A robust, scalable multi-robot control and coordination framework achieving high throughput for parcel sorting centers

Ch'ng, Chee Henn (2025) A robust, scalable multi-robot control and coordination framework achieving high throughput for parcel sorting centers. Master dissertation/thesis, UTAR.

PDF Download (1509Kb)

Abstract

This dissertation presents a robust, scalable multi-robot control and coordination framework for high-throughput parcel sorting centres. The research addresses the complexities of multi-robot systems in indoor settings, where homogeneous robots operate on a shared grid, focusing on enhancing pathfinding, coordination, and throughput. It begins by highlighting the need for advanced multi-robot systems in automation, identifying the limitations of existing multi-agent pathfinding (MAPF) solutions, such as static assumptions, computational overhead, and inflexibility in dynamic environments and human-robot interactions. To overcome these challenges, the research introduces a novel framework that separates the problem into path planning and resource allocation, simplifying the MAPF problem and optimizing robot coordination. The algorithmic model supports efficient path generation, dynamic resource allocation, and conflict resolution, with a focus on plan satisfiability and operational feasibility. The implementation strategy includes dynamic plan updates, operator selection, and traffic control mechanisms to enhance efficiency. Simulation experiments demonstrate the framework’s adaptability and effectiveness in managing multi-robot dynamics. Unlike traditional approaches, it employs a dynamic iterative allocation method that handles uncertainties and optimizes plans in real time, significantly reducing computational demands. Due to its assumption of a non perfect environment, unlike traditional algorithms, the framework is easier to implement. By balancing reactive and deliberative strategies, this framework bridges the gap between theoretical models and practical applications in robotics and automation. The research demonstrates significant improvements in system throughput, with pattern-matching allocators outperforming native allocators by up to 23.52%. The framework’s traffic control mechanism achieves a 58.24% higher throughput compared to systems without such controls. The research’s impact extends beyond parcel sorting centres, offering potential applications in warehouse management, manufacturing, and smart city logistics. Furthermore, its adaptability to dynamic environments and human-robot interactions paves the way for more seamless integration of robotic systems in human-centric workspaces, potentially revolutionizing collaborative robotics.

Actions (login required)