Max Robotics — Global Service Robot Directory + US Certification

About

Hengyue Liang is a robotics researcher specializing in robotic manipulation, deep reinforcement learning, and visual affordance learning, with a particular focus on solving complex, real-world grasping challenges. His most influential work, "A Deep Learning Approach to Grasping the Invisible" (2020), has garnered 115 citations and introduced the novel problem of locating and grasping initially hidden target objects through intelligent sequences of pushing and grasping actions — a significant leap beyond traditional grasping paradigms. Liang has also pioneered research into environment-aware manipulation, notably through his work on "slide-to-wall" grasping strategies, where robots leverage environmental fixtures such as walls and furniture to grasp objects otherwise infeasible with standard grippers. His development of Knowledge Induced Deep Q-Networks and target-oriented deep Q-learning frameworks demonstrates a consistent effort to integrate structured knowledge with self-supervised learning. Further contributions in attribute-based grasping enable robots to rapidly adapt to novel objects in cluttered scenes using transferable object attributes. Collectively, Liang's research advances the frontier of intelligent robotic manipulation, offering practical solutions to longstanding challenges in unstructured environments — making his work essential reading for robotics and AI researchers alike.