Joint Latency-Energy Optimization for Multitask Offloading in 6G Collaborative Robotic Networks

In sixth-generation (6G) networks, collaborative robotic networks enable connected robotic agents to perform tasks cooperatively, yet efficiently offloading tasks is critical to limit resource consumption in large-scale deployments. Compared to conventional mobile edge computing (MEC) approaches, collaborative robotic networks encounter unique challenges, including the high optimization computational complexity for large-scale robotic deployments and the necessity for coordinated, autonomous task assignment among agents. To address this, we formulate a multi-agent multitask offloading problem that jointly considers task assignment, resource management, offloading decisions, and computation frequency assignment, resulting in an NP-hard, nonconvex mixed-integer nonlinear programming (MINLP) problem. We then decompose the problem into resource allocation and task offloading subproblems, which are tackled through proximal policy optimization (PPO) and minimum-cost strategies respectively within a block coordinate descent (BCD) framework. Simulation results indicate that the proposed method achieves stable convergence, reducing latency by 61.95% and energy consumption by 11.82%, demonstrating strong applicability for large-scale collaborative robotic networks in future 6G scenarios, such as the industrial metaverse, autonomous swarm systems, and intelligent transportation.