// readme

Awesome Agentic Robotics

A reading list on how robots plan, act and improve through interaction. arXiv dates are first submissions. Project releases, workshop appearances and demos are dated separately. 🌟 marks verified open-source code.

GitHub star rankings by method and task

Each dated entry counts once by its primary link, including workshop papers and project releases. A paper added during a backfill appears in its original publication month. Term percentages use all titles in each period as the denominator.

Browse by method and task

S1 is the policy or skill that turns a goal into robot motion. Each paper appears once under its main demonstrated agent–S1 relationship. Task labels describe the downstream robot goal.

  • Agent Uses Robot β€” The agent issues robot commands or generates the control program that executes them.
  • Agent Controls Policies β€” The agent selects, sequences or supervises existing skills or policies. The executor translates each subgoal into motion.
  • [Agent Trains…