Design and Development of Decentralized Multi-Robot System for Mapping and Load Handling

Multi-robot systems have seen significant advancements in autonomous navigation, mapping, and task coordination. However, centralized communication architectures introduce latency, limiting real-time collaboration and scalability. This paper presents a decentralized multi-robot framework for mapping and load handling, integrating heterogeneous robots—a 13-DOF humanoid and TurtleBot3 units—to perform collaborative tasks without reliance on a central server. The system employs ROS1-based peer-to-peer communication, graph-based SLAM, and sensor fusion techniques to generate high-accuracy maps while enabling seamless humanoid navigation. The humanoid robot utilizes inverse kinematics for locomotion and YOLOv9-based vision processing for object recognition and weight estimation. Performance evaluations demonstrate a significant reduction in communication latency, improved mapping accuracy, and efficient obstacle manipulation. This work is particularly applicable in disaster response, industrial automation, and collaborative robotics, showcasing a scalable and adaptable approach for heterogeneous robot teams.