MimicIK: Real-Time Generative Inverse Kinematics from Teleoperation with FK Consistency

2026-06-13

AI Summary

This work addresses the challenges in real-time robotic inverse kinematics (IK)—namely, branch switching, instability near singular configurations, and the difficulty of simultaneously achieving accuracy, smoothness, and computational efficiency with learning-based approaches. The authors propose a generative IK framework trained on teleoperation data, which leverages conditional flow matching to learn motion priors in joint space and employs a two-step iterative minimal iteration policy (MIP) to predict continuous joint increments. A differentiable forward kinematics consistency loss is introduced to enforce physical plausibility. This approach is the first generative IK method to jointly achieve high accuracy, trajectory smoothness, robustness to singularities, and real-time performance. Evaluated on a real 6-DoF robotic dataset, it attains a mean position error of 4.65 mm, a 92.01% success rate within 10 mm, a trajectory spike rate of 7.99%, and an inference latency of only 6.74 ms, enabling stable real-time control at 20 Hz.

Abstract

Inverse kinematics (IK) remains a critical bottleneck for real-time robot manipulation. Classical numerical solvers achieve high geometric precision but often suffer from discontinuous branch switching and unstable behavior near kinematic singularities during closed-loop deployment. Meanwhile, learned IK approaches frequently struggle to balance spatial accuracy, motion smoothness, and real-time efficiency, particularly when trained on noisy human teleoperation data. We present \textbf{MimicIK}, a real-time generative inverse kinematics framework that learns smooth and robust joint-space motion priors from teleoperation demonstrations through conditional flow matching. Given the current joint configuration and a target end-effector pose, MimicIK predicts continuous delta-joint commands using an efficient two-step iterative refinement process based on a Minimal Iterative Policy (MIP) backbone. To enforce physical consistency, we further introduce an FK consistency loss, a differentiable forward-kinematics regularization that penalizes task-space deviations from the target pose during training. We evaluate MimicIK on a real-world 6-DOF robot dataset containing 8,848 teleoperation demonstrations. MimicIK achieves a mean position error of 4.65 mm, a 10 mm success rate of 92.01\%, and a trajectory spike rate of only 7.99\%. Compared with a UNet diffusion baseline, our method improves both spatial accuracy and motion smoothness while reducing inference latency from 21.66 ms to 6.74 ms. Furthermore, unlike deterministic MLP baselines that catastrophically diverge under out-of-distribution deployment, MimicIK remains stable near singular configurations and enables robust 20 Hz real-time control on deployment hardware.

Problem

• Inverse Kinematics
• Real-Time Control
• Teleoperation
• Motion Smoothness
• Kinematic Singularities

Innovation

• Generative Inverse Kinematics
• Conditional Flow Matching
• FK Consistency Loss
• Real-Time Robot Control
• Teleoperation Learning