Research On Human-automation Collaborative Path Planning For Multi-agent

Mission planning is a Human-Automation Collaborative process,Mission planning system is a classical Human-Automation Collaborative system that involves the integration of human decision making and an automated system.During developing such a collaborative system,the question often arises as to where,when,and how much humans and automation should be in the decision-making loop.Mission planning system will gradually face the increasingly complex,dynamic or uncertain planning environment,and the number of planning objectives has become increasingly large.The use of Human-Automation Collaboration technology to solve such problems is an effective method.This paper researches the problem of collaborative planning of multi-agent in the advanced planning stage,focusing on solving the problems of low efficiency,low success rate and inflexibility of Human-Automation Collaboration.The main research contents are as follows:1.Formal description of commander decision-making events in the process of Human-Automation Collaborative mission planning.The paper extends the cooperative problem solving(CPS)modal operator of Wooldridge and Jennings,and proposes a formalized representation of the decision-making in the process of multi-agent collaborative mission planning,which enables the decision-making in the collaborative planning of the mission can be integrated into the mission planning algorithm of the machine.2.Solving method of multi-agent oriented flexible Human-Automation Collaborative mission planning problem.A solving method of flexible Human-Automation Collaborative mission planning problem based on multi-agent is designed.Then multi-agent collaborative planning tasks in large-scale,dynamic and uncertain environments can be dealt with flexibly.It can be applied to the joint operations of multiple arms and services.It can solve the problems such as low planning efficiency,low success rate of planning results and inflexibility of Human-Automation Collaboration when the plan targets’ number is huge in the planning stage.For the multi-agent Human-Automation Collaborative mission planning problem,the performance evaluation parameters of the flexible Human-Automation Collaborative mission planning is proposed,which includes the overall system performance and behavior evaluation criteria.The performance of the solving method of the Human-Automation Collaborative mission planning problem is analyzed.3.The application of flexible complementary Human-Automation Collaborative architecture in multi-agent collaborative path planning.This paper develops a RRT algorithm suitable for Human-Automation Collaboration,improves the priority-based distributed priority routing algorithm,and develops a dynamic priority-based distributed priority programming algorithm for Human-Automation Collaboration.The above two improved algorithms in the planning process can be integrated with commanders’ formalized decision-making,achieves a flexible Human-Automation Collaboration in the planning process,effectively reduces multi-agent collaborative planning failure rate in the dynamic environment or uncertain environment.4.The development and experimental analysis of multi-agent oriented Human-Automation Collaborative path planning software system.In this paper,a dynamic priority-based distributed priority programming algorithm and a Human-Automation Collaborative RRT algorithm are applied in software system.A software simulation platform is built to evaluate the dynamic priority-based distributed planning algorithm and Human-Automation Collaboration for multi-agent path planning.Then this paper carries out a detailed experimental analysis.Experiments show that the flexible complementary Human-Automation Collaborative architecture and the dynamic priority planning algorithm proposed in this paper are reasonable and effective,which can reduce the conflict of the large number of agents and reduce the failure rate of multi-agent collaborative planning.When the number of agents in greater than 250,the failure rate can be reduced by more than 30% by the dynamic priority algorithm based on flexible complementary Human-Automation Collaboration;the dynamic priority algorithm based on flexible complementary Human-Automation Collaboration generates only 2.5% more path than the path generated by the static priority,and uses a planning time of 96% to 112% more than the planning time of the static priority.