项目名称: 基于机器学习的室外未知环境中移动机器人定位研究
项目编号: No.61473220
项目类型: 面上项目
立项/批准年度: 2015
项目学科: 自动化技术、计算机技术
项目作者: 王晓春
作者单位: 西安交通大学
项目金额: 80万元
中文摘要: 导航能力的高低是移动机器人智能化水平的重要体现,而定位问题则是移动机器人自主导航最重要的内容。在本项目中,我们将深入学习确定性环境条件下关于定位问题已取得的大量成果,在此基础上,研究如何使用单目视觉传感器,在应用环境中自然路标的坐标、形状等特征未知的前提下,通过对工作环境中的自然景物进行识别以完成定位。针对视觉导航中图像处理计算量大、实时性差这一瓶颈问题,及室外等非结构化环境的复杂性和多变性,研究出一种兼顾精度与速度的物体识别实时算法;针对单一路标抗干扰性和鲁棒性差的特点,提出一种基于自然景物组合作为特征路标的绝对自定位方法,以满足复杂非结构化应用环境下的自主式移动机器人应用的需要。
中文关键词: 视觉特征提取;聚类分析;图像分割;物体识别;机器学习
英文摘要: Navigation is an important ability of mobile robots. Localization in an environment is the very first step to achieve it. In this project, based on the extensive research already conducted for known indoor enviroments, we are going to design a natural landmark-based localization strategy for mobile robot working in an outdoor unknown environment. Particularly, we are going to design a real-time scene recognization scheme so as to use objects segmented in it as the natural landmarks and to explore the suitability of configural representation for automatic scene recognition in robot navigation by conducting experiments designed to infer semantic prediction of a scene from different configurations of its stimuli using a machine learning paradigm named reinforcement learning.
英文关键词: Visual feature extraction;Clustering;Image segmentation;Object recognition;Machine learning