Coordinated Search-and-Capture Using Particle Filters

This paper presents a search-and-capture (SAC) problem where multiple autonomous pursuer vehicles are deployed to capture evaders using the particle filter (PF). The PDFs of the evaders states are first represented as discrete sets of support vectors (particles). Using this representation, a coordinated SAC strategy is proposed by firstly defining the observation likelihoods for both detection and non-detection in the PF framework. Coordination is then achieved through the transmission of the likelihood function parameters of each pursuer to other pursuers to form combined observation likelihoods (COLs) followed by the derivation and sharing of weighted expected states (WESs) from the updated PDFs to provide control reference points for the pursuers. The proposed strategy is applied to two scenarios: first to multiple-pursuers single-evader and secondly to multiple-pursuer multiple-evaders. Results show that the proposed strategy allows the pursuers to successfully detect and capture the evaders in both scenarios