This week Ramona and I have spent a lot of time in the lab. We ran into some problems with noise. The sensors would sample fine when the motor was not on, but when we turned on the motor, the sampling would get messed up. After doing some trouble shooting, I saw that the problem seemed to be with the op-amps I am using for my sensor circuit. I decided to try using an external power supply to see if this would help the issue. The external power supply completely eliminated the problem. Since the noise only seemed to be affecting the op-amps that are supplied by 12 Volts and no other components in my circuit, I decided to use another battery to supply just the op-amps, and this solved the problem of noise from the motor. I also tested the shaft speed encoder with the software, and it is also working well. Later we encountered what appeared to be a software problem, but then we realized it was also being caused by noise. This time the noise was from the linear actuator. While the wheels were turning, I noticed that the sensors were sampling way too many times, and this was causing false readings. As it turns out, the linear actuator produces a lot of noise, so when the wheels are turning, it is causing the sensors' external interrupt to be triggered at the wrong times. To fix this problem, we disabled the external interrupt for the sensors while the wheels were turning, and then enabled the external interrupt after the wheels were finished turning.

Finally, we were ready to go outside and do some testing on Thursday. The testing was pretty successful. The vehicle was operating how we wanted it to. It would turn slightly left when sensor 1 = sidewalk and sensor 2 = grass, and it would turn slightly right when sensor 1 = grass and sensor 2 = grass. However, when the vehicle would turn left and get into the grass, the motor was not going fast enough to keep the vehicle moving, or when the vehicle would turn right and get onto the sidewalk, the motor would be going too fast. Due to these differences in grass and sidewalk, the vehicle was having problems following the edge of the sidewalk smoothly. We made some software changes to adjust for the different speeds, and we also looked at changing the wheel angle. We obviously need a better control algorithm for keeping the vehicle going along the edge of the sidewalk. Another change we made was moving the sensors more to the left. This way the sensors will detect grass and sidewalk before the vehicle has to leave the sidewalk for the grass. Perhaps if we can keep the vehicle on the sidewalk at all times, then controlling the vehicle will be a little easier. Unfortunately, we ran out of time, so we will have to test the new mounting of the sensors next week. All in all, I think this week's testing was pretty successful. Let's hope next goes as well.

Return to Note links