Go to IEEE ICRA 2026 Workshop AIMI homepageFrom Distances to Trajectories: Real-Time Signed Distance Function Mapping and Distance-Accelerated Motion Planning for UAVs
Keywords: Signed Distance Function, Path Planning, Quadrotor, UAV, Motion Planning, Aerial Inspection, Marine Infrastructure Inspection, Real-Time Mapping, Onboard Autonomy
TL;DR: A fully onboard UAV system for safe navigation in unknown environments using real-time signed distance function based motion planning.
Abstract: Autonomous aerial inspection of maritime infrastructure, such as large vessels and offshore assets, enables safer and more efficient operation in hazardous environments like confined ballast tanks. These settings involve complex, partially unknown geometry with tight clearances and limited visibility, posing significant challenges for real-time UAV navigation under strict onboard computational constraints. We present a unified framework for efficient onboard signed distance function (SDF) reconstruction and motion planning in unknown environments. Our approach reconstructs a non-truncated Euclidean SDF in real time by combining an explicit octree prior with an implicit neural residual, enabling accurate and differentiable distance estimates from streaming depth data. We further introduce Bubble$^\star$, a search-based planner that expands collision-free regions derived from SDF values. By operating on convex free-space regions directly, the method efficiently generates safe corridors for dynamically feasible trajectory optimization in confined environments. We demonstrate real-time, fully onboard autonomous flight in previously unseen environments, highlighting the effectiveness of tightly coupled SDF-based mapping and planning for UAV inspection of maritime assets.
Email Sharing: We authorize the sharing of all author emails with Program Chairs.
Data Release: We authorize the release of our submission and author names to the public in the event of acceptance.
Submission Number: 10
Loading