Autonomous Robot Localization Research Based On Stereoscopic Vision And Inertial Data Tight Coupling

To complete the set up of entire system in this paper,Nvidia's mobile processor TK1 was used as the core processor,Germany's IDS's industrial camera ueye series binocular stereo image acquisition system was combined with,and the product of Microstrain's 3DM line IMU inertial sensors was applied.The system above was placed on the differential drive wheeled mobile robot to realize the attitude detection of the robot and study the performance of several vision distance calculation methods in autonomous robot positioning.Finally,the positioning performance improvement was achieved by optimizing the data fusion.The main work are summarized as follows:First of all,a detailed introduction and in-depth study were made to explain the theoretical principle for robot,which achieved independent positioning through binocular stereo vision and inertial information.It mainly involved binocular vision odometer and the principle of inertial positioning,including the mathematical description(matrix,Euler angles,quaternions)of the rigid body(robot)pose description,the representation and transformation of the coordinate system,and the principle of inertial integration and related basic equations etc.The theory foundation of this dissertation was elaborated through the introduction of mathematical principles.Secondly,according to the set application model,the entire hardware core system was build.The selection of the entire hardware system,the process of building were described in detail and the various functions were analyzed in modular,including the establishment of stereoscopic vision,inertial information acquisition,the establishment and control of wheeled robots.The initialization of related equipment settings were completed,such as the camera calibration,the coordinate system fixation and conversion of entire system etc.For achieving the follow-up function,an excellent hardware platform with completed performance was set up.Again,write algorithms for the hardware.It mainly included the introduction and analysis of two commonly used visual distance calculation method,inertial positioning algorithm,the whole algorithm operation foundation(ROS),and in-depth study of algorithm implementation and principle.A comprehensive positioning method of visual and inertial information fusion coupling were proposed,and its principle and feasibility were analyzed.Finally,the whole system was used to collect the actual operating data and experimentally verified the relevant algorithms.It mainly related to the experimental environment,the specific data examples,visualization results showing,the relevant performance analysis charts,etc.,the system stability and real-time performance were continuously improved through the experiments.In summary,this paper introduced the system from the aspects of hardware and software by setting up the whole binocular stereo vision and inertial wheel mobile robot system,and put forward the performance improvement suggestions.The system was also used to verify the algorithm.