Belief-updated chance-constrained planning for hemorrhage-risk-aware robotic steering under uncertain vasculature: an in-silico study

Deif M. A., ELHATY I. A. M., Hafez M. A., Khishe M.

International Journal of Intelligent Robotics and Applications, 2026 (ESCI, Scopus)

Abstract

Robotic catheter steering near vascular structures is challenging because vessel geometry is uncertain and may change as new segmentation-like observations become available during motion. This paper introduces an in-silico hemorrhage-risk-aware belief-space planning framework that updates a Bayesian belief over uncertain vessel geometry and couples this belief to chance-constrained safety. Vessel-wall clearance is modeled probabilistically, and the chance constraint is converted into an adaptive deterministic margin governed by the current posterior covariance. A conservative per-vessel trajectory-level risk budget is allocated across time steps using a union-bound construction. The resulting nonconvex planning problem is solved through sequential convexification within a receding-horizon loop. Computational experiments on synthetic benchmark vascular anatomies show that belief-updated planning reduces empirical vessel-wall violations relative to a static-belief chance-constrained baseline, particularly under high uncertainty and slower observation updates. The results support the proposed mechanism as a computational planning approach, not as clinical, phantom, or physical ultrasound validation.