Research On The Vision/inertial Integrated Navigation Method Of The Electric Power Inspection Robot

With the development of power system in China and the concept of smart grid put forward,the electric power inspection robot,which aims to improve the efficiency of power grid inspection and to ensure the safe operation of power system,has been widely used.The electric power inspection robot not only solves the problem of low efficiency and high risk,overcoming difficulty in manual inspection in bad weather,but also improves the reliability of power grid operation by using intelligent detection devices such as HD camera,audio monitoring,infrared thermal imager,etc.In the process of inspection,autonomous navigation and positioning are the premise of the intelligent inspection.Therefore,it is of great practical significance to study the navigation and positioning method of robot with high positioning accuracy,good robustness and strong adaptability to the environment.This paper analyzes the current research on the inspection robot of power grid,and studies the navigation method of inspection robot based on the inertial assisted vision information for practical application scenarios such as substation and distribution station.When robot is unable to work independently in complex or unstructured inspection environment,a man-machine cooperative navigation method based on two-way information fusion is studied.The contributions of this paper are listed as follows:Firstly,this paper studies the feature-based visual navigation method.In terms of feature extraction and feature matching,the performance and efficiency of the mainstream algorithms are compared and analyzed.To solve the problem of mismatching in feature matching,the random sampling consistent method is used to eliminate the mismatching;in term of pose estimation,the optimization method of pose estimation based on EPn P is adopted.Experiment shows that the visual odometer has better positioning accuracy and real-time performance when the features are stable.Secondly,in order to solve the problem of poor robustness and scale drift in robot vision navigation,this paper studies the navigation method based on inertial assisted vision information.The inertial data is preprocessed by the method of pre integration.The initial parameters of the system are estimated by the visual inertial joint initialization.Then,the inertial information and visual information are fused by the tight coupling nonlinear optimization model based on the local map to estimate the position and attitude information of the robot.Navigation experiment is designed by simulating power grid application scenario,showing that the method is of good robustness and positioning accuracy.Finally,in the unstructured,complex and unknown environment or the scene with higher reliability requirement,a man-machine cooperative navigation method based on two-way information fusion is studied to solve the problem that inspection robot fails to complete the task alone and needs man-machine cooperative operation.This method uses the different error characteristics of pedestrian navigation system and robot navigation system to build bi-directional information fusion filters,which can synchronously correct the navigation information errors of both systems.Therefore,the proposed method improves the positioning and heading accuracy of both systems.