Indoor Location Algorithm And Realization Based On Kinect And Inertial Sensor

Mobile robot positioning technology is an important technology to realize robot autonomous navigation.Due to the continuous progress of computer vision technology and image processing technology,a lot of positioning technology based on vision sensor has been applied to robot positioning system.In visual positioning,in order to effectively deal with the phenomenon of ambiguity and difference of visual Angle,as well as the failure of inter-frame feature matching,the ORB feature extraction and matching algorithm with good stability is determined to be adopted for the total number of feature points matching and the total number of feature points collecting,and the algorithm is improved.Then,the PnP method is used to estimate the rotation matrix and translation vector,and the sparse point cloud map and trajectory are obtained by means of back-loop detection.Finally,the visual positioning of the robot is realized.Aiming at the phenomenon of large positioning error caused by the interaction of multiple influencing factors(itself and environmental factors)that must exist in the positioning of Kinect sensor,this paper proposes a dynamic combined positioning method based on integrated inertial navigation.The data of Kinect and IMU sensors were synchronized in time and space.Then,the extended Kalman filter was used to establish the framework of joint positioning fusion algorithm,and a process model was created based on the positioning information of ins.The visual information collected by Kinect sensor and the inertial navigation information of IMU sensors were used to construct the observation model.In the robot positioning test,the laboratory environment was selected,and the test results were as follows:compared with the positioning of independent visual sensors,the navigation and positioning mode combining visual and inertial information has certain advantages in positioning accuracy.