Wenlong Lu
Papers
5
Total Citations
34
H-Index
3
About
Wenlong Lu is a robotics researcher advancing the frontier of real-world deep reinforcement learning (RL) for contact-rich manipulation. His work centers on scaling RL from simulation to practical deployment, with key contributions in waste-sorting automation, contact-aware control, and simulation fidelity. In his highly cited work, “Deep RL at Scale: Sorting Waste in Office Buildings with a Fleet of Mobile Manipulators” (15 citations), Lu demonstrated a full-stack system for training and deploying RL policies on physical robots, tackling the challenges of bootstrapping real-world performance. He also introduced COCOI (12 citations), a contact-aware online context inference method that enables generalizable non-planar pushing, addressing the longstanding difficulty of adapting to complex contact physics. To improve simulation accuracy, Lu developed Midas, a multi-joint robotics simulator guaranteeing intersection-free, stable frictional contact. His research further explores robot morphology optimization for enhanced task performance and learning efficiency. With a growing citation impact and a focus on bridging simulation and reality, Wenlong Lu is shaping the future of autonomous manipulation in unstructured environments.
Research Focus
Key Achievements
Top Papers
- 1
- 2
- 3
- 4
- 5