Bridging Wuxi and Silicon Valley through global robotics research

Zhu's own academic work focuses on evaluating world‑model AI systems. He tests whether strong performance in simulation and benchmarks can translate into reliable robot behavior in the physical world.

"Before a robot goes into a home or a factory, we should be able to find out where it will fail," Zhu said.

He has co‑authored papers accepted by the journal Scientific Data and NeurIPS 2026, and works as a paper reviewer for top‑tier AI conferences.