Large-scale multi-robot task allocation via dynamic partitioning and distribution

Abstract

The beginning and development of modern intelligent robot makes robot task become more and more complicated. Task distribution and deployment for large-scale robots has become a major problem. Traditional distribution mechanisms are mainly Polling, SMS Push and IP Push, but for the task distribution of robotic devices, these mechanisms cannot fully meet the system's requirements. In this paper, we propose a "Push-Pull" distribution mechanism, and make a comparative experiment with the "Direct-Push" distribution mechanism based on MQTT protocol. Meanwhile, we construct a large-scale distribution and deployment system of robot task, which implements the distribution and deployment of large text tasks promptly and can guarantee the quality of task distribution. The experimental results show that the task package distributed by our system can run successfully on the robot operating system (ROS).

Mechanism Selection for Multi-Robot Task Allocation

“…Heap proposed sequential-singlecluster (ssc) auctions, an extension to ssi that uses an agglomerative clustering algorithm to create task bundles, which are then auctioned as in ssi [9]. Liu & Shell [16] develop a hybrid distributed-centralised approach to MRTA. The task set is first partitioned into subsets that are then solved in parallel using a centralised assignment algorithm [14].…”

A novel versatile architecture for autonomous underwater vehicle’s motion planning and task assignment

“…Iori and Ledesma (2015) modelled AUVs routing problem with a Double Traveling Salesman Problem with Multiple Stacks (DTSPMS) for a single-vehicle pickup-and-delivery problem by minimizing the total routing cost. Other methods also studied on efficient task assignment for single/multiple vehicle moving toward the destination such as graph matching (Kwok et al 2002), Tabu search (Higgins 2005), partitioning (Liu and Shell 2012), simulated annealing (Chiang and Russell 1996), branch and cut (Lysgaard et al 2004), and evolutionary algorithms (Gehring and Homberger 2001). Martinhon et al(2004) proposed stronger K-tree approach for the vehicle routing problem.…”

Shortest path set induced vertex ordering and its application to distributed distance optimal formation path planning and control on graphs

“…Graph theoretic approaches are quite popular here, probably because vehicles and inter-vehicle constraints can be represented naturally with vertices and edges of graphs. The second sub-problem put more emphasis on how to achieve a desired formation via planning [4,7,9,10,15,16,17,19,20,23,28,26,29,30], rather than to stabilize around a given formation. Among these, [15,16,17] appear to be mostly close to our effort in this paper (besides our earlier effort [30]).…”