Accepted by AROB 2021. For letting agents in traffic simulation behave more like humans, we propose a unified mechanism for agents learn to decide various accelerations on deep reinforcement learning and generate a traffic flow behaving variously to simulate the real traffic flow.
deep-reinforcement-learning multi-agent-systems multi-agent-reinforcement-learning a2c-algorithm generating-traffic-flows
-
Updated
May 7, 2021 - GAML