Performance Roundup

First Trial

Ultron did not perform his best on this Thursday night. The first movement was to turn around and run off the board.

Other teams also had a few issues. For example, a couple teams had drifting issues while their robot tried to navigate the board. Another fell right into the pit. However, some teams were successful in their search for the gold.

The motors and IR sensor functioned without error, although the light sensors could not be implemented. Both were broken (but only one replacement could be found in time) and our initial tests showed them to be inaccurate. The lack of light sensors explains why the robot tended to drift off the target path while turning.



Second Trial

Ken refined his algorithm after the first run, but the second attempt ended similarly to the first.

Plan B was then called into action– Theo quickly wrote his own algorithm and uploaded it to the robot. This simple algorithm successfully navigated the board, found the gold, and returned home. The robot decoded IR signals before each step.

Theo’s Algorithm:

read();
robot.go("N");	read();
robot.go("N");	read();
robot.go("S");	read();
robot.go("E");	read();
robot.go("N");	read();
robot.go("E");	read();
robot.go("W");	read();
robot.go("W");	read();
robot.go("S");	read();
robot.go("S");	read();

Lessons Learned

Did anything go wrong with Ultron? We ask ourselves this question in order to learn of our own strengths and shortcomings. Though we consider our robot to ultimately be a success, we do admit that the solution’s performance in competition was not as strong as we originally hoped.

  • We switched between several “Handy Board” microcontroller platforms during development. None of those boards ever worked, so we decided to switch to an Arduino board. This was a good decision. However, we wasted too much time with the Handy Boards at all.
  • The robot’s algorithm was not completed soon enough to test. We were able to test the robot on the game board and verify the motors and IR receiver, but we didn’t do any physical test-runs of the entire system until the first attempt in competition. An embarrassing failure resulted from the several undiscovered bugs in the algorithm.
  • Is Ultron intelligent? The algorithms prototypes that were discussed during early planning described system that was aware of its surroundings. Such a model could, if fully developed, make decisions about the world it knows when given relevant information. Is this any different than a human being making decisions? We fully explore the problem here.

To Improve:

  • Spend less time working on a board that will never work and much more time on a program.
  • Once an algorithm is developed, it should be tested on the robot in a competition scenario.