Development of an Original Theoretical Model for Cooperative Localization and Mapping in GPS-Denied Underground Mining Environments Using Multi-Robot Systems

Underground mining operations at solid mineral mining facilities in Zamfara and Plateau States, North West and North Central Nigeria respectively, represent GPS-denied, geometrically complex, and dynamically hazardous environments where autonomous robotic systems for ore body mapping, equipment inspection, and emergency response are highly desirable but whose deployment is constrained by the fundamental absence of reliable absolute position reference. This dissertation develops an original theoretical model for cooperative localization and simultaneous mapping in GPS-denied underground mining environments using heterogeneous multi-robot systems comprising wheeled ground robots and micro-aerial vehicles. A cooperative SLAM formulation is developed in which robots share local map fragments and relative pose measurements through opportunistic communication links subject to range, bandwidth, and duty-cycle constraints representative of underground wireless sensor deployments. A novel distributed information-theoretic map fusion algorithm is derived that maximizes the information content of the jointly maintained map under communication constraints, with theoretical optimality bounds derived using Fisher information matrix analysis. A relative pose measurement model exploiting ultra-wideband range measurements between robots is formulated and analytically characterized for bias and covariance under the multipath-rich signal propagation conditions of underground tunnels. The theoretical framework is experimentally validated in a simulated underground mining environment constructed at the National Metallurgical Development Centre in Jos, Plateau State. A three-robot heterogeneous team achieves 94.2 percent mapping completeness of the test environment within 18 minutes with absolute localization accuracy of 0.31 metres RMS, surpassing single-robot SLAM performance by 47 percent in mapping completeness. Keywords: cooperative SLAM, multi-robot, underground mining, GPS-denied, localization Nigeria