Co-building a New Paradigm of Physical Intelligence: Situation Awareness-based Jiuwen Symbiosis

Talk overview

This talk introduces Jiuwen Symbiosis, a Physical AI Agent framework that helps agents move from digital assistants to executors in the physical world, using natural language instructions and situational awareness to perform real-world tasks. It examines the challenges facing end-to-end embodied models such as VLA: limited generalization across robot embodiments, environments, and tasks; insufficient capabilities for long-horizon composite tasks; and low success rates and poor stability. Jiuwen Symbiosis proposes an end-to-end Agent system built on decoupled layers of physical perception, dynamic orchestration, physical execution, and observation feedback. Through Physical Harness, Adapter, and an SDK for heterogeneous robot embodiments, it aims to deliver stability and generalization for complex, long-horizon physical tasks.