Herd Monitoring with Autonomous Drones: a Decentralized k-Coverage-inspired Approach

This thesis explores the innovative intersection of autonomous drones with wildlife monitoring, specifically focusing on herd monitoring through a decentralized k-coverage approach, aiming for a significant improvement in the quality and scope of ecological data collection. The work includes an in-depth analysis of drone-captured footage telemetry, enabling the reconstruction of animal dynamics and drone engagement. Reproducing this complex situation involved enhancing the existing herd behavior model and implementing it using the Alchemist simulator environment to model multiple sophisticated scenarios and conditions. The significant contribution of this research lies in adapting aggregate algorithms for the Online Multi-Object k-Coverage (OMOkC) problem to the drone-based herd tracking scenario. The focus is on adopting a hierarchical clustering technique that optimizes target definition and assignment in a distributed network of drones while observing a highly dynamic environment. Through comprehensive simulations, the thesis assesses the performance of these algorithms, exploring the impact of clustering improvements and the effect of an adaptive hierarchical clustering algorithm. The implications of this work extend beyond the immediate application to herd monitoring, suggesting a paradigm shift in how we approach biodiversity conservation and ecosystem management in the era of autonomous systems.