Josh Kline's Robotics Project 2

[ Back to Index. ]

Problem Statement

I will be working on Lab Assignment I -- The Cleanup challenge using the RCX.

Notes

• The arena has edges on two sides only. That means the debris cannot be pushed off the other two edges. I have not come up with a way to avoid debris being pushed up and left against a wall.
• I have used basically the same physical robot from project 1 namely perspective my implementation of "Hank the Bumpertank" from the The Unofficial Guide to LEGO MINDSTORMS Robots.
• Debris to be cleared from the table serface consists of balled up paper.

Progress

1. I finally was able to get the brickOS to work properly. This was completed between the project 1 and project 2 periods.
2. I made some modifications to perspective including making the front fender larger to enable pushing of debris. I also added a light sensor that points downward and is located slightly behind the front fender. Pictures will be posted at a later date.
3. I created an initial program using the regular mindstorms operating system. This has demonstrated success in the converage task although it does not clear the arena quickly. The algorithm used is purly reactive and does not maintain any internal state.
   • If the front fender is hit hard enough it sets off the left or right bump sensor. This causes the robot to backup and turn slightly in the opposite direction.
   • IF the light sensor detects a white area (the masking tape on the border of the arena) the robot backup up and turns 180 degrees.
   • For the table top in question (the arena) imperically tests showed a wait time of 3.0 seconds will the motors are driving in opposite directions will produce a 180 degree turn.
4. Testing revealed that the table break in the middle of the arena makes a bump in some places. This hindered the double-bumper design because the bumper came too close to the table. I switched back to a single bumper design which should still be able to push paper balls.
5. I wrote a brickOS program to display the numeric value coming from the light sensor. Readings on the table are about 0x19-0x1a while readings on the tape are about 0x1e-0x1f. Readings on the brown stip dividing the tables are some other value which I just forgot but I concluded they will not pose a problem. I will check if that is true with further tests.
6. Implemented the same algorithm in c using brickOS libraries.

Goals/Todo List

• Find a way to prevent it from getting stuck in corners. Possibly remember each time a bumper gets hit and if the hit patern is alternating within a short period of time increase the angle of the turn, or reverse the direction of the turn.
• Possibly find a way to better mount the light sensor to get a wider range of values.