Receding task allocation method for modular robots during on-orbit assembly

Abstract

On-orbit assembly of large infrastructures plays a pivotal role in future space missions. Taking the process of extraction of one assembling object, its transportation and placement by one or several modular robots as a single-robot (SR) or multi-robot (MR) assembling tasks, this paper investigates the problem of designing a task allocation method, so as to realize the adaptive coordination among modular robots. A receding task allocation method is developed to solve this problem in a dynamic and decentralized manner. By proposing the receding-task strategy, the allocation over all tasks is formulated as the allocations at each triggering time over receding tasks to reduce the online computational complexity, where the receding optimization and triggering mechanism based on real-time information can improve the adaptability to a dynamic environment. In order to satisfy the precedence constraints from assembly sequence and the special task requirements about MR tasks, a decentralized algorithm which is an extension of the consensus based bundle algorithm is designed for the allocation of receding tasks, which can reduce the online computational burden of individuals. Numerical simulations are carried out to validate the performance of the proposed method.