Path Inference based on Voronoi Graph for Unmanned Maritime Vehicles

Design/methodology/approach: We propose a novel path inference approach based on Voronoi graph for unmanned maritime vehicles (UMV). We model the two-dimensional space with a flexible construction scheme of Voronoi graph, represent the inter-cell shortest distance with a delicate calculation of cell diameters, and infer the potential qualified navigation paths of UMVs w.r.t. the specified distance constraint based on a diameter path model with an efficient depth-first exploration algorithm. Purpose: Existing path planning approaches are generally focused on finding a single optimal path. However, in many practical scenarios, such as UMV patrolling and search and rescue, it is the comprehensiveness rather than the optimality that matters for the UMV paths. Inspired by the above problem, we propose a path inference method based on Voronoi graph to infer all the possible UMV paths satisfying the distance constraint. Findings: Experimental results indicate that our path inference approach based on Voronoi graph is able to perform more delicately and efficiently than the traditional path planning approach in terms of both space modeling and path exploration. The paths identified by our path inference method were more precise in distance estimation as well as more comprehensive when compared with the grid-based path planning method. In the simulated scenario of UMV search & rescue, we obtained the coverage rates of 63.14% and 100% against those of 14.5% and 81.07% for the grid-based method. Originality/value: Our path inference approach is able to provide valuable insights in various practical UMV applications.