A path planning study of autonomous mobile robot in dynamic environment

Abstract

In the dynamic environment, a path planning method of autonomous mobile robot based on fusing and genetic simulated annealing approach is introduced. In this paper ,the accurate dynamic environment data is acquired firstly by building ultrasonic data predictive model and the method based on multisensor fusing. Finally, on the based of path planning by using genetic algorithms, forming a genetic simulated annealing approach by inducting simulated annealing approach to search the global optimal path.The optimum problem of robot path planning is, according to some optimal rule, to search optimal stepped course from the start point to the end point and to avoid obstacle in its working environment. There are many paths planning approach dealing with the known environment. In recent year, people pay more and more attention to the problem of robot path planning in dynamic environment in which uncertain obstacle exist. Plenty of methods have been applied in dynamic path planning for mobile robot at home and abroad such as vector field method [1,2] , grid method [3] and neural networks [4] . These methods with advantages of practicability in calculation, easy control in kinetic optimization as well as real time path plan for robots, however, they cannot ensure to search the globe optimal path. Genetic algorithm [5] is efficient path planning method as its globe convergence and implicit parallelism. However the standard GA cannot ensure the quality and efficiency of solution of the path planning problem because of some limitations such as easy premature convergence and bad local searching ability. In this paper, a genetic simulated annealing algorithm [6,7] was presented by combining the simulated annealing approach to Genetic algorithms for solving the path planning problem. I. ULTRASONIC TRANSDUCER DATA FUSION The important premise to solve the path planning problem is to obtain accurate dynamic information. The accurate navigation and obstacle avoidance of autonomous mobile robot, which obtain environment information with ultrasonic transducer, were influenced by the apperceiving error of the ultrasonic transducer. The problem was solved by building transducer forecast model based on data fusion of transducers. A． Transducer Forecast Model Suppose there be 2m+1 ultrasonic transducers on the mobile robot which well-proportioned distributing by a semicircular form. The ultrasonic data of the ith transducer are ( ) i u k at the time of k. The data before the time of k make a vector, they are ( ) ( ) ( ) ( ) ( )

Real path planning based on genetic algorithm and Voronoi diagrams

“…In this regard, and in order to provide robust, flexible, tolerant to changes in the environment, reliable and computationally efficient solutions, many different approaches have been proposed to address this problem, including neural networks [1], reinforcement learning [2], bioinspired methods [3], [4], graphs [5], [6], [7], evolutionary algorithms [8], [9], [10], [11], [12], [13], [14], among others [15], [16], [17].…”