Nonlinear Longitudinal Controller Development and Real-time Implementation

Keywords longitudinal control, car platooning, back-stepping control, multi-surface dynamic sliding mode control, integral filters, realtime control code This report presents the development of nonlinear longitudinal controllers and their realtime implementation for Automated Highway Systems (AHS) at PATH, U. C. Berkeley supported by the project MOU 331. The demo'97 realtime code was first reconstructed for further development. Based on the new framework, the implementation and comparison of nonlinear controllers in the longitudinal control for cars platooning and other maneuvers have been carried out. Specifically, a dynamic sliding mode controller and a dynamic back-stepping multiple surface controller have been implemented, which are then compared to the previously implemented adaptive sliding mode controller. In the dynamic back-stepping control, a new type of integral filters have been developed, which were used for filtering measured signals and to estimate the derivative of these signals in realtime. These controllers have been implemented in realtime code and tested at low and high speed. Realtime run shows that each controller has its own characteristics. In general, the optimal dynamic back-stepping sliding surface controller has better performance by using the same set of measured data compared to previously implemented adaptive sliding mode controller. Emphasis has been put on the dynamic back-stepping multiple surface controller development.