Approximate Penetration Volume for Efficient Motion Planning in Deformable Environments

Abstract

Motion planning in environments with deformable objects, such as tissues and flexible materials, poses unique challenges in avoiding damage while accounting for object deformation. This paper presents a novel approach that approximates penetration volume capturing volumetric displacement during robot-object interaction using primitive shapes to efficiently model the environment. By constructing both enclosing and enclosed primitive skeletons, our method balances computational cost with accuracy and integrates this metric into samplingbased motion planning, generating a gradient field graph to guide robot paths while respecting environmental elasticity limits. Our method has potential applications in medical robotics, where safe and efficient interaction with soft tissues is critical, and in other domains requiring robots to navigate complex deformable environments. The contribution of our approach lies in its ability to improve planning efficiency and path quality, achieving significant reductions (8-30%) in planning time compared to exact and depth-based methods.

Department(s)

Computer Science

Document Type

Conference Proceeding

DOI

10.1109/MMAR65820.2025.11150916

Keywords

Collision Detection, Deformable Environments, Path Planning

Publication Date

1-1-2025

Journal Title

International Conference on Methods and Models in Automation and Robotics Mmar

Link to Full Text

Share

COinS