Adnan KhanAdnanProduct & AI
San Francisco Bay Area
Get in touch

Source note3 min read

Surgical WAM: A World-Action Model for Data-Efficient Surgical Robot Learning

Source

arXiv

arxiv.org/abs/2608.11204v1

Read the source

Source

  • Source name: arXiv
  • Source kind: paper
  • Authors / organization: Wenrui Bao, Tianyun Jiang, Zhiben Chen, Ser-Nam Lim, Peter D. Peng, Yuzhang Shang
  • Venue: arXiv: cs.RO
  • Published: 2026-08-11
  • Collected: 2026-08-12
  • URL: http://arxiv.org/abs/2608.11204v1
  • PDF: https://arxiv.org/pdf/2608.11204v1
  • DOI: UNKNOWN
  • Registry group: academic_apis_and_preprints
  • Evidence role: canonical_academic_index
  • Language: en
  • Country / region: UNKNOWN

Plain-Language Brief

The study introduces Surgical WAM, a model that uses unlabeled surgical videos for pre-training and then fine-tunes on a fixed number of labeled demonstrations. The goal is to improve the success rate in closed-loop surgical manipulation tasks. The results show an improvement in the average success rate by 14.3% across four simulated tasks, with the largest improvements on contact-rich and bimanual tasks.

What the Document Says

  • The study proposes Surgical WAM, a model for surgical robot learning that pre-trains on unlabeled videos and fine-tunes on a fixed number of labeled demonstrations.
  • Surgical WAM is based on Cosmos Policy and predicts future endoscopic observations and executable action chunks.
  • The model acts as a closed-loop, receding-horizon controller at deployment.
  • The study compares the performance on four simulated surgical manipulation tasks with and without video pre-training.
  • The results show an improvement in the average success rate by 14.3% when using video pre-training.

Key Concepts Mentioned

  • surgical robot learning
  • data-efficient learning
  • unlabeled videos
  • closed-loop control
  • simulated surgical manipulation tasks

Claims Made by the Document

  • Action-free video pretraining improves closed-loop surgical manipulation.
  • Surgical WAM is a unified generative model that jointly predicts future endoscopic observations and executable action chunks.

Evidence Used

  • The study presents results showing an improvement in the average success rate by 14.3% when using video pre-training.

Methods / Approach

  • Surgical WAM learns surgical visual dynamics from unlabeled videos and fine-tunes on a fixed number of labeled demonstrations.
  • At deployment, Surgical WAM acts as a closed-loop, receding-horizon controller.

Limitations Stated by the Authors

  • The study only evaluates the performance on simulated surgical manipulation tasks.

Limitations Noted by the Agent

Agent interpretation — not from the source.

  • The study does not provide information about the generalizability of the results to real-world surgical scenarios.
  • The study does not compare Surgical WAM with other similar methods.

Open Questions

  • How well does Surgical WAM perform in real-world surgical scenarios?
  • How does Surgical WAM compare to other methods for surgical robot learning?

Notes for Future Q&A

May include interpretation. Labeled where it does.

  • The study presents a method for improving closed-loop surgical manipulation by using unlabeled videos for pre-training and fine-tuning on a fixed number of labeled demonstrations.
  • The results show an improvement in the average success rate by 14.3% across four simulated tasks.

Raw Abstract / Excerpt

Learning reliable surgical manipulation policies is bottlenecked by the scarcity of action-labeled demonstrations: teleoperated surgical robot (e.g., dVRK) trajectories with synchronized kinematics are costly to collect, while surgical tasks demand precise contact handling, long-horizon reasoning, and bimanual coordination. Endoscopic video is comparatively inexpensive and abundant relative to synchronized video--kinematics trajectories, and a natural way to exploit it is to learn world models of surgical scenes. However, existing surgical world models use video primarily for simulation or policy evaluation, and rarely translate the learned dynamics into closed-loop control. This gap raises our central question: under a fixed budget of action-labeled demonstrations, does action-free video pretraining improve closed-loop surgical manipulation? To answer it, we introduce the Surgical World-Action Model (Surgical WAM), a unified generative model built on Cosmos Policy that jointly predicts future endoscopic observations and executable surgical robot action chunks. Surgical WAM first learns surgical visual dynamics from action-free video and is then fine-tuned on the fixed action-labeled budget; at deployment, it acts as a closed-loop, receding-horizon controller that executes a short prefix of each predicted action chunk and replans from the resulting observation. On a suite of four simulated surgical manipulation tasks, video pretraining improves the average success rate from 63.5% to 77.8%, including an absolute gain of 20 percentage points on PegTransfer, with the largest improvements on contact-rich and bimanual tasks. These results demonstrate that action-free video provides transferable visual dynamics priors for learning surgical robot control with limited action supervision, positioning data-efficient video pretraining as a practical path toward scaling up surgical robot learning.