Optimal coordination control of CAEVs under various traffic conditions
Problem to solve
- Developing a coordination mechanism that enables CAEVs to effectively cooperate in a traffic network while increasing overall efficiency, without significantly compromising the individual intentions of each vehicle

Figure 1. Concept of optimal coordination control
Technical challenges
- Resolving conflicts between the individual intentions of each vehicle
- Developing a coordination mechanism that can effectively operate across various traffic conditions
- Incorporating non-CAEV agents, such as human-driven vehicles and pedestrians, into the coordination framework
Specific research topics
- Developing a reinforcement learning (RL)-based coordination mechanism for CAEVs that ensures collision-free operations
- Designing a general framework for RL of CAEVs that can be applied across various traffic conditions
- Validating the effectiveness of the RL-based coordination mechanism in real-world driving scenarios
Experimental setup

Figure 2. Experimental setup
Control and AI techniques used in research
- Reinforcement learning (RL) algorithms
- Model predictive control (MPC) techniques
- Numerical optimization algorithms
Expected results
- The developed coordination mechanism shows promising potential to be adopted as a key technology for controlling CAEVs in the future.
Relevant references
- To be updated
Relevant research projects or grants
- 미래형자동차 핵심기술 R&D 전문인력양성 (한국산업기술진흥원, 2022 - 2027) [AI대학원 자율주행트랙]