基于ARM-Linux的爬壁机器人控制系统研究

发布时间：2018-01-14 10:25

本文关键词：基于ARM-Linux的爬壁机器人控制系统研究　出处：《天津理工大学》2017年硕士论文　论文类型：学位论文

【摘要】：本课题在船舶清洗爬壁机器人的平台上,研究了一种基于ARM-Linux控制平台的机器人控制系统。根据爬壁机器人对控制系统的需求,确定了爬壁机器人上位机和下位机的分布式控制方案。上位机和下位机都是以ARM9芯片作为控制核心。下位机外围电路还包括:Zigbee通信模块电路、模拟摄像头、远程图传系统发送端、高压水与真空度检测电路、高压水真空阀门与气动马达转速控制电路、MMA7455三轴加速度传感器电路。上位机还包括LCD显示屏电路、摇杆电路、远程图传系统接收端、Zigbee通信模块电路。爬壁机器人通过图像识别来实现爬壁机器人的自动运行。爬壁机器人实现首先要对工况图像进行采集,对图像进行灰度化和二值化来实现循迹引导线的提取,通过跟踪循迹引导线就可以完成爬壁机器人的自动运行。为了实现爬壁机器人跟踪循迹引导线的稳定、准确与快速,引入了强化学习Q-learning算法,以爬壁机器人循迹线在图像中的位置作为状态值,以左右两侧气动马达的气压差值作为动作值,实现了爬壁机器人的强化学习循迹,改进了传统PID等算法无法针对环境进行最优化动作策略选择的缺点,提高爬壁机器人在不同环境下循迹的准确性。爬壁机器人控制系统软件架构基于Linux操作系统,Linux操作系统具有可靠性高、稳定性高等优点,能够实现多任务的并发运行。采用Linux操作系统能够解决系统调度和工况图像实时显示的问题。Linux操作系统实现对硬件的操作需要在内核层进行,因此针对硬件系统进行了内核驱动程序开发和应用层程序开发。通过对爬壁机器人的无线通信的可靠性、图像数据的准确与稳定性、控制系统对爬壁机器人运动的控制效果测试可知,爬壁机器人控制系统的无线通信保证99%置信度的-610误码率,图像采集并解码后的视频数字信号波形稳定,图像显示无延迟,对爬壁机器人运动的控制能够保证稳定的状态,自动运行状态下能够在测试区域内实现强化学习循迹算法收敛,完成遍历。
[Abstract]:This topic in the vessel wall cleaning robot platform, this paper proposes a ARM-Linux control platform of the robot control system based on wall climbing robot. According to the requirement for control system, the distributed control scheme of climbing wall robot upper and lower machine. The upper computer and lower computer is based on ARM9 chip as the control core slave machine. Peripheral circuits include: Zigbee communication module circuit, analog camera, remote image transmission system transmitter, high-pressure water and vacuum detection circuit, high pressure water valve and pneumatic motor speed control circuit, the MMA7455 three axis acceleration sensor circuit. The PC also includes a LCD display circuit, rocker circuit, remote image transmission the receiver system, Zigbee communication module circuit. The wall climbing robot by image recognition to realize the automatic operation of the wall climbing robot. The robot must first carry on the working condition of image Collect, extract grayscale and binarization to achieve the tracking guide line of the image, by tracking the tracking guide line can be completed automatically. In order to realize the wall climbing robot climbing robot tracking tracking guide line is stable, accurate and fast, with reinforcement learning Q-learning algorithm, with the wall climbing robot tracking in line the location of the image as a state value to air pressure difference between the left and right sides of the pneumatic motor as the action value, realizes the reinforcement learning of wall climbing robot tracking, improved the traditional PID algorithm to select the optimal environmental action policy shortcomings, improve the accuracy of wall climbing robot tracking in different environments. The wall climbing robot control the system software architecture based on the Linux operating system, Linux operating system has high reliability, high stability, concurrent operation to achieve multi task operation using Linux. As to the.Linux operating system to solve the problem of system and operating system scheduling real-time image display to achieve the operation of hardware needed in the kernel layer, so the kernel driver development and application layer development in hardware system. The reliability of the wireless communication of the wall climbing robot, accuracy and stability of image data, the control system of the control effect of wall climbing robot test shows that the control system of the wall climbing robot wireless communication to ensure 99% confidence -610 BER degree, image acquisition and decoding of digital video signal waveform is stable, image display without delay, control of the wall climbing robot can guarantee the stable state, automatic operation in the test area to achieve reinforcement learning tracking algorithm convergence, complete traversal can state.

【学位授予单位】：天津理工大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP242

【相似文献】