Acceleration, Braking, and Steering Controller for Polaris GEM e2

Abstract

This paper discusses the design and simulation of lane keeping and velocity controllers for a Polaris GEM e2. The controllers were designed to be implemented in Florida Institute of Technology’s entry for the Intelligent Ground Vehicle Competition self-drive challenge. The software and physical implementation of actuators were included to control the acceleration, braking, and steering. The accelerator pedal was replaced with a set of digitally controlled potentiometers, the brake pedal was replaced with a linear actuator, and the steering column was replaced with a stepper motor. Each actuator was controlled by a Raspberry Pi 3. A sliding mode controller was designed and its performance was evaluated against a PID controller for lane keeping. The sliding mode controller had a consistently smaller lateral error, especially during changes to the radius of the curve. Velocity control was simulated using a derived dynamic model for the longitudinal motion of the car. An adaptive control method was designed for velocity control and compared to PID. The adaptive control method reduced the error in longitudinal position during both acceleration and deceleration.