Freeway Merging in Congested Traffic based on Multipolicy Decision Making with Passive Actor Critic
arXiv:1707.04489 · doi:10.1109/TIV.2019.2904417
Abstract
Freeway merging in congested traffic is a significant challenge toward fully automated driving. Merging vehicles need to decide not only how to merge into a spot, but also where to merge. We present a method for the freeway merging based on multi-policy decision making with a reinforcement learning method called {\em passive actor-critic} (pAC), which learns with less knowledge of the system and without active exploration. The method selects a merging spot candidate by using the state value learned with pAC. We evaluate our method using real traffic data. Our experiments show that pAC achieves 92\% success rate to merge into a freeway, which is comparable to human decision making.
6 pages, 5 figures. ICML Workshop on Machine Learning for Autonomous Vehicles
References in corpus (6)
- End to End Learning for Self-Driving Cars
- Trust Region Policy Optimization
- Learning Continuous Control Policies by Stochastic Value Gradients
- Continuous Inverse Optimal Control with Locally Optimal Examples
- Formulation of Deep Reinforcement Learning Architecture Toward Autonomous Driving for On-Ramp Merge
- Actor-Critic for Linearly-Solvable Continuous MDP with Partially Known Dynamics
Cited by in corpus (9)
- Merging Control Strategies of Connected and Autonomous Vehicles at Freeway On-Ramps: A Comprehensive Review
- A model of communication-enabled traffic interactions
- On Social Interactions of Merging Behaviors at Highway On-Ramps in Congested Traffic
- SA-Net: Deep Neural Network for Robot Trajectory Recognition from RGB-D Streams
- Deep Reinforcement Learning and Transportation Research: A Comprehensive Review
- A Survey of Deep Reinforcement Learning Algorithms for Motion Planning and Control of Autonomous Vehicles
- Anti-Jerk On-Ramp Merging Using Deep Reinforcement Learning
- Combining Reinforcement Learning with Model Predictive Control for On-Ramp Merging
- Interaction-Aware Behavior Planning for Autonomous Vehicles Validated with Real Traffic Data