RMRL: Robot Navigation in Crowd Environments With Risk Map-Based Deep Reinforcement Learning
- 4 October 2023
- journal article
- research article
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Robotics and Automation Letters
- Vol. 8 (12) , 7930-7937
Abstract
Achieving safe and effective navigation in crowds is a crucial yet challenging problem. Recent work has mainly encoded the pedestrian-robot state pairs, which cannot fully capture the interactions among humans. Besides, existing work attempts to achieve “hard” collision avoidance, which may leave no feasible path to the robot in human-rich scenarios. We suppose that this can be addressed by introducing the local risk map and thus incorporate the risk map into the deep reinforcement learning architecture. The proposed map structure contains the crowd interaction states and geometric information. Meanwhile, a “soft” risk mapping of pedestrians is proposed to promote the robot to generate more humanlike motion patterns, and the riskaware dynamic window is designed to enhance the robot's obstacle avoidance ability. Experiments show that our method outperforms the baseline in terms of navigation performance and social attributes. Furthermore, we successfully validate the proposed policy through real-world environments.Keywords
Funding Information
- National Natural Science Foundation of China (62173248, 62233013, 62073245)
- Suzhou Key Industry Technological Innovation-Core Technology R&D Program (SGC2021035)
- Special funds for Jiangsu Science and Technology Plan (BE2022119)
- Suzhou Key Research and Development Project (SGC2021069)
This publication has 31 references indexed in Scilit:
- Socially aware motion planning with deep reinforcement learning2017
- Assessing the social criteria for human-robot collaborative navigation: A comparison of human-aware navigation planners2017
- Modeling cooperative navigation in dense human crowds2017
- Decentralized non-communicating multiagent collision avoidance with deep reinforcement learning2017
- Walking Ahead: The Headed Social Force ModelPLOS ONE, 2017
- DROW: Real-Time Deep Learning-Based Wheelchair Detection in 2-D Range DataIEEE Robotics and Automation Letters, 2016
- Socially compliant mobile robot navigation via inverse reinforcement learningThe International Journal of Robotics Research, 2016
- Human–Robot InteractionHuman Factors: The Journal of the Human Factors and Ergonomics Society, 2016
- Probabilistic mapping of dynamic obstacles using Markov chains for replanning in dynamic environments2008
- Dynamic Obstacle Avoidance in uncertain environment combining PVOs and Occupancy GridProceedings 2001 ICRA. IEEE International Conference on Robotics and Automation (cat. No.01ch37164), 2007