Inspection of confined spaces, such as aircraft wings and ship compartments, is essential for maintaining safety and reliability in critical infrastructure. However, these environments are difficult and hazardous for human workers, and existing robotic methods often struggle to operate efficiently in tight, complex spaces filled with structural constraints. Current exploration strategies tend to generate motion patterns that work well in open areas but are poorly suited for confined environments with limited maneuverability and restricted communication. The research funded by this award seeks to develop new coordination strategies that enable teams of robots to explore and inspect constrained spaces more effectively and safely. By improving inspection efficiency and reducing the need for human entry into confined structures, this research supports US competitiveness in the aerospace and ship manufacturing sectors, enhances workers safety and lowers maintenance cost of industrial and civil infrastructure. The project will also advance workforce development by providing specialized research training and education to undergraduate and graduate students in decentralized control of networked systems and multi-robot systems. Building on prior advances in robotic exploration, the goal of this research is to develop a new mathematical framework for coordinating multiple robots in topologically constrained environments. The central technical contribution is the design of time-discounted ergodic controllers on spatially discretized region graphs, enabling efficient exploration in confined spaces where continuous-space methods are difficult to apply. The research addresses the following challenges: (i) developing decentralized methods for estimating and updating time-varying target distributions over graphs using shareable information metrics, without requiring transmission of all the collected data; and (ii) accounting for agent heterogeneity by incorporating differences