Files
Abstract
This dissertation advances the state of the art in persistent multi-robot systems by addressing fundamental challenges that arise in real-world deployments.Existing approaches frequently rely on idealized assumptions—such as unlimited energy resources, continuous GPS availability, and access to expensive sensors—that rarely hold in practice. To overcome these limitations, this work introduces novel algorithms and frameworks that enable energy-efficient, GPS independent coordination for robust path planning and coverage in dynamic and resource-constrained environments. We investigate the impact of different information function formulations on energy consumption in Informative Path Planning (IPP), with a focus on maintaining model fidelity and prediction confidence. Using Gaussian Process Regression, we systematically evaluate how these functions influence energy efficiency while achieving core IPP objectives, such as accurate environmental modeling and reliable predictions. This analysis provides valuable insights into how sampling parameters can be tuned to balance energy expenditure with informative data collection. Building on these insights, we introduce an energy-aware IPP algorithm that dynamically assigns energy-efficient trajectories based on real-time energy levels, effectively reducing overall energy consumption and enhancing system longevity. In addition, we propose a novel control law for multi-robot coverage that accounts for both heterogeneous and time-varying energy depletion rates, as well as the current energy levels of individual robots. To address these dynamics, we develop a weight-adaptation mechanism that reallocates regions of interest in real time based on each robot’s energy availability, thereby extending the system’s operational duration. We further present a framework for coverage control in GPS-denied environments by leveraging low-cost onboard sensors and a shared anchor point for localization. The anchor point is established using relative sensing among robots and serves as a common spatial reference for coordination. To further relax the assumption of a known anchor point location and environmental distribution, we integrate IPP with coverage control in GPS-denied environments. To support operation under these constraints, we design a three-phase anchor-oriented controller. In the first phase, the robots localize themselves by identifying the anchor point. In the second phase, the controller guides the robots through informative path planning to gather environmental data. Finally, in the third phase, the robots transition to full coverage, ensuring spatial completeness of the area of interest. This phased approach enables robust and efficient operation without relying on external localization systems. Together, these contributions provide comprehensive solutions to critical challenges in enabling persistent multi-robot operation in resource-constrained,GPS-denied environments. The proposed strategies significantly improve energy efficiency, extend system longevity, and adapt path planning and coverage algorithms to dynamic real-world conditions. The effectiveness of our approaches is validated through both theoretical analysis and real-world experiments, laying a strong foundation for future advancements in multi-robot systems.