This paper proposes a deep reinforcement learning (DRL)-based algorithm in the path-tracking controller of an unmanned vehicle to autonomously learn the path-tracking capability of the vehicle by interacting with the CARLA environment. To solve the problem of the high estimation of the Q-value of the DDPG algorithm and slow training speed, the controller adopts the deep deterministic policy gradient algorithm of the double critic network (DCN-DDPG), obtains the trained model through offline learning, and sends control commands to the unmanned vehicle to make the vehicle drive according to the determined route. This method aimed to address the problem of unmanned-vehicle path tracking. This paper proposes a Markov decision process model, including the design of state, action-and-reward value functions, and trained the control strategy in the CARLA simulator Town04 urban scene. The tracking task was completed under various working conditions, and its tracking effect was compared with the original DDPG algorithm, model predictive control (MPC), and pure pursuit. It was verified that the designed control strategy has good environmental adaptability, speed adaptability, and tracking performance.

To improve the handling stability of automobiles and reduce the odds of rollover, active or semi-active suspension systems are usually used to control the roll of a vehicle. However, these kinds of control systems often take a zero-roll-angle as the control target and have a limited effect on improving the performance of the vehicle when turning. Tilt control, which actively controls the vehicle to tilt inward during a curve, greatly benefits the comprehensive performance of a vehicle when it is cornering. After analyzing the advantages and disadvantages of the tilt control strategies for narrow commuter vehicles by combining the structure and dynamic characteristics of automobiles, a direct tilt control (DTC) strategy was determined to be more suitable for automobiles. A model predictive controller for the DTC strategy was designed based on an active suspension. This allowed the reverse tilt to cause the moment generated by gravity to offset that generated by the centrifugal force, thereby significantly improving the handling stability, ride comfort, vehicle speed, and rollover prevention. The model predictive controller simultaneously tracked the desired tilt angle and yaw rate, achieving path tracking while improving the anti-rollover capability of the vehicle. Simulations of step-steering input and double-lane change maneuvers were performed. The results showed that, compared with traditional zero-roll-angle control, the proposed tilt control greatly reduced the occupant’s perceived lateral acceleration and the lateral load transfer ratio when the vehicle turned and exhibited a good path-tracking performance.

The rolling control of a car that focuses on reducing the roll angle passively has limited performance of increasing handling stability, passing speed, ride comfort, and rollover prevention while turning. This project presents a method for controlling an automobile to tilt toward the turning direction using active suspension. A 6-degree-of-freedom vehicle model with a 2-degree-of-freedom steering model and a 4-degree-of-freedom tilting model is established. The active tilt sliding mode controller, which causes zero steady-state tilt angle error, is established after the desired tilt angle is determined by dynamic analysis. Simulation results confirm the effectiveness of the control method. The proposed controller reduces the perceived lateral acceleration and the lateral load transfer rate, thereby effectively improving handling stability, ride comfort, and vehicle speed, meanwhile decreasing the possibility of rollover while turning.

Automobile roll control aims to reduce or achieve a zero roll angle. However, the ability of this roll control to improve the handling stability of vehicles when turning is limited. This study proposes a direct tilt control methodology for automobiles based on active suspension. This tilt control leans the vehicle’s body toward the turning direction and therefore allows the roll moment generated by gravity to reduce or even offset the roll moment generated by the centrifugal force. This phenomenon will greatly improve the roll stability of the vehicle, as well as the ride comfort. A six-degrees-of-freedom vehicle dynamics model is established, and the desired tilt angle is determined through dynamic analysis. In addition, an H_∞ robust controller that coordinates different performance demands to achieve the control objectives is designed. The occupant’s perceived lateral acceleration and the lateral load transfer ratio are used to evaluate and explain the main advantages of the proposed active tilt control. To account the difference between the proposed and traditional roll controls, a simulation analysis is performed to compare the proposed tilt H_∞ robust control, a traditional H_∞ robust control for zero roll angle, and a passive suspension system. The analysis of the time and frequency domains shows that the proposed controller greatly improves the handling stability and anti-rollover ability of vehicles during steering and maintains acceptable ride comfort.

1. The obstacle-avoidance trajectory-tracking control of autonomous vehicles is currently a popular research area in the industry. Whether the obstacle is stationary or moving, it is necessary to p...

For suspension systems with asymmetric damping during the extension and compression strokes, the equilibrium position of the vehicle body will lift or drop. Inspired by this phenomenon, a novel idea of control the vehicle height by adjusting asymmetric damping via adjustable-damping semi-active suspensions is proposed. Meanwhile, a methodology to implement this idea is also presented. By accounting for nonlinear factors, such as the nonlinear constraints of the output damping force of an adjustable-damper and the logic control conditions of the asymmetric damping adjustment, this control system was designed as a hybrid system. Hybrid model predictive control theory was used to design this controller, and the switching control problem was transformed into a continuous-rolling finite-time-domain optimisation control problem. The feasibility of this idea was proven by simulations for single-wheel vehicle height control. The proposed idea makes it possible to use semi-active actuators for vehicle height adjustment to realise the control effect achieved by active actuators. Moreover, it has lower energy consumption, lower cost and faster response than active suspension, and provides a new method for short-term vehicle height and attitude control under some specific vehicle conditions, such as high-speed steering, and emergency obstacle avoidance, etc.