Cone Navigation
The team programmed an RC car to drive autonomously around the MIT basement and avoid cones along its path. In order for the car to know its current location we used a particle filter that took the Lidar, wheel Odometery, and the map of the MIT tunnels to predict where it was. The map of the MIT tunnels was created using the SLAM algorithm.
Tunnel Racing
The team also programmed the RC car to race autonomously around the MIT tunnels. In this video you can see the car doing a lap around the STATA basement in roughly 35 seconds. In order to achieve these speeds that car had to be smart about how it drove. Driving at full speed around a turn would cause the car to loose control. In order to get the car to slow down, it used a look ahead point on the path two meters ahead of itself. Once it finds a point if figures out the turning radius that it would have at that point. Then with the turning radius it calculate the maximum speed the car can travel without slipping. Then the car decides what its acceleration/deceleration has to be to reach that speed in time. In order, to drive safely at those speeds a PID controller would not work. Instead the car used the Geometric Ackerman model to steer towards a pursuit point which it selected using its current speed.