Pinxin Long

About

Pinxin Long is a robotics researcher whose work spans decentralized multi-robot navigation, autonomous scanning, and robotic manipulation. He is best known for pioneering the application of deep reinforcement learning to multi-robot collision avoidance, tackling one of the field's most persistent challenges: enabling robots to navigate safely and efficiently in complex, dynamic environments without centralized coordination or explicit communication. His landmark 2018 paper, "Towards Optimally Decentralized Multi-Robot Collision Avoidance via Deep Reinforcement Learning," has accumulated over 530 citations, while its 2020 follow-up on complex navigation scenarios has garnered more than 320, underscoring the sustained influence of this research thread. Long's contributions extend beyond navigation; his early work on quality-driven autonomous 3D scanning and coupled scene reconstruction demonstrated a strong foundation in perception and robotic autonomy. More recently, he has addressed practical challenges such as robot navigation in dense pedestrian crowds and developed optimization-based frameworks for autonomous excavator trajectory generation. With additional contributions to robotic picking systems, Long's research profile reflects a rare breadth — from theoretical reinforcement learning frameworks to real-world deployment — making his work essential reading for anyone exploring autonomous robotics and human-robot coexistence.

Research Focus

Key Achievements

Top Papers

1. 1
   Towards Optimally Decentralized Multi-Robot Collision Avoidance via Deep Reinforcement Learning

   533 citations · 2018
2. 2
   Distributed multi-robot collision avoidance via deep reinforcement learning for navigation in complex scenarios

   322 citations · 2020
3. 3
   Deep-Learned Collision Avoidance Policy for Distributed Multiagent Navigation

   169 citations · 2017
4. 4
   Quality-driven poisson-guided autoscanning

   78 citations · 2014
5. 5
   Autoscanning for coupled scene reconstruction and proactive object analysis

   76 citations · 2015
6. 6
   Fully Distributed Multi-Robot Collision Avoidance via Deep Reinforcement Learning for Safe and Efficient Navigation in Complex Scenarios

   69 citations · 2018
7. 7
   Getting Robots Unfrozen and Unlost in Dense Pedestrian Crowds

   68 citations · 2019
8. 8
   Optimization-Based Framework for Excavation Trajectory Generation

   58 citations · 2021
9. 9
   Towards Optimally Decentralized Multi-Robot Collision Avoidance via Deep Reinforcement Learning

   34 citations · 2017
10. 10
    DoraPicker: An autonomous picking system for general objects

    31 citations · 2016

Key Collaborators

Contact & Links