Zidan Wang

Papers

Total Citations

H-Index

About

Zidan Wang is an emerging robotics and artificial intelligence researcher whose work sits at the intersection of robot learning, task planning, and large language models. His research focuses on two closely related challenges in modern robotics: enabling robots to learn feasible, reliable behaviors from demonstrations, and empowering them to plan and act intelligently in novel environments without task-specific training. In his work on "Cold Diffusion on the Replay Buffer," Wang investigates a critical but often overlooked limitation of Learning from Demonstrations (LfD) — the feasibility of the plans these systems generate. By grounding learned behaviors in known good states, his approach pushes toward safer and more practically deployable robot systems. Complementing this, his "Wonderful Team" framework leverages multi-agent Vision Large Language Models (VLLMs) to achieve zero-shot high-level robotic planning, allowing robots to interpret real-world visual contexts and execute complex tasks in entirely new environments without prior exposure. Though early in his research career — with both papers accumulating citations since 2023 and 2024 respectively — Wang's contributions address foundational problems in robot autonomy and generalization, areas of rapidly growing importance as robotics moves toward real-world deployment.