Max Robotics — Global Service Robot Directory + US Certification

About

Zehang Weng is a robotics researcher whose work lies at the intersection of dexterous manipulation, deformable object handling, and interactive perception. His most influential contribution is DexDiffuser (2024, 24 citations), a novel framework that uses conditional diffusion models to generate, evaluate, and refine dexterous grasps from partial point clouds—a significant advance for robots needing to grasp complex objects with multi-fingered hands. Weng has also made foundational contributions to modeling scenes containing both rigid and deformable objects, developing graph-based task-specific prediction models (2021) that capture complex interaction dynamics. His work on interactive perception for deformable object manipulation (2024) addresses the dual challenge of manipulating soft objects while using that interaction to improve perception under occlusion. By creating simulation environments and novel datasets for these challenging scenarios, Weng is enabling robots to move beyond simple rigid-object grasping toward the nuanced manipulation skills needed for real-world applications like household assistance and manufacturing. His research is particularly notable for bridging generative AI with physical robot control, opening new pathways for robots to handle the soft, deformable objects that populate human environments.