Hybrid Car-Following Strategy Based on Deep Deterministic Policy Gradient and Cooperative Adaptive Cruise Control

• Published in: IEEE transactions on automation science and engineering Vol. 19; no. 4; pp. 2816 - 2824
• Main Authors: Yan, Ruidong, Jiang, Rui, Jia, Bin, Huang, Jin, Yang, Diange
• Format: Journal Article
• Language: English
• Published: New York IEEE 01.10.2022; The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
• Subjects: Adaptive control; Autonomous driving; Car following; Constraint modelling; cooperative adaptive cruise control (CACC); Cooperative control; Cruise control; deep deterministic policy gradient (DDPG); Differential equations; hybrid strategy; Markov processes; Performance degradation; Reinforcement learning; Training
• ISSN: 1545-5955; 1558-3783
• DOI: 10.1109/TASE.2021.3100709

Deep deterministic policy gradient (DDPG)-based car-following strategy can break through the constraints of the differential equation model due to the ability of exploration on complex environments. However, the car-following performance of DDPG is usually degraded by unreasonable reward function design, insufficient training, and low sampling efficiency. In order to solve this kind of problem, a hybrid car-following strategy based on DDPG and cooperative adaptive cruise control (CACC) is proposed. First, the car-following process is modeled as the Markov decision process to calculate CACC and DDPG simultaneously at each frame. Given a current state, two actions are obtained from CACC and DDPG, respectively. Then, an optimal action, corresponding to the one offering a larger reward, is chosen as the output of the hybrid strategy. Meanwhile, a rule is designed to ensure that the change rate of acceleration is smaller than the desired value. Therefore, the proposed strategy not only guarantees the basic performance of car-following through CACC but also makes full use of the advantages of exploration on complex environments via DDPG. Finally, simulation results show that the car-following performance of the proposed strategy is improved compared with that of DDPG and CACC. Note to Practitioners-This article presents a new car-following strategy, which avoids the impact of deep deterministic policy gradient (DDPG) performance degradation on the system. In the proposed strategy, DDPG is replaced with cooperative adaptive cruise control (CACC) when the performance of DDPG is worse than that of CACC. Meanwhile, a switching rule is designed to guarantee that the change rate of acceleration is smaller than the threshold. Simulation results show that the performance of hybrid car-following strategy has been improved compared with that of only using CACC or DDPG. Moreover, the proposed strategy has the advantages of low computational burden, high real-time performance, and good scalability.