Zhanguang Zhang
Zhanguang Zhang is a researcher at Huawei, affiliated with the company's Noah's Ark Lab. He is known for his work in embodied AI, robotic task and motion planning, and vision-language navigation, with recent research focusing on world action models and their robustness compared to vision-language action models. He has also contributed to research applying graph neural networks to SAT solver selection, work that was accepted to KDD 2024.
“The headline result is that WAMs achieve meaningfully better robustness under visual and language perturbations than standard VLAs... The paper attributes this directly to 'spatiotemporal priors inherited from their world model backbones'.”
Source→AI-extracted from podcast / newsletter / paper summaries. May contain errors.