基于LabVIEW的多通道视觉测速系统设计

目标运动速度与运动状态的快速预判是机器人避障、目标检测与跟踪的关键。经典的初级运动检测器（elementary motion detector，EMD）模型具有在局部范围内的速度矢量敏感特性，在此基础上结合波峰检测算法，提出一种动目标运动速度和运动状态的快速测速法。为验证所提方法的可行性和有效性，以线性排列的光敏传感器阵列为视觉检测单元，在LabVIEW开发环境中，利用雷赛SMC3380运动控制器和研华PCI-1747U数据采集卡，搭建了具有动目标运动控制及动目标信号采集功能的多通道视觉测速系统；通过多通道数据采集及动目标快速测速处理，实现了目标运动速度和运动状态的判别。实验结果表明：多通道视觉测速系统运行稳定，能够实现最大速度为80 mm/s（绝对误差小于±2 mm/s，相对误差小于±3%），以及最大加速度为20 mm/s²的匀加速运动状态的动目标检测与判断。为机器人视觉感知器的动目标快速预判研究提供了新思路。

Design of multi-channel visual velocity measurement system based on LabVIEW

The fast prediction of the movement velocity and status of a target is critical for robot to avoid obstacle, detect and track target. Based on the elementary motion detector (EMD) model which was advantaged in sensitivity in motion speed detection in local area, a fast detection method for detecting moving target's velocity and status was proposed. The method combined with the EMD and peak detection algorithm. In order to verify the feasibility and effectiveness of the fast detection method for moving target, a multi-channel visual velocity measuring system was designed. In this system, it contained visual inspection unit, multi-channel data acquisition and target motion control device. The visual inspection unit was designed with linear optical sensor array. The multi-channel data acquisition (Advantech PCI-1747U data acquisition) and three-axis motion controller (Leadshine SMC3380) were programmed both in the LabVIEW development environment. Target motion device controlled object moving in scheduled way, and the multi-channel data device gathered object motion information data from visual inspection unit, which could predict the target's velocity and status after processed by the fast detection method. The experiment results showed that the multi-channel visual speed measuring system run stably and achieved in the maximum velocity of 80 mm/s moving target detection (absolute error less than ±2 mm/s, relative error less than ±3%) and the maximum acceleration of 20 mm/s² acceleration measurement. It provides a new idea for the research of rapid prediction of moving targets by robot vision sensors.