摘要
当前应用于机器人的多传感器数据融合研究大都集中在机器人自主导航和定位问题上,很少涉及利用机器人进行目标跟踪的研究。现有的基于网络时延的融合算法大都直接套用传统的同步数据融合方法,因此会产生数据等待、资源浪费以及较差的实时性能等问题。本文提出结合预测估计以及递推加权融合技术,设计出一种新的能适应网络时延的多传感器预测加权融合算法。该算法不仅能很好的解决现有基于时延的数据融合算法存在的诸多弊端,而且拥有良好的实时预测功能。文中给出了新算法的推导过程,并通过计算机仿真算例来显示新算法的实用性和优越性。
The existing researches of multisensor data fusion applied in robots focus mostly navigation and orientation, and a few researches about the object tracking using robots are reported. Because the existing fusion methods based on network-delay often use the traditional synchronous fusion algorithms directly, so some problems are be produced, such as information delay, resource free, and bad real-time performance etc.. Aiming at above problems, this paper combines the predictive estimate with the technology of sequential weighted fusion at the basis of the existing research, accordingly a new multisensor predictive weighted fusion method which can adapt to network-delay is proposed. The new method can not only avoid the disadvantages existing in the current data fusion method founded on network delay but also gain the better real-time predictive function. It presents the process to deduce the sequential predictive weighted method based on network-delay. Moreover, the computer simulation and theoretical analysis are used to show the practicability and advantage of the proposed fusion method.
出处
《微计算机信息》
北大核心
2006年第12Z期185-187,共3页
Control & Automation
基金
国家863计划先进制造与自动化技术领域机器人技术主题(2005AA420062
2002AA420110-5
关键词
多机器人系统
传感器
目标跟踪
线性无偏估计
预测加权融合
multi-robot system,sensor,object tracklng,linear unbiased estimate,predictive weighted fusion