Harris Chan

About

Harris Chan is a leading researcher at the intersection of robotics, artificial intelligence, and natural language processing, with a focus on enabling robots to understand and execute complex, semantically rich instructions. His most influential work, "Inner Monologue: Embodied Reasoning through Planning with Language Models" (2022, 206 citations), pioneered the use of Large Language Models (LLMs) for real-world robotic planning, demonstrating how LLMs can reason about environmental feedback to guide manipulation tasks—a breakthrough that bridges high-level language understanding with low-level physical control. Chan also advanced robotic skill acquisition through instruction augmentation with vision-language models, notably in his 2023 paper (29 citations), where he developed methods to propagate semantic knowledge from models like CLIP across large, unlabeled robot datasets, drastically reducing the need for expensive human annotation. This work enables robots to generalize from limited demonstrations to unseen instructions, a critical step toward deployable, language-guided automation. As a core contributor to the Google Robotics team, Chan’s research has been instrumental in shaping how embodied agents reason, plan, and learn, earning him recognition as a rising star in embodied AI.

Research Focus

Key Achievements

Top Papers

1. 1
   Inner Monologue: Embodied Reasoning through Planning with Language Models

   206 citations · 2022
2. 2
   Robotic Skill Acquisition via Instruction Augmentation with Vision-Language Models

   29 citations · 2023
3. 3
   Robotic Skill Acquisition via Instruction Augmentation with Vision-Language Models

   2 citations · 2022

Key Collaborators

Contact & Links