in the competition.
To see how we did, check out the Round-Up Results.