An optimized feature selection using bio-geography optimization technique for human walking activities recognition

A bipedal walking robot is a kind of humanoid robot. It mimics human behavior and is devised to perform human-specific tasks. Currently, humanoid robots are not capable to walk properly like human beings. In this paper, a technique to identify different human walking activities using a human gait pa...

Full description

Saved in:
Bibliographic Details
Published in:Computing Vol. 103; no. 12; pp. 2893 - 2914
Main Authors: Semwal, Vijay Bhaskar, Lalwani, Praveen, Mishra, Manas Kumar, Bijalwan, Vishwanath, Chadha, Jasroop Singh
Format: Journal Article
Language:English
Published: Vienna Springer Vienna 01.12.2021
Springer Nature B.V
Subjects:
Activity recognition
Algorithms
Artificial Intelligence
Classification
Computer Appl. in Administrative Data Processing
Computer Communication Networks
Computer Science
Decision trees
Feature recognition
Gait
Gait recognition
Geography
Heels
Human performance
Humanoid
Information Systems Applications (incl.Internet)
Locomotion
Machine learning
Optimization
Optimization techniques
Regular Paper
Robots
Software Engineering
Support vector machines
Walking
Bio-geography Optimization
Wearable sensor
Ensemble learning
68T40
Human gait activity recognition
Gait analysis
ISSN:0010-485X, 1436-5057
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:A bipedal walking robot is a kind of humanoid robot. It mimics human behavior and is devised to perform human-specific tasks. Currently, humanoid robots are not capable to walk properly like human beings. In this paper, a technique to identify different human walking activities using a human gait pattern is suggested. Human locomotion is a manifestation of a change in the joint angle of the hip, knee, and ankle. To achieve the aforementioned objective, firstly, 25 different subject’s data is collected for identification of seven different walking activities, namely, natural walk, walking on toes, walking on heels, walking upstairs, walking downstairs, sit-ups, and jogging. Next, the important features for gait activity recognition are selected using bio-geography based optimization, in which, classification accuracy is considered as a fitness function. Finally, we have explored six machine learning algorithms for the classification of gait activities, namely, support vector machine (SVM), K-nearest neighbor (KNN), random forest (RF), decision tree (DT), gradient boosting (GB), and extra tree classifier (ET). All these algorithms have been tested rigorously and achieve high accuracy of 91.64% in RF, 90.41% in SVM, 82.6% in KNN, 86.51% in DT, 88.34% in ET & 89.97% in GB respectively on our HAG dataset. The proposed technique is also validated on the WISDM data-set for comparative analysis.
Bibliography:ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 14
ISSN:0010-485X
1436-5057
DOI:10.1007/s00607-021-01008-7