Wavelet-Based Feature Extraction for Behavior Recognition in Mobile Robots

This paper introduces a dynamic data-driven method for behavior recognition in mobile robots. The core concept of the paper is built upon the principle of symbolic dynamic filtering (SDF) that is used to extract relevant information in complex dynamical systems. The objective here is to identify the robot behavior from time-series data of piezoelectric sensor signals from the pressure sensitive floor in a laboratory environment. A symbolic feature extraction method is presented by partitioning of two-dimensional wavelet images of sensor time-series data. The K-nearest neighbors (k-NN) algorithm is used to identify the patterns extracted by SDF. The proposed method is validated by experimentation on a networked robotics test bed to detect and identify the type and motion profile of mobile robots.

Download Full-text

A seamless economical feature extraction method using Landsat time series data

Earth Science Informatics ◽

10.1007/s12145-020-00564-4 ◽

2021 ◽

Author(s):

Chao Chen ◽

Liyan Wang ◽

Jianyu Chen ◽

Zhisong Liu ◽

Yang Liu ◽

...

Keyword(s):

Time Series ◽

Feature Extraction ◽

Extraction Method ◽

Time Series Data ◽

Series Data ◽

Feature Extraction Method

Download Full-text

Feature extraction from time series data

Journal of Computational Methods in Sciences and Engineering ◽

10.3233/jcm-2009-0239 ◽

2009 ◽

Vol 9 (s1) ◽

pp. S99-S110 ◽

Cited By ~ 2

Author(s):

Durga Toshniwal

Keyword(s):

Time Series ◽

Feature Extraction ◽

Time Series Data ◽

Series Data

Download Full-text

Automatic feature extraction and selection for classification of cyclical time series data

XXX. Messtechnisches Symposium ◽

10.1515/9783110494297-009 ◽

2016 ◽

pp. 61-68

Keyword(s):

Time Series ◽

Feature Extraction ◽

Time Series Data ◽

Series Data ◽

Automatic Feature Extraction ◽

Feature Extraction And Selection ◽

Cyclical Time ◽

Selection For

Download Full-text

Feature Extraction of Motion from Time-series Data by using Attractors

IEEJ Transactions on Electronics Information and Systems ◽

10.1541/ieejeiss.132.975 ◽

2012 ◽

Vol 132 (6) ◽

pp. 975-982

Author(s):

Takuma Akiduki ◽

Zhong Zhang ◽

Takashi Imamura ◽

Tetsuo Miyake

Keyword(s):

Time Series ◽

Feature Extraction ◽

Time Series Data ◽

Series Data

Download Full-text

Fusion Process Neural Networks Classifier Oriented Time Series

Journal of Computational and Theoretical Nanoscience ◽

10.1166/jctn.2019.6946 ◽

2019 ◽

Vol 16 (10) ◽

pp. 4059-4063

Author(s):

Ge Li ◽

Hu Jing ◽

Chen Guangsheng

Keyword(s):

Neural Network ◽

Neural Networks ◽

Time Series ◽

Feature Extraction ◽

Time Series Data ◽

Learning Algorithm ◽

Series Data ◽

Time Varying ◽

Synthetic Control ◽

Classification Feature

Based on the consideration of complementary advantages, different wavelet, fractal and statistical methods are integrated to complete the classification feature extraction of time series. Combined with the advantage of process neural networks that processing time-varying information, we propose a fusion classifier with process neural network oriented time series. Be taking advantage of the multi-fractal processing nonlinear feature of time series data classification, the strong adaptability of the wavelet technique for time series data and the effect of statistical features on the classification of time series data, we can achieve the classification feature extraction of time series. Additionally, using time-varying input characteristics of process neural networks, the pattern matching of timevarying input information and space-time aggregation operation is realized. The feature extraction of time series with the above three methods is fused to the distance calculation between time-varying inputs and cluster space in process neural networks. We provide the process neural network fusion to the learning algorithm and optimize the calculation process of the time series classifier. Finally, we report the performance of our classification method using Synthetic Control Charts data from the UCI dataset and illustrate the advantage and validity of the proposed method.

Download Full-text