Vision-guided Unified Task and Motion Planning for Multiple Robots in Cluttered Environments

Efficient coordination of multiple mobile and fixed base robot arms to perform independent or cooperative tasks in a shared workspace while avoiding collisions is a challenging research problem that has been studied across various robotic fields. Although robotic systems are commonly used in industrial applications such as automotive production, they typically operate independently with limited collaboration between two or more units. In this study, we propose a unified methodology for motion and task planning for multiple robot arm systems with high degrees of freedom that are simultaneously performing different tasks in a shared workspace. Our methodology introduces an integration module between the high-level task planning module and low-level motion planning module to achieve reliable and collision-free execution of tasks. We also introduce the concept of a dynamic shared space graph (D-SSG) to assess the cooperation between arm pairs that share certain parts of the workspace, which is essential for selecting arm sequences and scheduling each arm to perform a task or sub-task.

I. Introduction

Currently, a considerable number of applications involving robots employ platforms that are equipped with dual-arm and/or multi-arm manipulators (DaM) that operate within a shared workspace to improve their manipulation efficiency compared to single-arm platforms [1], [2]. To operate effectively, these platforms must be capable of planning arm motions that allow objects to be transferred between various locations in the workspace, especially since these platforms can consist of n-arm robotic systems . Moreover, the challenge of task scheduling arises when several objects are in motion simultaneously, as outlined in [3]–[5].