The Implementation Of A Patrol System Based On Mobility Robot

With the development of embedded and other related technologies, videosurveillance technology and intelligent robotics-related implementations areincreasingly being put forward, and its products are gradually integrated intopeople’s daily life. Due to the existence of a fixed camera dead, poor flexibility andother reasons, will also become a way to integrate the two current trends products.This paper select Forlinx’s OK6410A ARM11family development board tostudy the embedded Linux kernel and driver development system interrupts, and onthis platform design and development of infrared sensors, ultrasonic gun, camera,and other peripherals DC motor applications and drivers program, and theinstallation of software and hardware platform built for embedded Linux, by readingthe data on the peripheral hardware platform to achieve control of the driver.This paper focuses on image processing in robot patrol functions relatedalgorithms: target separation, image Unicom, target identification, target tracking.For the target separation, we propose a modeling background difference method anddynamically updated based on the background of improved algorithms. Unicom interms of image using convex hull algorithm to obtain the best results. For the targetobject recognition method using Hog features improved target recognition accuracy.for target tracking algorithm using improved CamShift. This paper’s improvementsachieved good results.This paper also studied the robot motion control algorithm based on fuzzycontrol method, the use of sensor data and image processing as a result of the case isdesigned to fit the input mode and the tracking mode patrol two driving strategicplan.