Shiyang Lu

Rutgers Sexual and Reproductive Health and Rights, Rutgers, The State University of New Jersey

Papers

Total Citations

H-Index

About

Shiyang Lu is a robotics researcher whose work spans robotic manipulation, tensegrity systems, and 3D scene understanding. His most impactful contribution is **ARMBench** (32 citations), a large-scale, object-centric benchmark dataset developed at Amazon for robotic manipulation in warehouse environments—addressing the critical challenge of automating operations with diverse, unstructured objects. Lu also advances control of **cable-driven tensegrity robots**, which offer exceptional strength-to-weight ratios and resilience but are notoriously difficult to control due to high dimensionality and complex dynamics. His work on **Real2Sim2Real transfer** (15 citations) leverages differentiable physics engines to bridge simulation and reality, while his **6N-DoF pose tracking** method enables precise state estimation for these deformable robots. In 3D perception, Lu introduced **OVIR-3D**, an open-vocabulary instance retrieval method that requires no 3D training data, allowing robots to identify objects from natural language queries. With a focus on bridging simulation and real-world deployment, Lu’s research directly impacts warehouse automation, disaster response, and generalizable robotic manipulation—making him a rising figure in embodied AI and robot learning.