In this project, we designed and built an intelligent competing race car targeting at outperforming video race car games through the tangible car’s real-time interaction with the physical world. The car could either be controlled wirelessly with a phone as a steering wheel or autonomously track the lanes with four digital and analog reflective sensors. I developed a complete codebase for gravity sensing, microprocessor manipulation, real-time obstacle detection, wireless communication and user interface.

We proposed a Linear-Quadratic Regulator(LQR) controller which exploits the path information from camera data to help vehicles follow lanes. The high-level LQR controller generates the reference velocity and steering angle inputs by manipulating the deviation of the vehicle’s center of mass position from the lane center. A low-level PID control strategy is proposed to convert the reference values to motor and servo PWM signals with a first order model whose parameters are determined experimentally. The control system was tested with the Berkeley Autonomous Race Car with reliable performance.