Anatomical Landmark-Guided Deep Reinforcement Learning for Autonomous Gastric Navigation

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Wu, Haoxuan, Yuan, Sishen, Gao, Haitao, Li, Zhen, Zuo, Xiuli, Ren, Hongliang
Format: Preprint
Published: 2026
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866914545831772160
author Wu, Haoxuan
Yuan, Sishen
Gao, Haitao
Li, Zhen
Zuo, Xiuli
Ren, Hongliang
author_facet Wu, Haoxuan
Yuan, Sishen
Gao, Haitao
Li, Zhen
Zuo, Xiuli
Ren, Hongliang
contents Wireless capsule endoscopy (WCE) enables painless visualization of the gastrointestinal tract, but its diagnostic potential is limited by incomplete mucosal coverage and poor transferability of existing navigation methods across patient anatomies. We propose a transferable, anatomical landmarkguided deep reinforcement learning (AL-DRL) framework for autonomous gastric navigation. Leveraging a lightweight edgecontour-depth fusion module, our policy operates on stable, lowdimensional landmark coordinates rather than high-dimensional video streams, effectively bridging the sim-to-real gap. In simulations across eight patient-derived models, the method achieves over 97% coverage within 50 seconds, significantly outperforming vanilla PPO, SAC, and DQN agents. A two-stage sim-to-real pipeline with an adaptive dynamic programming controller actively mitigates physical disturbances. Ex-vivo experiments demonstrate a mean coverage of 87% and a 53% reduction in procedure time compared with expert manual control.
format Preprint
id arxiv_https___arxiv_org_abs_2605_08269
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Anatomical Landmark-Guided Deep Reinforcement Learning for Autonomous Gastric Navigation
Wu, Haoxuan
Yuan, Sishen
Gao, Haitao
Li, Zhen
Zuo, Xiuli
Ren, Hongliang
Robotics
Systems and Control
Wireless capsule endoscopy (WCE) enables painless visualization of the gastrointestinal tract, but its diagnostic potential is limited by incomplete mucosal coverage and poor transferability of existing navigation methods across patient anatomies. We propose a transferable, anatomical landmarkguided deep reinforcement learning (AL-DRL) framework for autonomous gastric navigation. Leveraging a lightweight edgecontour-depth fusion module, our policy operates on stable, lowdimensional landmark coordinates rather than high-dimensional video streams, effectively bridging the sim-to-real gap. In simulations across eight patient-derived models, the method achieves over 97% coverage within 50 seconds, significantly outperforming vanilla PPO, SAC, and DQN agents. A two-stage sim-to-real pipeline with an adaptive dynamic programming controller actively mitigates physical disturbances. Ex-vivo experiments demonstrate a mean coverage of 87% and a 53% reduction in procedure time compared with expert manual control.
title Anatomical Landmark-Guided Deep Reinforcement Learning for Autonomous Gastric Navigation
topic Robotics
Systems and Control
url https://arxiv.org/abs/2605.08269