Adaptive Actuator Delay Compensation for a Vehicle Lateral Control System

Steering actuator lag is detrimental to the performance of lateral control systems and often leads to oscillation, reduced stability margins, and in some cases, instability. If the actuator lag is significant, compensation is required to maintain stability and meet performance specifications. Many recent works use a high-level approach to compensate for delay by utilizing model-based methods such as model predictive control (MPC). While these methods are effective when accurate models of both the vehicle and the actuator are available, they are susceptible to model errors. This work presents a low-level, adaptive control architecture to compensate for unknown or varying steering delay and dynamics. Using an inner-loop controller to regulate steer angle commands, oscillation can be reduced, and stability margins can be maintained without the need for an accurate vehicle model. The Smith Predictor (SP) control scheme is implemented in the inner-loop to mitigate the effects of the communication delay between the controller and the steering actuator. An algorithm will be presented to estimate both the communication delay between the controller and actuator and the steering dynamics. These estimates will be used to adapt the inner-loop SP to maintain gain and phase margins while reducing oscillation. Estimating the steering lag allows the algorithm to compensate for unknown or changing steering dynamics and communication delay. Results are presented both from simulation and from real-time experiments on a vehicle outfitted with drive-by-wire (DBW) hardware.