论文部分内容阅读
提出一种基于距离传感器的结构化特征的动态、自组织提取方法.该方法由3个部分组成:主动感知行为的设计,时空信息的降维处理及路标的自组织提取.设计基于沿墙走的“主动感知行为”来获得高相关性的感知时空序列信息;给出基于变化检测和激活强度的活性神经元来对时空序列信息降维;最后提出一种二维动态增长自组织特征图方法,实现环境路标的自组织提取和识别.实验结果验证该方法的有效性.
This paper presents a dynamic and self-organizing extraction method based on the structural characteristics of distance sensors.The method consists of three parts: the design of active perception, the dimension reduction of space-time information and the self-organizing extraction of road signs.The design is based on the walk along the wall Based on the “active sensing behavior” of the “active sensing behavior” to obtain highly correlated perceptual space-time sequence information; given active detection neurons on the basis of change detection and activation intensity to reduce the dimensional space-time sequence information; finally proposed a two-dimensional dynamic growth self-organizing feature Map method to realize self-organization extraction and identification of environmental road signs.The experimental results verify the effectiveness of the method.