A Two-Layer Approach for Predictive Optimal Cruise Control

Bauer, Kai-Lukas; Gauterin, Frank

doi:10.4271/2016-01-0634

Cited by 14 publications

(3 citation statements)

References 7 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…After the ego vehicle completes a round of interaction with the environment, the policy neural network is optimized based on the trajectory information collected from this interaction (lines 8-19). When the control policy is at a high safety level, optimization is conducted using the TRPO algorithm (lines [11][12]; at a medium safety level, the PCPO algorithm is employed for optimization (lines 13-15); and at a low safety level, a linear backtracking method is used to minimize the safety value (lines [16][17]. Finally, the parameters of the reward and safety value neural networks are updated using gradient-based methods to minimize the gap between estimated and true values (line 18).…”

Section: Algorithm Overviewmentioning

confidence: 99%

“…Compared to traditional control theory, MPC offers improved control performance, and ACC technology based on MPC has made significant progress in recent years [ 11 , 12 , 13 ]. Starting from [ 14 ], MPC-based methods have become the dominant approach because MPC optimizes a multi-objective cost function, including fuel economy, driver comfort, and safety during driving.…”

Section: Introductionmentioning

confidence: 99%

“…Compared to traditional control theory, MPC offers improved control performance, and ACC technology based on MPC has made significant progress in recent years [11][12][13].…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Adaptive Cruise Control Based on Safe Deep Reinforcement Learning

Zhao,

Wang,

Che

et al. 2024

Sensors

View full text Add to dashboard Cite

Adaptive cruise control (ACC) enables efficient, safe, and intelligent vehicle control by autonomously adjusting speed and ensuring a safe following distance from the vehicle in front. This paper proposes a novel adaptive cruise system, namely the Safety-First Reinforcement Learning Adaptive Cruise Control (SFRL-ACC). This system aims to leverage the model-free nature and high real-time inference efficiency of Deep Reinforcement Learning (DRL) to overcome the challenges of modeling difficulties and lower computational efficiency faced by current optimization control-based ACC methods while simultaneously maintaining safety advantages and optimizing ride comfort. Firstly, we transform the ACC problem into a safe DRL formulation Constrained Markov Decision Process (CMDP) by carefully designing state, action, reward, and cost functions. Subsequently, we propose the Projected Constrained Policy Optimization (PCPO)-based ACC Algorithm SFRL-ACC, which is specifically tailored to solve the CMDP problem. PCPO incorporates safety constraints that further restrict the trust region formed by the Kullback–Leibler (KL) divergence, facilitating DRL policy updates that maximize performance while keeping safety costs within their limit bounds. Finally, we train an SFRL-ACC policy and compare its computation time, traffic efficiency, ride comfort, and safety with state-of-the-art MPC-based ACC control methods. The experimental results prove the superiority of the proposed method in the aforementioned performance aspects.

show abstract