Ming Liu
Hong Kong University of Science and Technology, ETH Zurich, City University of Hong Kong, Tongji University, University of Hong Kong, Monash University, University of Bristol, Shandong University of Science and Technology, HKUST Shenzhen Research Institute, Nanjing Forestry University, Engineering Systems (United States)
Papers
150
Total Citations
7,203
H-Index
44
About
Ming Liu is a prominent robotics and autonomous systems researcher whose work spans mobile robot navigation, simultaneous localization and mapping (SLAM), deep reinforcement learning, and autonomous driving. His most influential contribution, "Virtual-to-real Deep Reinforcement Learning" (2017, 800 citations), revolutionized mapless robot navigation by demonstrating that sparse sensory inputs combined with continuous steering commands could enable robust real-world deployment — a landmark advance in sim-to-real transfer. His sustained work on RGB-D SLAM in dynamic environments, culminating in multiple highly cited publications (2016, 2018), significantly improved the reliability of localization systems operating among moving objects. Liu further pioneered the intersection of federated learning and robotics through his "Lifelong Federated Reinforcement Learning" framework (2019, 266 citations), enabling cloud-connected robots to collaboratively accumulate and share navigation experience. His contributions extend to socially aware navigation using generative adversarial imitation learning, point cloud registration benchmarking, and monocular 3D object detection for autonomous driving. Collectively, his body of work has accumulated over 2,500 citations, reflecting deep and wide influence across the robotics and computer vision communities. His research consistently bridges theoretical innovation with practical deployment, making him a defining voice in intelligent mobile robotics.
Research Focus
Key Achievements
Top Papers
- 1
- 2Improving RGB-D SLAM in dynamic environments: A motion removal approach384 citations · 2016
- 3
- 4Challenging data sets for point cloud registration algorithms259 citations · 2012
- 5A deep-network solution towards model-less obstacle avoidance211 citations · 2016
- 6Motion removal for reliable RGB-D SLAM in dynamic environments202 citations · 2018
- 7
- 8Robotic Online Path Planning on Point Cloud166 citations · 2015
- 9Ground-Aware Monocular 3D Object Detection for Autonomous Driving160 citations · 2021
- 10A robot exploration strategy based on Q-learning network149 citations · 2016