两轮自平衡机器人多传感器数据融合方法研究  

A Multisensory Data Fusion Method of the Two-Wheeled Self-Balanced Robot

作　　者：王晓宇[1] 闫继宏[1] 臧希喆[1] 秦勇[1] 赵杰[1] 

机构地区：[1]哈尔滨工业大学机器人研究所,哈尔滨150001

出　　处：《传感技术学报》2007年第3期668-672,共5页Chinese Journal of Sensors and Actuators

基　　金：长江学者和创新团队发展计划资助(IRT0423)

摘　　要：为了对机器人运行状态进行有效的识别,提出一种基于支持向量机的多传感器数据两级融合方法,从分类的角度实现了运行状态识别,解决了识别正确率较低的问题。将此方法应用于两轮自平衡机器人进行运行状态识别实验,当每种状态采集的独立样本数超过20个时,正确率可以达到98%以上.实验结果表明应用该方法可以对两轮自平衡机器人的运行状态进行有效、可靠的识别,能够满足两轮自平衡机器人快速机动过程中的实时性要求.To recognize the running state of the robot efficiently, a SVM-based multisensory two-graded data fusion method is presented. The running state recognition is realized from the view of classification. The problem that the classified accuracy is low is solved. The method is applied to the two-wheeled self-balanced robot and the experiments of the running state recognition are conducted. When the individual sample number of each running state exceeds twenty the accuracy of fusion method will be above 98%. Experimental results demonstrate that the running state of the two-wheeled self-balanced robot could be recognized efficiently and reliably. The real-time requirement will be suitable in the fast and maneuverable process.

关 键 词：支持向量机 两轮自平衡机器人 多传感器数据融合 两级融合 运行状态识别 

分 类 号：TP242[自动化与计算机技术—检测技术与自动化装置] TP212.9[自动化与计算机技术—控制科学与工程]