摘要
针对移动机器人上的声纳传感器感知周围环境具有很大的不确定性和不精确性问题,提出DSmT融合机技术(包括信息过滤器、融合算子和冲突分配器),在建立声纳信度赋值模型的基础上,首先利用基于证据支持贴近度函数的信息过滤器过滤掉部分不一致信息,然后利用经典DSm组合规则和PCR5冲突分配规则,融合16个声纳的不确定感知信息,使移动机器人能够准确地感知非结构环境.最后通过让机器人在一个相对拥挤具有家居特征的实验室环境中运行,得到了仅次于激光传感器感知环境的效果,充分地表明了该方法的有效性.
There being sonar＇s uncertainty and imprecision in sensing environment. Thus, Dezert-Smarandache theory （DSmT）-based fusion machine including information filter, fusion operator, and conflict redistributor is proposed to construct map for mobile robot. On the basis of establishing sonar＇s model of generalized basic belief assignment, at first, the information filter based on evidence support measure of similarity function is applied to filter out that inconsistent information. And then, the uncertain information from 16 sonar sensors is fused by applying classical DSm rule and proportional conflict redistribution rule No. 5 （PCR 5）. Finally, when pioneer 3DX mobile robot is controlled to run in a crowded experimental environment with domestic habitation features, a 2D map with sonar sensors inferior to that with laser range finder is built by new technology. Therefore, this experiment shows that new technology is very valid in map building adequately.
出处
《华中科技大学学报（自然科学版）》
EI
CAS
CSCD
北大核心
2009年第12期64-67,共4页
Journal of Huazhong University of Science and Technology(Natural Science Edition)
基金
国家自然科学基金资助项目(60804063)
