CollOR: Distributed collaborative offloading and routing for tasks with QoS demands in multi-robot system

Multi-access edge computing (MEC) offers prospective opportunities for robots that have various computational tasks to execute in Industry 4.0, smart cities and many other fields. Computational offloading, as the key of this emerging paradigm, can leverage resourceful edge infrastructure nearby to provide computational support for these resource-limited robots. However, existing computational offloading schemes not only rarely consider that robots are geo-distributed, but also ignore the diversified quality of service (QoS) of robotic applications. Thus in this paper, the collaborative computational offloading problem for multi-robot system in MEC is studied, and a distributed scheme is proposed to jointly optimize offloading and routing for computational tasks of robots. First, we introduce the collaborative communication model of robots, construct the multi-criteria utility function related to both the QoS attributes of applications and the energy efficiency of robot, and formulate the joint offloading and routing problem. Afterwards, the QoS preference analysis mechanism is established to characterize the heterogeneous requirements of robotic applications. Next, by designing a novel collaborative robotics framework, we realize the distributed processing and adaptive strategy -making of multi-robots, and integrate the game-theoretic approach to propose a utility-aware collaborative computational offloading algorithm. Finally, simulation results validate that the proposed algorithm has significant performance improvement over the representative algorithms in different evaluation metrics, and can achieve stable and scalable performance in the scenario of the increasing demand for QoS and the growing scale of robots.

Automated summary

This paper studies the computational offloading problem for multi-robot systems in MEC and proposes a distributed scheme to optimize the offloading and routing of computational tasks. By introducing a collaborative communication model and a QoS preference analysis mechanism, and using a game-theoretic approach, a utility-aware collaborative computational offloading algorithm is proposed, which significantly improves performance and achieves stability and scalability under different demands and scales.