一种移动机器人跟踪机动目标的方法  

A Maneuvering Target Tracking Scheme for Mobile Robot

作　　者：蒋荣欣[1] 陈耀武[1] 

机构地区：[1]浙江大学数字技术研究所,浙江杭州310027

出　　处：《江南大学学报（自然科学版）》2009年第1期10-14,共5页Joural of Jiangnan University (Natural Science Edition)　

基　　金：国家863计划项目(2003AA1Z2130)

摘　　要：针对机器人跟踪机动目标,提出了一种完整探测、估计的方法。利用单目视觉定位被跟踪目标的方位,再融合激光数据来获取目标的空间位置。基于"当前"统计模型,将获取到空间位置作为观测信息,采用自适应卡尔曼滤波算法,对机动目标进行跟踪,并准确预测其位置、速度及加速度信息。为验证本方案,使用一个Pioneer 3-AT作为主动机器人,及一个AmigoBot机器人作为被跟踪目标进行实验。结果显示,提议的方案可行,其精度满足实际应用的要求。An integrated scheme is proposed for detecting and tracking maneuvering target. The robot uses a vision sensor to obtain the target orientation by object detection technique. The target localization can be determined by fusing the orientation information with the laser scanner distance data. The current Statistical Model is adopted to model the target motion state, and the localization data is used for the observation model. Based on these two models, the robot can track the moving target and predict the target position, speed, and acceleration using the adaptive Kalman filter algorithm. To validate the proposed scheme, a tracking target experiment has been implemented with a Pioneer 3-AT robot and 1 AmigoBot robot. The result clearly shows that the proposed scheme is valid and appropriate for applying to the actual usage.

关 键 词：“当前”统计模型 卡尔曼滤波 机动目标跟踪 

分 类 号：TP242.6[自动化与计算机技术—检测技术与自动化装置]