Exposure Conscious Path Planning for Equal Exposure Corridors

TL;DR

This work tackles minimizing visual exposure to the environment during robot navigation, revealing that exposure minimization is non-Markovian when the state is limited to the current position. It introduces a two-graph region framework (traversability and exposure) and analyzes multiple A*-based planners (Exposure Score, Binary, Saturation) to quantify the trade-offs between optimality and computation time. A key contribution is the concept of equal-exposure corridors, which characterize all paths that do not increase exposure and enable efficient local and team-based planning. Experimental results in Boxes and Hills demonstrate the practical performance of the approaches, with Binary and Tau=1 Saturation offering tight optimality gaps and Exposure Score achieving fast precomputation, highlighting the potential for exposure-aware navigation in dynamic, real-world settings.

Abstract

While maximizing line-of-sight coverage of specific regions or agents in the environment is a well-explored path planning objective, the converse problem of minimizing exposure to the entire environment during navigation is especially interesting in the context of minimizing detection risk. This work demonstrates that minimizing line-of-sight exposure to the environment is non-Markovian, which cannot be efficiently solved optimally with traditional path planning. The optimality gap of the graph-search algorithm A* and the trade-offs in optimality vs. computation time of several approximating heuristics is explored. Finally, the concept of equal-exposure corridors, which afford polynomial time determination of all paths that do not increase exposure, is presented.

Figures (8)

• Figure 1: When delivering a package the robot should take the solid black line route over the dashed red one, so as to minimize the likelihood of being seen by a malicious agent.
• Figure 2: Example graph formulation demonstrating an exposure graph (a) and traversability constraint graph (b) for a problem where the exposure function is visual line-of-sight.
• Figure 3: An example line-of-sight exposure scenario where the optimal path $F \rightarrow H$ does not contain the optimal sub-path $F \rightarrow E$ when the objective is to minimize the number of positions line-of-sight is shared with.
• Figure 4: Overview of the environments tested. Boxes (a) provides large areas of zero-gradient and gradient spikes. Hills (b) supplies a well-defined gradient but several local exposure minima.
• Figure 5: Ratio of computation times for the A* Exposure Search, A* Binary, A* Saturation, and Best-First-Search implementations compared to an exposure-agnostic shortest-path A* implementation. The A* Saturation time is the average of the tested $\tau$ values.