Path Planning and Tracking for Autonomous Cars

Autonomous vehicles have the potential to increase safety, efficiency, and reliability in transportation services. Keeping that goal in mind, this work entails path planning and tracking for a small car model in both simulation and hardware. Specifically,given the starting position and orientation of the car, the objective is to navigate to a target position and orientation. A Dubins path is one way to map out the desired path, which can consist of a maximum of 3 segments, where each segment is either a curve or a straight line. This path can vary depending on how sharp of a turn the car can make, but all Dubins paths can be narrowed down to one of two general paths: a Curve-Curve-Curve path or a Curve-Straight-Curve path. Once the path is planned, simulation trials are run to track the path as closely as possible. A controller is implemented to adjust the steering angle based on the error between the car’s current and target locations. Once the error between the desired and actual path is within some satisfactory bound, the trajectory is executed on the real world model by mapping the simulation car’s steering angle value at each time step to the real system.