Planering och styrning av en flotta autonoma mobila robotar med centraliserad rörelsespårning

Abstract

Autonomous mobile robots are increasingly utilized in logistical workflows and material flow management. The process of automating heavy transportations leads to increased efficiency of material handling while simultaneously decreasing manual labour. This study investigates the coordination of a scalable fleet of autonomous mobile robots (AMR) as they transition from logistical patterns to dynamic scenarios. External real time data about the location of a moving target is obtained from a drone based surveillance system whereas the fleet is traced in real time by a motion capture system with high precision localization. The realisation of the system is achieved through a strategy of efficient path planning and collision avoidance. An A* path planning strategy A* is implemeted which utilizes euclidean distance heuristics to minimize the cost between start and goal positions. This determines the optimal trajectory from inital, to designated position. As the fleet scales, the increasing number of agents leads to collisions and conflicts such as deadlocks. This issue is addressed through a multi agent path finding (MAPF) strategy evaluating a local collision resolver A* (LCRA*). LCRA* is based on the model lifelong priority based search (LPBS) and the benefit of using LCRA* is that it significantly improves computational performance as it does not require global time optimization to handle collisions locally. This results in a smooth transition between a logistical pattern mode or an intruder mode in order to collectively encircle the moving target.