Publication

• Pinxin Long, Wenxi Liu, Jia Pan. Deep-Learned Collision Avoidance Policy for Distributed Multi-Agent Navigation. IEEE Robotics and Automation Letters, 2(2), 656-663, 2017.

Bibtex

@article{long2017deep,

  title={Deep-Learned Collision Avoidance Policy for Distributed Multiagent Navigation},

  author={Long, Pinxin and Liu, Wenxi and Pan, Jia},

  journal={IEEE Robotics and Automation Letters},

  volume={2},

  number={2},

  pages={656--663},

  year={2017},

  publisher={IEEE}

}

Workshop Presentations

• Pinxin Long, Xinyi Liao, Wenxi Liu, Hao Zhang, Jia Pan. Deep-Learned Collision Avoidance Policy for Distributed Multi-Agent Navigation. NIPS Workshop on Learning, Inference and Control of Multi-Agent Systems, 2016.
• Pinxin Long, Xinyi Liao, Hao Zhang, Wenxi Liu, Jia Pan. Exploring Deep Networks for Reactive and Distributed Collision Avoidance Control among Multiple Robots. ICRA Workshop on Multi-robot Perception-Driven Control and Planning, 2017.

Please also check our new work on deep reinforcement learning for multi-robot collision avoidance: https://sites.google.com/view/drlmaca/

Project Summary

High-speed, low-latency obstacle avoidance that is insensitive to sensor noise is essential for enabling multiple decentralized robots to function reliably in cluttered and dynamic environments. While other distributed multi-agent collision avoidance systems exist, these systems require online geometric optimization where tedious parameter tuning and perfect sensing are necessary.

We present a novel end-to-end framework to generate reactive collision avoidance policy for efficient distributed multi-agent navigation. Our method formulates an agent’s navigation strategy as a deep neural network mapping from the observed noisy sensor measurements to the agent’s steering commands in terms of movement velocity. We train the network on a large number of frames of collision avoidance data collected by repeatedly running a multi-agent simulator with different parameter settings. We validate the learned deep neural network policy in a set of simulated scenarios with noisy measurements and demonstrate that our method is able to generate a robust navigation strategy that is insensitive to imperfect sensing and works reliably in all situations. We also show that our method can be well generalized to scenarios that do not appear in our training data, including scenes with static obstacles and agents with different shapes.