Constraint-Aware Occupancy Map Abstraction for Lightweight Path Planning in Hazardous Environments
Radiation-aware navigation is critical for enabling mobile robots to operate safely and reliably in hazardous environments. By accounting for radiation exposure during planning, the cumulative dose experienced by a robot can be reduced, improving system longevity and mission robustness. This paper presents a cost-aware navigation framework that abstracts an occupancy map into a sparse graph, uses radiation intensity to bias where graph nodes are placed, and weights each edge by the line integral of intensity along it, a relative dose proxy, so that each query is an A* search on the graph. The method is evaluated in simulation over 750 trials in five environments against distance-only grid A* and a full-grid planner that minimises the same objective. In trials where the shortest route crosses elevated intensity, the proposed method reduces mean cumulative radiation cost by 28.9% for a 6.7% increase in path length, against 31.5% for the full-grid planner, while each query runs 15–24 times faster; building the graph pays for itself after 4–10 queries on the same map. On ray-traced fields with wall shielding, the reductions are 62.3% and 65.2%. An ablation shows that the reduction comes from the cost-aware edge weights: radiation-aware node placement moves nodes but does not reduce exposure further, and letting node separation grow with intensity shrinks the graph at the cost of higher exposure.
Authors
- Paolo Paoletti (ORCID: https://orcid.org/0000-0001-6131-0377)
- Mario Gianni (ORCID: https://orcid.org/0000-0001-5410-2377)
- Ipek Caliskanelli (ORCID: https://orcid.org/0000-0002-2641-3570)
- Andrew A West (ORCID: https://orcid.org/0000-0003-4553-8640)
- David Batty (ORCID: https://orcid.org/0009-0003-5375-421X)
Institutions
- University of Liverpool (GB)
- University of Manchester (GB)
- United Kingdom Atomic Energy Authority (GB)
Publication Details
- Journal
- Applied Sciences
- Published
- 2026-11-18
- DOI
- https://doi.org/10.3390/app162211428
- Primary Topic
- Robotic Path Planning Algorithms
- Type
- article
- Field-Weighted Citation Impact
- 0.00
Funders
- Engineering and Physical Sciences Research Council