有限状态自动机辅助的行人导航状态匹配算法

发布时间：2018-05-05 02:36

本文选题：行人导航 + 有限状态自动机　；参考：《测绘学报》2017年03期

【摘要】：行人导航状态的自动识别是行人导航研究的一个难点问题,对提升行人导航软件服务的精准反馈与改善导航性能至关重要,此方面已有的研究工作很少。本文提出了一种基于有限状态自动机的行人导航状态匹配算法,其核心思想是在识别行人动作基础上匹配行人当前导航状态。利用谷歌眼镜及智能手机采集的多种传感器数据对行人动作进行识别,得到其动作特征参数;然后将行人导航状态分为熟悉、陌生及迷路3类,根据有限状态自动机理论建立状态转移模型,设计基于该模型的行人导航状态匹配算法;最后,实现状态匹配算法,通过试验对该算法的有效性进行验证。试验结果表明,该算法能够较好地识别行人导航过程中的状态转移,其中对熟悉向陌生状态转移识别准确度较高,对迷路状态识别准确度达到90%。
[Abstract]:Automatic recognition of pedestrian navigation state is a difficult problem in pedestrian navigation research. It is very important to improve the accurate feedback of pedestrian navigation software service and improve navigation performance. In this paper, a state matching algorithm for pedestrian navigation based on finite state automata is proposed. Its core idea is to match the current navigation state of pedestrians on the basis of identifying pedestrian movements. Using the data of various sensors collected by Google Glass and smart phone to identify the pedestrian movement and obtain its motion characteristic parameters, the pedestrian navigation status is divided into three categories: familiar, unfamiliar and lost. According to the theory of finite state automata, a state transition model is established, and a state matching algorithm for pedestrian navigation based on this model is designed. Finally, the state matching algorithm is implemented, and the validity of the algorithm is verified by experiments. The experimental results show that the algorithm can recognize the state transition in pedestrian navigation process, and the accuracy of recognition of familiar to unfamiliar state is higher, and that of labyrinth is 90%.
【作者单位】：武汉大学测绘遥感信息工程国家重点实验室;四川省第二测绘地理信息工程院;
【基金】：国家自然科学基金面上项目(413714420)~~
【分类号】：P208;P209

【相似文献】