Research On Agent Path Planning Method With Linear Temporal Logic Constraints

Path planning is the basic guarantee that an agent can complete its tasks autonomously,it has always been a hotspot and a difficult point in the field of intelligent navigation technology and research.The traditional path planning method has a single task mode,it cannot meet the complex task requirements including temporal sequence and cyclicity and so on,while the linear temporal logic language has high similarity with natural language and extremely rich ability to express the behavior of agents,which can describe large-scale complex task requirements in various practical applications.Therefore,it is of great practical value and theoretical significance to do the research on agent path planning method with linear temporal logic constraints that can meet the requirements of different complex tasks in a finite deterministic system.The main research contents of the thesis are summarized as follows:(1)In order to realize the multi-point patrol function of the hospital drug distribution agent,an optimal patrol path planning method with linear temporal logic constraints is proposed according to the actual application requirements of the drug distribution link in the hospital.(2)For the fixed-point operation and point-to-point driving problem of unmanned campus sightseeing vehicles(agents)in colleges and universities,an improved optimal patrol path planning method with linear temporal logic constraints is proposed,which removes unnecessary path search link and reduces the computational complexity of the algorithm.(3)For the complex task requirements including strict temporal sequence and obstacle avoidance,this paper proposes a global optimal path planning algorithm with linear temporal logic constraints to realize the global path planning for the agent,ensuring that the agent can achieve security obstacle avoidance in complex environments and complete multi-point visit tasks in strict temporal sequence.(4)For the problem of agent path optimization depending on time-varying parameters(dynamics)in a finite deterministic system,this paper propose a receding horizon control strategy that satisfies both properties for weighted deterministic transition systems and full linear temporal logic specifications.The control objective is to maximize the collected time-varying rewards associated with states of the system while satisfying the high level linear temporal logic task specification.What is more,this paper allows a more general cost function by the introduction of the terrain factor.Meanwhile,iteratively re-planning and optimizing the cost function in the finite horizon by forcing certain terminal constraints in the optimization problem ensure correctness of the produced infinite trajectory and recursive feasibility of the optimization problem solved at each time step.In this paper,MATLAB simulation experiments are respectively carried out for the above methods.The simulation results demonstrate the feasibility and effectiveness of the proposed methods.