cs.CVNov 17, 2026

Distilling World Action Models for Efficient Bimanual Manipulation via Anchored Path Matching

Authors: Jialong Liu, Junjie He, Yandu Chen, Bin Zhou, Heng Su, Peijun Tang, Jianan Wang, Changhao Chen

Organizations: Astribot · The Hong Kong University of Science and Technology (Guangzhou) · Harbin Institute of Technology · The Hong Kong University of Science and Technology · 

Abstract

World Action Models (WAMs) jointly generate future video and robot actions through iterative denoising. Few-step distillation reduces the number of model evaluations, but its coarse updates can move the student away from the teacher's trajectory. We propose Anchored Path Matching (APM), which supervises endpoint and short-transition predictions from states reached by the student. From each state, we continue the teacher to zero noise and use its endpoint and first intermediate state as targets. Modality-routed adapters learn to match both targets from the same starting state. When the two gradients conflict, anchor-priority updates preserve the endpoint gradient and remove the opposing component of the local gradient within each modality. One checkpoint supports 1-, 2-, and 4-step joint generation. The sampling budget can be chosen at deployment without retraining or teacher queries. On RoboTwin 2.0, we train only on Clean demonstrations and evaluate directly on the domain-randomized Random setting. At one step, APM improves closed-loop success over continued base distillation at matched inference budgets by 2.0 percentage points on Clean and 1.4 on Random. Video peak signal-to-noise ratio (PSNR) rises from 30.169 to 32.317 dB, with gains in structural similarity (SSIM) and learned perceptual image patch similarity (LPIPS). On two real-robot tasks, APM improves success over the one-step teacher and reduces control-cycle latency by 21.3×21.3\times relative to the eight-step teacher.Project page: https://jialonglong.github.io/Anchored-Path-Matching/.

Figures & tables

Explore similar work

CardsList
  1. AnyStep-WAM: Budget-Aligned Distillation and Adaptive Inference for World Action Models

    Sep 27, 2026Rui Wang, Xiangyu Wang, Donglin Yang +4Adaptive InferenceDiffusion Model Distillation

  2. Flash-WAM: Modality-Aware Distillation for World Action Models

    Jun 3, 2026Arman Akbari, Ci Zhang, Arash Akbari +6Video Diffusion ModelsAction-Conditioned World Models

  3. WAM-OPD: Sharpening World Action Models via On-Policy Distillation

    Sep 28, 2026Panjun Liu, Xiaohan Lei, Shiqi Zhang +8Robot Policy LearningRobot Policy Adaptation