Spatially Intelligent Patrol Routes for Concealed Emitter Localization by Robot Swarms

TL;DR

The paper addresses locating concealed RF emitters with a robot swarm without assuming emitter power or frequency. It introduces spatially intelligent patrol graphs optimized by differential evolution and triangulation-based localization using least-squares, supporting both omnidirectional and directional antennas. Results show directional sensing dramatically improves success rate (≈$98.75\%$) and localization accuracy (≈$1.0$–$1.3$ m) versus omnidirectional sensing (≈$80.25\%$, ≈$1.7$–$1.9$ m), with triangular patrols delivering the best efficiency per meter. The work highlights that the physical interaction between robots and the environment, through patrol geometry and sensing modality, is crucial for effective electromagnetic surveillance by swarms.

Abstract

This paper introduces a method for designing spatially intelligent robot swarm behaviors to localize concealed radio emitters. We use differential evolution to generate geometric patrol routes that localize unknown signals independently of emitter parameters, a key challenge in electromagnetic surveillance. Patrol shape and antenna type are shown to influence information gain, which in turn determines the effective triangulation coverage. We simulate a four-robot swarm across eight configurations, assigning pre-generated patrol routes based on a specified patrol shape and sensing capability (antenna type: omnidirectional or directional). An emitter is placed within the map for each trial, with randomized position, transmission power and frequency. Results show that omnidirectional localization success rates are driven primarily by source location rather than signal properties, with failures occurring most often when sources are placed in peripheral areas of the map. Directional antennas are able to overcome this limitation due to their higher gain and directivity, with an average detection success rate of 98.75% compared to 80.25% for omnidirectional. Average localization errors range from 1.01-1.30 m for directional sensing and 1.67-1.90 m for omnidirectional sensing; while directional sensing also benefits from shorter patrol edges. These results demonstrate that a swarm's ability to predict electromagnetic phenomena is directly dependent on its physical interaction with the environment. Consequently, spatial intelligence, realized here through optimized patrol routes and antenna selection, is a critical design consideration for effective robotic surveillance.

Figures (5)

• Figure 1: Spatially explicit observation coverage for omnidirectional (column 1) and directional (column 2) sensing geometric patrol routes. (a) (b): Triangular (c) (d): Square (e) (f): Hexagonal. Cross-hatched areas are regions where if an emitter were present an exact source location can be extrapolated. Internal orthogonal lines drawn to their minimum required length. External coverage in reality would extend equally as far out as the minimum distance required internally. Directional lines are drawn to the minimum required length for full coverage, with slope equal to the antenna offsets $\pm \psi$. Triangulation line colors represent coverage per antenna. Rotation coverage at the vertices for directional is not shown.
• Figure 2: Waypoint generation for 4 triangular and 4 hexagonal patrol shapes utilizing each of the sensing strategies. Randomly generated source locations across the 100 trials are included, along with the whether a successful localization prediction was made by the agents.
• Figure 3: Localization error between predicted and true source location, and percentage success rate of forming a successful prediction for both sensing strategies across each patrol shape configuration.
• Figure 4: Localization success rate based on power and frequency of signal sources for each sensing strategy.
• Figure 5: Heatmap of omnidirectional localization failures. Source locations are discretized into a 20$\times$20 grid. Failure counts of zero also include locations where no emitters were placed. Emitter locations are always inside the map.