连续信号模式分类中的范式转变:面向两轮移动机器人的移动骑行辅助系统
机器人学
2015-06-17 v1
摘要
在本研究中,我们描述了一种可部署于普及的智能手机和平板电脑上的骑行辅助应用的开发。该骑行辅助应用具有信号处理和模式分类模块,可对实时信号模式分类产生近乎 100% 的识别准确率。我们引入了一种构建具有极强判别能力的训练字典的新框架,从而消除了在训练样本上人工干预识别模式的需要。我们通过提供另一项研究的结果来验证所提方法的识别准确率,该研究中追踪并识别了用于操控机器人轮椅的手部姿态与手势。
引用
@article{arxiv.1506.04810,
title = {Paradigm Shift in Continuous Signal Pattern Classification: Mobile Ride Assistance System for two-wheeled Mobility Robots},
author = {Ali Boyali and Naohisa Hashimoto and Osamu Matsumoto},
journal= {arXiv preprint arXiv:1506.04810},
year = {2015}
}
备注
This paper introduce a training approach for continuous signal pattern classification and its application to braking state classification of a mobility robots. In our previous journal article, we didn't employ a training method thus this paper is an improvement of a previously published journal article