| The obtain of the indoor robot's position in real-time and accurately becomes the premise as well as the foundation of robot operation.The warehouse corridor environment belongs to a special indoor environment.The whole warehouse is large,while the location area in a single corridor is limited.Besides,the similarity between different corridors is high.Because of the particularity of environment,localization algorithm should have strong description ability for robot position distribution.Monte Carlo localization method uses particles to describe position distribution of robot.Particle set can effectively cover location space and flexibly adjust the range of distribution.However,there are still following shortcomings in the direct application.Firstly,the method lacks effective particle number adjustment strategy,too many particles lead to low positioning efficiency while too few particles lead to low positioning accuracy.Secondly,there is no way to identify corridors,and the range of particle distribution is difficult to achieve flexible adjustment.This paper takes Monte Carlo localization method as the research content,aiming at the shortage of location algorithm in warehouse and corridor environment,the content of improved parts is as follows.(1)A modified method of likelihood model for laser radar in motion state.In view of the effect of the motion of the robot on the measured data during the laser radar scanning period,a likelihood observation model considering the motion characteristics is proposed,which compensates the error caused by the motion to the observation model and improves the accuracy of the localization algorithm.(2)A corridor feature and grid mixed map considering the location relationship between packages.According to the characteristics of warehouse corridor environment,a package chain map describing the packages and their relationship is set up,combines with grid map,the whole environment is well described.The establishment of the mixed map is the basis for distinguishing different corridors and locating accurately.(3)A sampling strategy which dynamically adjusts the number and scope of sampling.In order to solve the efficiency problem of Monte Carlo localization algorithm,an adaptive bin size KLD sampling strategy is proposed.This strategy adjusts the sampling number dynamically by adjusting the size of the sampling unit.Besides,according to the search result of the location relation of the package,the strategy dynamically adjusts the size of the sampling area.Finally,the proposed method is applied to the Monte Carlo location algorithm to complete the position tracking and global positioning in the experimental warehouse environment.The experimental results show that the adaptive Monte Carlo localization algorithm improves the efficiency of the algorithm while ensuring the positioning accuracy,and meets the location requirements in the warehouse corridor environment. |
PDF Full Text Request