Adaptive robust control for reliable trajectory tracking of autonomous vehicle in uncertain driving environment

External disturbances, parameter perturbance, data delay and control lag provoke significant model mismatches. If not properly compensated, they can greatly deteriorate the control performance of autonomous vehicle, such as reduction of tracking accuracy or even loss of stability in extreme. However, existing approaches barely consider these uncertainties together. In the light of this, an adaptive strategy is presented for trajectory tracking control of autonomous vehicle to simultaneously cope with aforementioned factors. First of all, given the dynamic or kinematic characteristics among path, vehicle and steering actuator, an integrated dynamic model is constructed. To handle the control lag of the steering actuator, a first-order model is utilized to approximate the dynamics of the steering subsystem, which is then integrated into the vehicle dynamics to reformulate the tracking model as a lag-free one. Then, the hierarchical robust tracking controller is proposed to acquire reliable control commands. To prevent the system breakdown in the presence of data delay, the delay-dependent criterion is designed via linear parameter varying technique and integral inequality approach. Moreover, the controllers also consider both the H infinity index and the guaranteed cost one to guarantee the effectiveness and robustness of tracking commands. Subsequently, to enhance the adaptability of algorithm, a feedback gains scheduling mechanism is proposed to adaptively tune tracking commands among different robust gains leveraging the phase plane approach. Finally, several comparative cases are conducted in the hardware-in-the-loop platform to verify that proposed strategy has better capability on trajectory tracking in uncertain driving conditions.

Automated summary

An adaptive strategy is presented to address the impact of external disturbances, parameter perturbance, data delay, and control lag on the control performance of autonomous vehicles. By constructing an integrated dynamic model, utilizing a first-order model to handle control lag, and designing delay-dependent criteria, the proposed strategy enhances adaptability and robustness. The strategy is validated through comparative cases in uncertain driving conditions.