Hebing Liu

This article proposes a three-layer framework for autonomous driving systems with optimized trajectory generation and tracking, ensuring optimal energy efficiency during the whole process. The third-order minimize curvature method is built in the first layer, which generates the personalized reference path by smoothing the human driving path to reduce its curvature. The energy-efficient strategy in the second layer is mainly through adopting the motor efficiency map to achieve the best efficiency interval of the operation motor's speed and acceleration, resulting in an optimized trajectory with motor efficiency consideration (OTHDM). The third layer integrates the OTHDM with the model predictive control module built based on vehicle dynamics to generate the optimal steering angle and achieve accurate tracking. The performance is validated through detailed numerical analysis and real human driving data collected by Honda Research Institute, including dozens of different driving scenarios and 19 343 tracks. Various test environments have been established in CarMaker. The experimental results indicate that our method can ensure low energy cost without energy recovery and battery hardware control in various scenarios. The energy saving rate can reach about 5%, up to more than 7%.

In this paper, an energy-efficient autonomous driving framework combining adaptive soft actor-critic (ASAC) evaluation reinforcement learning (RL), and model predictive control (MPC) strategies is proposed. It enables autonomous driving vehicles (AVs) to obstacle-freely track the desired path that the improved A*-based path planner generated. As the core work is to maintain energy-efficient performance, the motor efficiency map is employed to help reduce energy consumption quickly and accurately in an MPC-based predictive path generator and the ASAC-based path-tracking modules. At the same time, the final control input is adapted to the vehicle kinematics, and the generalization performance of the frame is improved. In this work, the CarMaker tool trains and tests on a map of multiple scenarios. Experimental results show that the proposed method is superior to other control methods in path tracking performance, running time, and energy efficiency. Compared to human driving data, this ASAC method can reduce the energy consumption of autonomous vehicles by 9.432% without any energy feedback mechanism. The method can also deal with various disturbances, such as complex road conditions and vehicle mass changes, and track the path accurately.

A two-layer model predictive control (MPC) algorithm with curvature adaptive is introduced and adopted in path tracking, especially for high-speed autonomous driving. Whether the vehicle can stably reach a safe driving speed in advance is the main consideration of rollover and speed-overshoot avoidance, especially when the trajectory with large curvatures. Thus, the outer layer of the proposed controller, which is built based on the vehicle kinematic model, generates an optimal vehicle velocity. And, the inner layer controller that is established according to the vehicle dynamics provides an optimal front wheel angle obtained combined with the optimal tracking trajectory generated by the path planner. The cross-track error, which is considered an important judging criterion of tracking and obstacle avoidance, is chosen here to validate the control performance. With the MATLAB/Simulink-Carmaker platform for modeling, the two-layer MPC has good performance in a continuous curve, simple obstacle avoidance, and complex obstacle avoidance scenarios. Notably, when the average driving speed reaches 108 km/h , the cross-track error can be controlled within 0.21 m when employing the proposed MPC method. In contract the conventional MPC method yields a cross-track error exceeding 3.4 m , while an LQR-based strategy results in a maximum error of 0.64 m . The proposed two-layer MPC algorithm with curvature adaptivity significantly enhances path tracking performance, particularly for high-speed autonomous driving. By effectively controlling the cross-track error and considering various driving scenarios, it successfully ensures safe and stable driving speeds, thereby mitigating the risks associated with rollover and speed overshoot.