RL's precision ceiling isn't RL, it's training only in sim.
Replying to @xiaolonw and @chris_j_paxton
Any online optimization (IK, ID, MPC) for low-level control should be abandoned. A RL policy can fundamentally track more precisely, more robustly, and adaptively to unknown disturbances and uncharacterized noise, especially for low-cost hardware, even on a manipulator.
