多智能体博弈的纺织车间搬运机器人任务分配

Task allocation of handling robot in textile workshop based on multi-agent game theory

针对当前纺织品智能化生产加工过程中搬运机器人所面临的大规模复杂动态任务分配问题,提出了一种基于智能体博弈理论的分布式自主决策框架。根据纺织品实际生产过程任务环境建立任务模型,在考虑到任务的距离和时间优先级等因素下建立目标函数,以智能体的目标效用函数最优为策略选择依据,引入博弈论中的Nash均衡理论对问题求解,并对该决策框架进行了实验验证。实验结果表明:在该决策框架下任务的分配相比同类分布式任务分配算法能够得到全局最优解,具有较强的可扩展性、良好的鲁棒性、收敛性能,同时对动态任务分配同样具有良好结果表现。

A distributed autonomous decision-making framework was proposed based on multi-agent game theory.The framework is used to solve the problems of handling robot in the process of intelligent textile production and processing,which are large-scale and complex dynamic task allocation problems.To start with,the task model was established according to the actual task environment of textile production.Taking into account of the task distance and time priority,the target utility function of the agent was then used as the policy selection basis,and the equilibrium theory of the game was introduced to solve the problem.Eventually,the decision framework was verified by experiments.The experimental results show that the global optimal solution of the task allocation in this decision framework can be better achieved in comparison to the similar distributed task allocation algorithms.In summary,the proposed algorithm has high scalability,good robustness,and convergence performance.Furthermore,the proposed algorithm has excellent performance for dynamic task allocation.