Research paper
Noncommunicating agents · Selecting avoidance actions with a value network
A learned value network estimates time to goal from neighboring positions and velocities to support decentralized avoidance. Simulations compare path efficiency with ORCA when other agents’ intentions are not communicated.
Reviewed 2026-11-05 · Source and bibliography checked
- Original title
- Decentralized non-communicating multiagent collision avoidance with deep reinforcement learning
- Publisher
- IEEE
- Authors
- Yu Fan Chen, Miao Liu, Michael Everett, Jonathan P. How
- DOI
- 10.1109/icra.2017.7989037
- Publication date
- 2017-05
- Source language
- en
- Edition
- 2017 IEEE International Conference on Robotics and Automation (ICRA); 2017-05; DOI 10.1109/icra.2017.7989037
- Access
- Publisher metadata and open author manuscript
- Publication status
- Published
Related resources and engineering notes
Source overview
A learned value network estimates time to goal from neighboring positions and velocities to support decentralized avoidance. Simulations compare path efficiency with ORCA when other agents’ intentions are not communicated.
Use and review
Author abstract and bibliography checked. Path-efficiency gains do not establish universal collision avoidance or hardware safety.
Sources & editions
Decentralized non-communicating multiagent collision avoidance with deep reinforcement learning ↗
IEEE · 2017 IEEE International Conference on Robotics and Automation (ICRA); 2017-05; DOI 10.1109/icra.2017.7989037
ICRA2017 bibliography; author abstract https://arxiv.org/abs/1609.07845.