Max Robotics — Global Service Robot Directory + US Certification

About

Jiayuan Mao is a leading researcher at the intersection of robotics, artificial intelligence, and cognitive science, with a focus on enabling robots to achieve compositional generalization and robust physical reasoning. Her major contributions center on developing neuro-symbolic frameworks that bridge high-level semantic understanding with low-level manipulation skills. In her highly cited work on *Programmatically Grounded, Compositially Generalizable Robotic Manipulation*, she pioneered methods that integrate large-scale vision-language models with programmatic reasoning, allowing robots to decompose complex tasks into reusable skills. Her research on *KALM* introduces keypoint-based abstraction using large models for imitation learning, achieving generalization to novel object configurations—a critical challenge in robotics. Further, her work on *What's Left?* advances concept grounding by enhancing foundation models with logic-based reasoning, moving beyond 2D images to richer physical domains. She also created the *HandMeThat* benchmark, which evaluates human-robot communication in ambiguous social and physical environments. With over 15 citations across her most prominent papers, Mao’s work is shaping the future of generalizable, semantically-aware robotic systems that can understand and act in the real world.