COOR-PLT: A hierarchical control model for coordinating adaptive platoons of connected and autonomous vehicles at signal-free intersections based on deep reinforcement learning
Abstract
A two-layer hierarchical model is developed to coordinate adaptive CAV platoons. DRL is applied to implement centralized and decentralized control strategies. The optimal size of CAV platoon is determined by considering multiple objectives. COOR-PLT has satisfactory convergence performances and the capability of avoiding deadlocks. COOR-PLT outperforms state-of-the-art methods on reducing travel time and fuel consumption in various traffic conditions.
- Publication:
-
Transportation Research Part C: Emerging Technologies
- Pub Date:
- January 2023
- DOI:
- 10.1016/j.trc.2022.103933
- arXiv:
- arXiv:2207.07195
- Bibcode:
- 2023TRPC..14603933L
- Keywords:
-
- Connected and autonomous vehicle (CAV);
- Signal-free intersection;
- Adaptive platoon;
- Multi-agent coordination;
- Hierarchical control;
- Deep reinforcement learning;
- Computer Science - Machine Learning;
- Computer Science - Multiagent Systems;
- Electrical Engineering and Systems Science - Systems and Control
- E-Print:
- This paper has been submitted to Transportation Research Part C: Emerging Technologies and is currently under review