Vehicle following with minimal memory

A logical step on the path to full automation of an automobile is following a lead vehicle. Most studies in vehicle following use the chase vehicle's estimated speed and steering angle to calculate a series of approximate waypoints in an absolute coordinate frame to create a path for the chase vehicle to follow. However, state estimation requires increased computational ability as well as a robust noise model. This paper takes the minimal approach, attempting to find effective paths and steering strategies with minimal information storage through use of appropriate curve fitting parameters and measurements. The resulting control algorithm is tested through simulations on constructed test tracks and real-world road data.