摘要
采用以DSP+ARM微控制器为核心的嵌入式实时操作系统,设计了一种基于嵌入式系统和机器视觉定位的室内移动机器人。利用视觉导航图像处理技术、形态学方法和一种基于尺度空间理论的Harris角点检测方法,借助陀螺仪和加速度计的惯性导航技术进行地图的匹配定位,并按环境的变化情况更新地图以实现导航。基于超声波传感器设计了避障模块,实现了自主避障。设计了一种基于Zig Bee技术的无线通信模块,实现了机器人的智能控制,增加了机器人之间以及机器人和服务器之间的信息交换。软件核心算法采用多传感器融合技术,将D-S理论和人工神经网路相结合;在非线性化系统中,利用BP神经网路多层前馈网络的反相传播学习方式,很好地实现了模式识别。与其他机器人系统相比,该系统具有独立操作性强、功能多样化、扩展性强等特点,克服了目前机器人存在的成本高、功耗大、实时性差和定位不准确的问题。
By adopting the embedded real -time operating system with DSP +ARM micro controllers as the core, the indoor mobile robot based on embedded system and machine vision positioning is designed. By using visual navigation image processing technology,the morphological method,and the Harris corner detection method based on scale space theory,and with aid of inertial navigation technology of gyroscope and accelerometer, the map matching positioning can be conducted, and the map can be updated in accordance with the variation of environment for implementing navigation. Based on ultrasonic sensors, the obstacle avoidance module is designed to realize automatic obstacle avoidance. The wireless communication module is designed based on ZigBee technology,to achieve intelligent control for robot,and implement information exchange between robots and between robot and server. By using multi-sensor fusion technology, the D-S theory and artificial neural network is combined. In nonlinear system,with help of inverse propagation learning mode of BP neural network multi-layer feedforward network,pattern recognition can be implemented well. Comparing with other robot systems,this system possesses powerful independent operability,functional diversification,and strong extensibility, and it overcomes the disadvantages of current robots, such as high cost, large power consumption,poor real time performance and inaccurate positioning.
出处
《自动化仪表》
CAS
2017年第2期49-52,共4页
Process Automation Instrumentation
关键词
机器人
智能控制
嵌入式系统
神经网络
传感器
视觉定位
导航
无线通信
Robot
Intelligent control
Embedded system
Neural network
Sensor
Vision positioning
Navigation
Wireless communication