SEAR: Self-Evolving Agentic Robotics
Frontier language-model agents can now improve robots on their own: they write control code, test it, and keep what works as skills and memories for the next task. We introduce SEAR (Self-Evolving Agentic Robotics), a framework and benchmark built around two questions. Is the agent really learning, or just trying more...
Xin-Yi Wang, Bang-Zheng Li, Han-Yu Wang et al.
· 0 citations