Location Detection System Based On Machine Vision And Inclination Sensor And Its Verification

In the next few years,coal will remain the main energy in China's energy system.The automation of comprehensive excavating face in coal mine can greatly improve the production efficiency.At present,China's underground comprehensive excavating work is generally dominated by single roadway excavation with boom-type roadheader,so the location detection of boom-type roadheader is the basis and premise of the automation of comprehensive excavating working face.Only by obtaining stable and reliable real-time location of roadheader,can the automation and intelligent transformation of comprehensive excavating face be carried out on this basis.In this paper,a location detection method of boom-type roadheader based on machine vision and inclination sensor is proposed,and research is carried out on pose parameter solving,laser target parameter optimization,laser target automatic calibration,laser spot target recognition and other technologies.The detailed research content and work results are as follows:The overall scheme of the location detection system of the boom-type roadheader based on machine vision and inclination sensor is proposed.The indoor experimental prototype and workshop experimental prototype are manufactured.The location detection software system is designed.On the basis of the overall scheme,the modeling analysis of the underground comprehensive excavating working face is carried out,and the calculation formula of the inverse solution of the five position and pose parameters is given.Aiming at the working conditions of the location detection system and the small volume demand of laser target in practical application,genetic algorithm is used to optimize the laser target parameters.The optimization parameters are simplified by the symmetry principle,and the pose combination is analyzed to simplify the constraint equation to improve the efficiency of the algorithm.The optimization algorithm is verified by the forward kinematics simulation and the comparison of multiple optimization data Effectiveness.In the machine vision part,Zhang Zhengyou's calibration method is used to calibrate the camera's internal and external parameters and correct the distortion.An automatic calibration method of laser target based on checkerboard calibration board is proposed to reduce the measurement error caused by the size difference between the measurement model and the theoretical model.The dynamic target detection method based on background difference method is used to detect laser spot target.By means of image preprocessing,the influence caused by weak interference such as large area background and shadow area change is removed.The recognition method of laser spot target based on HOG feature and support vector machine classifier is used for laser spot recognition under strong interference condition.Finally,the center of the laser spot target is positioned and the pixel coordinate of the center of the laser spot is obtained.Finally,a series of experimental design and data analysis are carried out.The effectiveness of Kalman filter method in the real-time angle measurement of inclination sensor are verified by the data filtering experiment of inclination sensor.The angle measurement errors and offset measurement errors of the location measurement system of the boom-type roadheader are obtained through indoor experiments,which proves that the system can meet the requirements of measurement accuracy.Through workshop experiments,the recognition rate and processing time of the image processing module for the laser spot target under different interference conditions are obtained,which proves that the system has good reliability and high real-time performance.