A navigation method based on the partially observable markov decision process （POMDP） for smart wheelchairs in uncertain environments is presented in this paper. The design key factors for the navigation system of a smart wheelchair are discussed. A kinematics model of the smart wheelchair is given, and the model and principle of POMDP are introduced. In order to respond in uncertain local environments, a novel navigation methodology based on POMDP using the sensors perception and the user＇s joystick input is presented. The state space, the action set, the observations and the sensor fusion of the navigation method are given in detail, and the optimal policy of the POMDP model is proposed. Experimental results demonstrate the feasibility of this navigation method. Analysis is also conducted to investigate performance evaluation, advantages of the approach and potential generalization of this paper.
研究了音频信息处理中一项重要的预处理工作:语音音乐分类。针对语音信号处理中遇到的实际问题,选择合适的音频特征和分类器来对音频数据进行语音和音乐分类。采用二级系统,选择优化低能量率(ModifiedLow Energy Ratio,MLER)以及梅尔频谱倒谱系数(Mel Frequency Cepstral Coefficients,MFCC)作为音频特征,通过贝叶斯分类和混合高斯分类器进行分类。最后,使用上下文分类器对分类结果进行修正。实验结果表明,这种分类方法准确率和速度都较好。
