Multi agent mobile robot system for active collaborative SLAM application

Abstract

This thesis presents the design, simulation, and real-world implementation of an Active Collaborative Simultaneous Localization and Mapping (AC-SLAM) framework for multi-robot systems, built on ROS2 and validated in NVIDIA Isaac Sim and on physical hardware. The framework combines Active SLAM, which guides autonomous exploration using a prior connectivity graph, with Collaborative SLAM, where multiple robots explore simultaneously and merge their occupancy maps into a unified global map. Pose-graph SLAM, via SLAM Toolbox, provides real-time map construction, while Nav2 handles navigation and obstacle avoidance. A BFS partitioning of the prior connectivity graph assigns exploration regions to each robot. The system is deployed on a custom-built differential-drive robot and a Festo Robotino, coordinated through a control platform.