Paper Info
Title
Spatiotemporal Life-Log Mining of Wheelchair Users' Driving for Visualizing Accessibility of Roads
Abstract
This paper introduces a novel system for computational estimation and visualization of a possibility of accidents and incidents using wheelchair users' driving life-logs with three-axis accelerometers mounted on smart devices such as smartphones. Three wheelchair users participated in outdoor driving experiments held in urban center of Tokyo and all their actions were recorded in time-series of acceleration values and filmed on a digital video camera. In total, four units of iPods, which attached on left and right wheels, seats, and user's body, recorded approximately 1,800,000 accelerations signals during 10,000 seconds. In this paper, we analyze and classify life-logs of three wheelchair users' driving as the first step of computational estimation. We employed Support Vector Machine for classification, and created the supervised data from the video by human judgments. The life-logs were classified into moving/resting state and rough/flat state of the ground surface with finding optimal window size from 0.5 sec to 10 sec. As the result of classifications, estimation of moving/resting was achieved 99.8% accuracy rate and estimation of rough/flat surface was achieved 88.3% accuracy rate. Also estimations of driving difficulty were visualized on Google Map, and were evaluated by comparing with actual states of roads about wheelchair driving difficulty.
Year
DOI
Venue
2013
10.1109/ICDMW.2013.98
Data Mining Workshops
Keywords
Field
DocType
resting state,digital video camera,ground surface,flat state,spatiotemporal life-log mining,flat surface,accuracy rate,outdoor driving experiment,visualizing accessibility,computational estimation,actual state,wheelchair user,wheelchair users,machine learning,data visualisation,time series,accelerometers,support vector machines,data mining
Wheelchair,Data mining,Computer vision,Data visualization,Smart device,Visualization,Computer science,Accelerometer,Support vector machine,Acceleration,Artificial intelligence,Log mining
Conference
ISSN
ISBN
Citations 
2375-9232
978-1-4799-3143-9
1
PageRank 
References 
Authors
0.48
11
2
Name
Order
Citations
PageRank
Yusuke Iwasawa12610.78
Ikuko Eguchi Yairi26714.16