This thesis presents the design, implementation, and validation of a novel leader assisted localization framework for a heterogeneous multi-robot system (MRS) with sensing and communication range constraints. It is assumed that the given heterogeneous MRS has a more powerful robot (or group of robots) with accurate self localization capabilities (leader robots) while the rest of the team (child robots), i.e. less powerful robots, is localized with the assistance of leader robots and inter-robot observation between teammates. This will eventually pose a condition that the child robots should be operated within the sensing and communication range of leader robots. The bounded navigation space therefore may require added algorithms to avoid inter-robot collisions and limit robots’ maneuverability. To address this limitation, first, the thesis introduces a novel distributed graph search and global pose composition algorithm to virtually enhance the leader robots’ sensing and communication range while avoiding possible double counting of common information. This allows child robots to navigate beyond the sensing and communication range of the leader robot, yet receive localization services from the leader robots. A time-delayed measurement update algorithm and a memory optimization approach are then integrated into the proposed localization framework. This eventually improves the robustness of the algorithm against the unknown processing and communication time-delays associated with the inter-robot data exchange network. Finally, a novel hierarchical sensor fusion architecture is introduced so that the proposed localization scheme can be implemented using inter-robot relative range and bearing measurements.

The performance of the proposed localization framework is evaluated through a series of indoor experiments, a publicly available multi-robot localization and mapping data-set and a set of numerical simulations. The results illustrate that the proposed leader-assisted localization framework is capable of establishing accurate and nonoverconfident localization for the child robots even when the child robots operate beyond the sensing and communication boundaries of the leader robots.