Show-Harness: Just a VLM Agent Can Play Robots
Frontier vision-language models (VLMs) can already perceive, reason, and plan remarkably well, yet translating this intelligence into physical action remains challenging.
In this edition of the Singapore Embodied AI Colloquium (SEAIC), we explore Show-Harness, an Embodied Harness that investigates whether the missing piece is not necessarily a stronger robot policy, but a better interface between general-purpose models and robots.
Through a compact semantic action interface, Show-Harness enables frontier VLM agents to directly "play" different robots across tasks and embodiments, while also supporting lightweight fine-tuning of smaller models for faster deployment.
The session will cover the motivation behind Show-Harness, its system design, real-robot results, and broader perspectives on how general-purpose agents may increasingly enter the physical world.
(Technical Level: 100)
More About the Speaker
Yanzhe Chen is a PhD student at the National University of Singapore, advised by Prof. Mike Zheng Shou. His research interests include agents, robotics, and multimedia. He previously interned at TikTok and Kuaishou, and received his undergraduate degree from Wuhan University and his master’s degree from Peking University. He is currently a teaching assistant for Robot Perception at NUS and is involved in projects including Show-Harness and Code2Video, a code-centric paradigm for automated educational video creation.
More About the Series
Singapore Embodied AI Colloquium (SEAIC) is a focused technical research series built around researcher-driven discussions and sharing on frontier topics in embodied AI, robot learning, and physical AI.
Get involved: Learn more about Lorong AI | Speaker Sign-up | WhatsApp Community | LinkedIn | X