The Design And Fabrication Of An Indoor Positioning Tracking Robot Based On Multi-sensor Fusion Technology

With the rapid development of science and technology,mobile robot technology is becoming more and more mature.Compared with repeating simple mechanical activities initially,the robot has developed to an intelligent mobile robot with independent perception and decision-making ability,in which positioning plays a vital role.Unlike traditional positioning equipment such as compass,inertial navigation,wheeled odometer,and so on,visual odometer has the advantages of wide sensing range and large amount of information,which has become an important research direction in the field of robotics.The RGB-D visual odometer studied in this paper is a visual odometer based on depth camera.It can collect both RGB and depth image information at the same time,and the positioning accuracy is higher.But because of the dependence on the image information,it will have the case of matching failure in the special scene,and the cumulative error of the system can not be corrected because only a single sensor is dependent on it.Aiming at the indoor environment,this paper mainly studies the location algorithm of multi sensor fusion to improve the positioning accuracy and develops a RGB-D camera based robot.It includes the configuration of the robot operation core system,the selection of environmental sensing sensors,the debugging of the control core and the motor control,and the human-computer interaction interface based on the python GUI etc.Then,the positioning and tracking algorithm is studied on the completed mobile robot platform.First,an extended Calman filter is used to fuse the attitude information of RGB-D vision positioning and inertial navigation positioning,and the improvement of system positioning is studied by multi-sensor fusion.Second,in view of the failure of the system due to the failure of the visual odometer under special circumstances,we propose an improved location algorithm.It is mainly through the logarithm setting threshold for the feature matching of the visual odometer.When the feature matching logarithm is less than the threshold,the visual mileage is invalid.At this time,the inertial navigation and the wheel type integrated orientometerare used to complement the long-term stability of the system.This paper also uses visual tracking technology based on RGB-D camera.Firstly,the image is filtered and de-noised,the following target is automatically identified and the center of mass is obtained.Then the position of the centroid relative robot is obtained by coordinate transformation,and the following target is finally realized.Finally,the positioning precision of the wheel inertial range meter is compared with the RGB-D visual odometer and the visual inertial navigation meter in the same situation.It is found that the positioning accuracy of the RGB-D vision inertial integrated odometer is relatively optimal.Through the experiment on the positioning function of the system under special circumstances,it shows that the system can use the wheel inertial integrated mileage for the supplementary positioning in the case of the failure of the visual odometer,which proves the feasibility of the improved positioning algorithm.Through the analysis of the experimental results,it is found that the following robot can follow the target more accurately and steadily,proving the feasibility of the following algorithm.