Lab 8: Stunts!

Finally, I will use the Kalman filter from lab 7 to perform a stunt with my robot.

Task A: Position Control

Since I chose Task A in lab 6, my task for lab 8 is to perform a flip when the robot reaches the black padded area of the track and then drive in a straight line back to the starting line as fast as possible. In order to have the highest chance of for the robot to flip successfully, it must be driven at the fastest possible speed, which in this case is a pwm of 255. However, extremely high speeds run the risk of not getting enough raw TOF sensor readings to accurately gauge the distance of the robot from the wall, so the Kalman filter from lab 7 is incorporated to more accurately predict the robot's real time position. Below is a code snippet of the relevant code:


At first, instead of doing a flip when the robot reversed its velocity, it would start spinning to the left. I realized that the speed at which the right and left motors stop spinning and change direction is different, and the left motor changes direction more quickly than the right motor, causing the robot to spin instead of flip. I was able to determine this through recording slo-mo videos of the wheels as the robot changes direction in midair.



To fix this issue, I included a short delay in my code which drives the car backwards as follows:

This effectively helped both of the motors to switch direction simultaneously. Thanks to this adjustment, the robot was able to perform a flip as it switched from driving forward to backward rapidly. Here are some videos of the robot successfully performing the stunt and the associated position and pwm graphs:



Position vs Time PWM vs Time

The reason there is a 2 second gap in the data is that I used a delay to time when the robot should stop after performing the flip so that in case my robot goes rogue, it won't have that much time to become an occupational hazard. The data before the gap is from the sensor readings right up until the flip, and the data afterwards just shows that the robot is no longer facing the wall and is detecting whatever object is in its way.



Here are some bloopers:


Home