Cooperating agents approach to task execution planning

Purpose: The aim of the paper is to investigate the behaviour of the cooperating agents having to agree task execution. Design/methodology/approach: A heuristic method is proposed for scheduling tasks that should be carried out by a group of agents. An important issue is that the agents have to complete their negotiations to commit to the task and carry out it in real time. This task was solved by splitting the problem solving into two layers which make it possible to obtain the rough solutions first and further improve them next as much as possible. Findings: The first layer proposed refers to tasks assigned static priorities and its goal is to meet the real time requirements. Next, the agents can proceed to obtain the optimized solution, provided there is time to do it before the task execution has to begin. Research limitations/implications: Analysis of the negotiations procedure was done and model examples were worked out to develop a planning system based on these design requirements. Practical implications: Implementing the real time task execution planning in artificial agents brings them closer to the job shop floor real life requirements, making it possible to develop software entities mimicking reactions of humans and capable of joint task execution planned on the fly as needed. Originality/value: Analysis of the real time task planning and execution of groups of agents.