A computer vision-based framework for the synthesis and analysis of beamforming behavior in swarming intelligent systems

This work proposes a computer vision-based framework for analyzing and synthesizing the collaborative radiation behavior from swarming clusters of intelligent systems. Radiating sensor nodes with inertial measurement units and optical identification features represent the networked cluster of radiators. These create a set of object-distinguishable nodes on a reticulating platform capable of arbitrary spatial distributions. Node discovery and tracking algorithms based on open-source computer vision libraries use image and depth-of-field information from multi-spectral cameras. These locate nodes and their volumetric distribution within the cluster. An automated system then derives the weighted phases for collaborative beamforming from the resulting nodal distribution. Measured and simulated radiation patterns are gathered and compared to demonstrate the capability and accuracy of the proposed framework and to explore its usability in swarm applications.