Real-Time Coverage Path Planning for Bone-Aware Robotic Ultrasound Scanning
Yu Zhang, Minqi He, Dong Han, Shengwei Guo, Shaohua Zhang, Lin Yan, Hongdi Liu, Dahu Zhu
Abstract
Robotic ultrasound systems offer significant potential for automated organ coverage scanning, yet acoustic shadows from bone structures can cause incomplete scans and missed diagnoses. This work presents a real-time coverage path planning method for bone-aware robotic ultrasound scanning. A voxel based perception map is built by segmenting organ regions with a U-Net network and back-projecting acoustic shadows to estimate bone locations. Ultrasound image data are then fused with real time probe poses and mapped into 3D space. The proposed method runs online over a continuously updated perception map, selecting the next path segment in real time by maximizing motion efficiency and information gain. Phantom experiments demonstrate that the proposed method effectively avoids bone obstruction, while human experiments confirm its clinical feasibility. On average across the kidney and spleen experiments, the proposed system achieved 87.69% of ultrasound physician performance while requiring only 1.22 times the scanning time, demonstrating strong potential for clinical application.
BibTeX
@inproceedings{ral2026_realtimecoverage,
title = {Real-Time Coverage Path Planning for Bone-Aware Robotic Ultrasound Scanning},
author = {Yu Zhang and Minqi He and Dong Han and Shengwei Guo and Shaohua Zhang and Lin Yan and Hongdi Liu and Dahu Zhu},
booktitle = {RA-L 2026},
year = {2026}
}