Study On Driver-automation Shared Control Strategies For Intelligent Vehicles

Nowadays,driver-automation shared control is at the forefront and hot spot in the field of intelligent driving technologies.Driver-automation shared control framework can synthesize both drivers' and automated driving systems' capabilities(e.g.perception,decision-making,control,and etc.),in which the driver and the automated driving systems can simultaneously control the vehicle motion to complete dynamic driving tasks,therefore,it will be a solution to the technologies limitation,social dilemma,and driving safety facing to intelligent driving systems with the high level of automation,and improve the driving safety and comfort performances.However,there still exist many challenges in the previous related researches,including that the driver-automation control authority allocation strategies were designed based on the part of the driver-vehicle-road related factors,or the strategies were unreasonably designed,and it is hard to solve the driver-automation path and control signals conflicts in driver-automation shared control schemes.To solve these issues in driver-automation shared control systems,this paper focuses on four aspects,including the driving intention,driver-automation control authority allocation,driver-automation interaction,and driver-automation conflicts management.Then,five studies have been completed.The innovations and main contents of this thesis are listed as follows:First,to predict accurate driving intention results for the design of shared control systems,the driving intention data collection experiment is designed based on the driver-in-the-loop simulation platform,and the driving intention datasets are established.Then,the support vector machine and an inductive multi-label classification method with an unlabeled data method are adopted to construct driving intention prediction models.Besides,the driving intention prediction models are trained and optimized based on the constructed driving intention datasets.We also design testing experiments for the verification of the driving intention prediction models.The results show that the constructed models are effective to predict driver's driving intention.Second,a shared fuzzy controller is innovatively proposed to address the control authority allocation strategies towards path-following problems and it is constructed using the fuzzy control method based on the driver's lane-changing intention,risk assessment,and the evaluation of path-following performance.The linear-time-varying model predictive control method is proposed for the path-following controller,and the control signal is calculated to follow the reference path.According to the comparison results between the driver's lane-changing intention and the desired lane-changing maneuver,three different shared fuzzy controllers are designed to determine the control authority between the driver and automated driving system,namely,the consistent,advanced inconsistent,and lagged inconsistent fuzzy controllers.The results of case studies show that the proposed shared fuzzy controller can dynamically adjust the control authority between the driver and path-following controller according to the driving intention comparison results,improve the driver's path-following performance,and ensure driving safety performance during obstacles avoidance.Third,a hierarchical driver-automation cooperative path planning and shared control framework is innovatively proposed to address path planning and path following problems for different driving intentions.By comprehensively considering driving intention and driving environment information,an optimal cooperative path planning method is proposed to generate the reference path in accordance with the driver's driving intention for the path-following controller.To adjust the control authority between the human driver and the path-following controller,the cooperative driving modules and cooperative coefficients are designed based on the time to collision and the parameter that the time required from the start position of the lane-changing to the obstacle to collide.The experiment results show that the proposed hierarchical shared control framework can obtain the optimal cooperative planned path considering the driving intention and calculate the reasonable control authority allocation based on the collision risk assessment,which assist the human driver avoid obstacles safely.Fourth,to deal with the driver-automation conflicts,a driver-automation dynamic interaction strategy is innovatively proposed based on the game theory and the improved driving safety field.The interaction model between a human driver and the path-following controller is constructed using the noncooperative Nash game theory.The driving safety field is improved by adjusting the shape of the potential fields and introducing the coordinate transformation matrix,which makes the driving safety field adapt to different driving scenarios for the evaluation of the driving risk level of shared control systems.To address driver-automation conflicts in different driving risk situations,the driver-automation dynamic interaction strategy is proposed to adjust the path-following error weight of the human driver's cost function,which is designed based on the driving risk level evaluated from the improved driving safety field.Therefore,the condition of the Nash equilibrium between the human driver and the path-following controller will be changed.Comparative studies prove that this driver-automation dynamic interaction strategy can reduce the driver-automation conflicts and obtain the best performance index among three aspects,including the performances of the driving safety,control workload,and road curvature.Fifth,a two-layer fuzzy shared control framework is innovatively proposed to address both the control authority allocation and driver-automation conflicts.The motion planning controller for shared control is innovatively proposed based on the potential-field-driven model predictive control method and the driver-vehicle-road dynamics model,which converts the path planning and path following problems into one optimization problem for the automated driving system.The driver's control action and target path are considered for the design of the cost function of the proposed motion planning controller,and the tuning weight is also proposed to adjust the trade-off between the motion-planning related cost and the driver-related cost.Besides,a two-layer fuzzy shared strategy for the shared control is designed using a fuzzy control method based on the evaluation of driving risk levels and conflict situations,and the values of the tuning weight and cooperative coefficient are dynamically determined.The results of comparative studies show that the proposed framework can improve the driving safety performance during obstacle avoidance,and obtain the great performance of driver-automation conflict management.These research contents of this thesis can provide certain theoretical references and technical supports for the deep studies of the driver-automation shared control methods and the advanced intelligent driving technologies.