Research On Indoor Robot Location Technology Based On Distance Measurement Information

With the development of artificial intelligence and the development of robots,robots are widely used in many fields,such as outer space exploration,medical service,smart home,industry and so on.And not only that,robot has gradually entered our daily life.However,obstacle layout changes and the dynamic changes in the indoor environment make the location of robot in indoor environment is unknown.In order to make the robot can efficiently complete the various tasks in the indoors environment,autonomous robot location is a prerequisite.Therefore,it is of great significance to research on indoor location technology of robot.This paper describes a design of a robot indoor location system based on range sensors.The design uses sonar range sensor and laser radar range sensor as sensing system to detect environmental information in the robot experiment scenes for data acquisition and analysis in order to complete the indoor environment localization.Because single sonar information can only provide partial environmental information in most cases,in order to improve the reliability of the data joint the multiple sonar information to make the robot location more accurate.The obtained sonar information are transformed into 2D shapes and extracted feature by the ring projection method,and the inner relations of the multiple sonar information are extracted and the classification of the scenes is carried out in the improved K Nearest Neighbor classifier;A method of aligning the sonar information is proposed to solve the problem caused by the rotation of the robot,in view of the sparseness of the sonar sensor distribution and the small measurement range,the joint kernel sparse coding method is designed to fuse the inner relations of multiple sonar information and realize the classification of the scenes.In order to verify the indoor scene recognition ability of the two methods,the recognition ability of each experiment is described and analyzed by drawing the confusion matrix when the optimal classification is drawn.The location of the robot requires not only the position of the robot,but also the orientation of the robot.The support vector machine and the extreme learning machine are introduced into the robot orientation prediction analysisto determine the indoor orientation of the robot.The support vector machine regression and the extreme learning machine regression model are established.According to the data characteristics of the robot's range sensor,the output and input factor of the prediction model are determined andthe influence of the model parameters on the prediction performance is analyzed.The above-mentioned indoor scene positioning and the orientation prediction of the robot are tested on the robot.The result shows that a good location effect is made.In the future research,this algorithm will be applied to the real-time operation of the robot,so that the robot can effectively locate itself in the process of performing the task.