Cooperatively Scheduling Hundreds of Fetch and Freight Robots in an Autonomous Warehouse

Abstract

Homogeneous robotic sorting systems, such as the famous Kiva system which follows the shelves-to-workers mode, have been successfully used in warehouses. However, these systems generally have shortages in two-folds, i.e., (1) redundant moves of shelves, and (2) unavoidable manual operations. To overcome these shortages, an alternative solution is using heterogeneous robots (fetch and freight robots) cooperatively to accomplish sorting tasks. In this field, the existing works mostly focus on the design of robots, while there is no public literature (to our best knowledge) which studies how to coordinately schedule a large number of fetch and freight robots. To fit this blank, this paper first proposes a cooperative algorithm to schedule hundreds of heterogeneous robots. The algorithm adopts a cloud-edge-terminal architecture, where the cloud is responsible for allocating tasks and robots, the edge computing units monitor the status of each regional area, while the terminals (robots) are able to plan their own paths (guided by the edge computing units) and resolve conflicts by bidding mechanisms. A simulation platform is developed, based on a large amount of simulations (with up to 330 robots) are carried out to analyze the impacts of several key components of the algorithm and confirm the superiority of our algorithm.