TR2019-142
Model-based deep reinforcement learning for CACC in mixed-autonomy vehicle platoons
-
- "Model-based deep reinforcement learning for CACC in mixed-autonomy vehicle platoons", IEEE Conference on Decision and Control (CDC), DOI: 10.1109/CDC40024.2019.9030110, December 2019, pp. 4079-4084.BibTeX TR2019-142 PDF
- @inproceedings{Chu2019dec,
- author = {Chu, Tianshu and Kalabić, Uroš},
- title = {Model-based deep reinforcement learning for CACC in mixed-autonomy vehicle platoons},
- booktitle = {Proc. IEEE Conference on Decision and Control},
- year = 2019,
- pages = {4079--4084},
- month = dec,
- doi = {10.1109/CDC40024.2019.9030110},
- url = {https://www.merl.com/publications/TR2019-142}
- }
,
- "Model-based deep reinforcement learning for CACC in mixed-autonomy vehicle platoons", IEEE Conference on Decision and Control (CDC), DOI: 10.1109/CDC40024.2019.9030110, December 2019, pp. 4079-4084.
-
Research Areas:
Abstract:
This paper proposes a model-based deep reinforcement learning (DRL) algorithm for cooperative adaptive cruise control (CACC) of connected vehicles. Differing from most existing CACC works, we consider a platoon consisting of both human-driven and autonomous vehicles. The humandriven vehicles are heterogeneous and connected via vehicleto-vehicle (V2V) communication and the autonomous vehicles are controlled by a cloud-based centralized DRL controller via vehicle-to-cloud (V2C) communication. To overcome the safety and robustness issues of RL, the algorithm informs lowerlevel controllers of desired headway signals instead of directly controlling vehicle accelerations. The lower-level behavior is modeled according to the optimal velocity model (OVM), which determines vehicle acceleration according to a headway input. Numerical experiments show that the model-based DRL algorithm outperforms its model-free version in both safety and stability of CACC. Furthermore, we study the impact of different penetration ratios of autonomous vehicles on the safety, stability, and optimality of the CACC policy.
Related News & Events
-
NEWS MERL researchers presented 8 papers at Conference on Decision and Control (CDC) Date: December 11, 2019 - December 13, 2019
Where: Nice, France
MERL Contacts: Mouhacine Benosman; Scott A. Bortoff; Ankush Chakrabarty; Stefano Di Cairano
Research Areas: Control, Machine Learning, OptimizationBrief- At the Conference on Decision and Control, MERL presented 8 papers on subjects including estimation for thermal-fluid models and transportation networks, analysis of HVAC systems, extremum seeking for multi-agent systems, reinforcement learning for vehicle platoons, and learning with applications to autonomous vehicles.