PolicyIK provides high-precision inverse kinematics via reinforcement learning
PolicyIK uses goal-conditioned RL to achieve millimeter-precision IK for robot arms, outperforming classical controllers and enabling 95.7% success in simulated pick-and-place with π0.5.