Estimation Of Collecting End Pose Of Bridge Detection Robot Based On Sequence 2D Structured Line Features

With the rapid development of the construction of China's roads,highways,railways and high-speed railways,the number of bridges has also increased rapidly,and China has become the first bridge country in the world in the filed of bridges.When we increase the bridge construction,we also need to pay attention to the detection and maintenance of the bridge,and reduce the hidden dangers of the bridge accidents.At present,the bridge repair are artificial or semi-manual-maintenance,and it is ineffective,risky,with missed inspection and false inspection.Therefore,with the rapid development of sensor technology,robot technology and image processing technology,the research and development of high-tech and automated bridge detection robot has naturally become a very important subject.This paper relies on the research project of "intelligent bridge detection robot equipment and system",which is based on the joint research project of the instructor and the Hunan Provincial Traffic Science Research Institute.The goal is to develop the autonomous bridge detection robot and achieve the intelligent bridge detection.In the bridge detection robot system,the acquisition of the end pose information not only provides real-time information for the acquisition path planning,but also provides a priori knowledge for the subsequent splicing of the bridge image.It is the foundation of the bridge detection robot system to complete the autonomous bridge detection.In order to complete the real-time pose estimation of the end of bridge detection robot,based on the analysis of the actual environment of bridge,we researched the pose estimation scheme and algorithm of the end.The main work is as follows:(1)The research on bridge structure line detection based on sequence 2D structured line feature is carried out.Based on the analysis of the structural line characteristics and the line contour model of the bridge,the detection of the structured line feature based on the Hough transform and the LSD is studied.An algorithm for merging straight segments is proposed,as well as the detection algorithm of structured corner feature and structured contours Feature based on structured line feature.And the simulation experiment is designed to verify the algorithm.(2)Based on the detection of 2D structured line features,the end pose estimation algorithm is studied.Firstly,the geometric transformation of photography in European space and the process of the end pose estimation is studied.Then,the initial pose estimation is achieved under the hypothesis-test framework,and a new straight-line distance measurement rule is proposed.Thirdly,the continuous pose estimation is achieved under the least square framework,and a weighted least squares algorithm is proposed.(3)Based on the analysis of the bridge detection robot system,the pose estimation system of collecting end of the bridge detection robot is designed.Based on the 2D structure of the bridge,a hardware system based on raspberry and Hokuyo single line lasers is used to develop the end pose estimation scheme based on the characteristics of the sequence 2D structured line.Finally,the application of pose estimation system based on the algorithm proposed in this paper in practical bridge detection is introduced.