Probabilistic Task Assignment for Specialized Multi-Agent Robotic Systems

摘要

This paper introduces a probabilistic approach for assigning specialized individual agents among a robotic swarm to corresponding constrained tasks. Based on the assumption that each individual agent possesses specialized capabilities, the proposed approach evaluates probabilistic fitting of the available robot individuals based on the requirements imposed by the current task, which takes the form of a recognized target object in a specific environment. A formal matching scheme is developed to evaluate a task-agent fitting score among all available agents. It assigns the most qualified and available specialized robotic agent as the best responder to perform the recognized task. A simulation study is presented to validate the efficiency and robustness of the proposed approach.