Autonomous Topological Optimisation for Multi-robot Systems in Logistics

Abstract

Multi-robot systems (MRS) are currently being introduced in many in-field logistics operations in large environments such as warehouses and commercial soft-fruit production. Collision avoidance is a critical problem in MRS as it may introduce deadlocks during the motion planning. In this work, a discretised topological map representation is used for low-cost route planning of individual robots as well as to easily switch the navigation actions depending on the constraints in the environment. However, this topological map could also have bottlenecks which leads to deadlocks and low transportation efficiency when used for an MRS. In this paper, we propose a resource container based Request-Release-Interrupt (RRI) algorithm that constrains each topological node with a capacity of one entity and therefore helps to avoid collisions and detect deadlocks. Furthermore, we integrate a Genetic Algorithm (GA) with Discrete Event Simulation (DES) for optimising the topological map to reduce deadlocks and improve transportation efficiency in logistics tasks. Performance analysis of the proposed algorithms are conducted after running a set of simulations with multiple robots and different maps. The results validate the effectiveness of our algorithms.