Tracing Robot Based On The Autonomy Of Embedded Linux

At present, manual operation is used in the most adverse circumstances and dangerous environment of the industrial scene, however, its high-risk, poor security lead to the urgent need for intelligent guided robot research and development. And now, the research development on guiding robot is mostly based on the MCU and DSP, so there is a lot of limitation, such as the relative single function, weak expansibility, and so on. Thus it can not make the guided robot accomplish more and more complicated tasks with higher intellectualized character. Therefore, this paper based on the intelligent character of robot, adopts more advanced embedded MCU as a host-controlled chip, makes transplanted Linux operating system as a robot's brain, and research the speediness and stability of the traveling automatic line-tracking robot on the black tracking line of the white course. The main research contents in this paper include the following aspects:1, It has constructed the intelligent robot's visual guidance system, designing and developing each hardware module of the robot.2, Model analysis to turning of servo, based on the robot's body axis relative to the direction and lateral deviation of the path midline as input design a fuzzy controller for the turning of servo. Model analysis to robot's speed control designs a classical PID controller and a single neural network-based PID controller.3, It has successfully transplanted Linux2.6 kernel to S3C2440 core chip, meanwhile, transplanted a matched BIOS startup program UBOOT and made the root file system JFFS2, and has accomplished drive and application procedures based on Linux2.6 kernel according to the robot's hardware design, and in the end, made testing, comparing and analysis to the self line-tracking robot by various calculating methods.This paper has used embedded Linux system to the intelligent guided robot. It is proved by experiments that the robot can accomplish automatic line-tracing task fleetly and stably after input the road surface information collected by the CCD senor. It is also proved that robot's speed can be controlled well by single neural network PID controller in curve paths, according to the testing and comparing to the two calculation methods of robot's speed controlling on reality condition.