Font Size: a A A

Robot Path Planning For Rescuing Trapped People With Limited Life Strength

Posted on:2017-04-24Degree:DoctorType:Dissertation
Country:ChinaCandidate:N GengFull Text:PDF
GTID:1108330509454781Subject:Control theory and control engineering
Abstract/Summary:PDF Full Text Request
Robot rescue path planning is the key to guarantee the rescue missions to carry on smoothly and effectively, which can be attributed as the problem of robot path planning. It can be described as: after a disaster, one or more robots are used to rescue multiple trapped people, a rescue path is planned to rescue as many targets as possible in a limited time. Therefore, the premise of rescue path planning is to plan an effective path. For the problem of robot path planning, it is a NP hard problem. Recently, there are many achievements about robot path planning, some traditional methods only can be employed into some simple rescue environments, while for the complex environments, they are no longer suitable to solve such rescue problems. Since Particle Swarm Optimization(PSO) algorithm is simple, has few parameters and quick convergence speed, it has been successfully applied into many path planning problems, we study the problem of robot rescue path planning based on PSO, and design the specific strategies for different rescuing scenarios.Firstly, the PSO method for the problem of path planning of single robot for single type of rescue tasks is proposed. Firstly, the function of the targets’ life strength which changes over time is given; then, the fitness function related to the life strength is established which is the number of rescued targets; then, the modified PSO is designed to solve the fitness function, including: particle’s position and velocity update method, particle’s local and global best solutions update method, correction for the unfeasible solutions which can guarantee the particles feasible after update. We apply the proposed method into three different rescue scenarios, and compare it with the PSO method without modifying to verify the effectiveness of the proposed method.Then, considering the real situation, when more targets need to rescue, the efficiency of single robot is much low, if we use multiple robots to perform the rescue mission instead of single robot, it can surely improve the rescue efficiency. Hence, we study the rescue problem of path planning of multiple robots for single type of rescue tasks based on PSO. Firstly, Petri net with time constraint is established, then, PSO is employed to optimize the established Petri net to get its best fired transitions, and then the fitness function is formulated by combining the characteristic of the problem, and the particle’s global best and local best update methods are given as well. Besides, the particle decode method is provided to make all the transitions be totally assigned to each robot without overlap, which can avoid the generation of unfeasible solutions. Finally, employing the proposed method to solve different rescue scenarios, and comparing it with other methods, the simulation results show that the proposed method is effective and can get better solutions than other algorithms.Normally, the rescue type is not single, multiple rescue types of rescue tasks are common in real situation, and the above mentioned problem cannot solve the rescue problems in such situation. Therefore, we study the problem of path planning of multiple robots for multiple types of tasks based on PSO. Firstly, taking the multiple types of tasks and the targets’ limited life strength into consideration, the priority set of each task for each robot is given, based on it, the robot’s rescue sequence can be modified by it; then, the average rescuing time for each target is treated as the fitness function, and the life strength is considered as the constrains, thus, the mathematical model of the problem is established. For the PSO method, we try to give two different PSO algorithms to solve two different rescue scenarios.Firstly, for the centralised optimal rescue problem, all the robots can communicate with each other, herein, the centralised PSO method is employed to solve this problem, it includes: particle’s decode method, particle’s local and global best update methods, and its local search method. Secondly, for the distributed optimal rescue problem, not all the robot can communicate with each other, their communication is limited, thus in this paper, we use distributed method and propose multi-swarm PSO method to solve the problem, where one swarm is used to optimize the rescue sequence for one robot, and the robots communicate with each other by using different topologies. The modified multi-swarm PSO method includes: particle’s decode method, particles’ position and velocity update method, particle’s local and global best update methods, and the communicate strategy between different robots and so on. Finally, the two methods are employed into many kinds of rescue scenarios, and compared them with Genetic Algorithm, the traditional PSO method, Consensus-Based Bundle Algorithm(CBBA), Performance Index(PI), and modified PI(PI with softmax, PI-softmax; PI with e-greedy, PI-e-greedy), the simulation results show that the centralized PSO method is always the best among all the above listed algorithms, and the proposed distributed PSO method is always the best among all the above listed distributed algorithms.Further, taking every factor in the rescue scenario into consideration, the targets’ life strengths we got are uncertain, but with little different from the actual values, therefore, we treat each life strength as an interval, and formulate their interval functions, which change over time. Based on it, the method for path planning of multiple robots for uncertain multiple types of tasks based on PSO is given, it mainly includes: interval comparison method, particle’s global update method and local search method. We apply the proposed method to solve the rescue problem in different scenarios, and the obtained results verify the effectiveness of the proposed method.Finally, considering other more complex scenario in real situation, PSO is employed to solve the problem of uncertain path planning of multiple robots for multiple types of tasks in an uncertain life strength scenarios. The mathematical model of the objective function for the proposed problem, the particles’ coding and decoding methods, the particle’s local and global solutions update methods and local search method are given, and also we employ the proposed method to solve different rescue scenarios to verify its effectiveness.The results we obtain in this dissertation can not only enrich the theoretical knowledge of robot rescue path planning, but also improve the efficiency of robot path planning, thus greatly save the cost and time of rescue, therefore, have important theoretical and practical values.
Keywords/Search Tags:Robot, Life strength, Rescue, Path planning, Particle Swarm Optimization
PDF Full Text Request
Related items