A Design Of Automatic Tracking Robot System Based On Machine Vision

With the continuous advancement of urbanization,the construction industry has unprecedented opportunities for development.As one of the important branches of the construction industry,interior decoration has experienced rapid development in residential decoration and public building decoration.Floor tile caulking is an indispensable part of interior decoration,people usually using a glue gun to squeeze the grout/caulking agent,this method is not only time-consuming and laborious,but also easily overflows a large amount of materials.So it's causes waste of materials and make trouble for the cleaning work.Aiming at the above problems,an automatic tracking robot based on machine vision is designed.With this robot as the carrier,it can be combined with different glue gun extrusion equipment to realize floor tile caulking work.This robot consists of three parts: image processing unit,motion control unit and remote monitoring unit.Image processing unit refers to real-time shooting of ground information through camera,then using ITTI saliency target detection algorithm,Canny edge detection and Hough transform to extract floor tiles,and the gap edge information is extracted by the straight line fitting,finally the direction,angle and distance calculation in the local path are used to determine the deviation between the robot and the navigation line and the correction command is sent to the motion control unit through the serial port.The motion control unit is a vehicle-type robot system that is controlled by the STM32,and controls the robot to perform operations such as forward,backward,left turn,right turn,left shift,and right shift according to the correction command received from the image processing unit.The remote monitoring unit receives the real-time image of the camera and the processing result of the image processing unit through the PC client,which is convenient for the construction personnel to perform remote monitoring of multiple devices.The image processing unit runs on the NVIDIA embedded development board Jetson Tegra K1 and is equipped with Ubuntu14.04+OpenCV3.1.0.The experimental results show that the algorithm can accurately identify the floor tile gap and extract the navigation line whether it is a smooth floor tile or a floor tile with a complex background.The ROS(Robot Operating System)is used to integrate the three parts.The test result shows that all parts of the system communicate is correctly,and the set functions can be accurately completed with good reliability,accuracy and real-time.It provides a firm foundation for the complete realization of the subsequent caulking robot.