Characterizing Roughness of Pedestrian Pathways Using Crowd Sourced Data and Support Vector Machine Analysis
Over two million people in the United States use a wheelchair for their primary means of mobility and rely on functional and accessible sidewalks to participate in their communities. The Americans with Disabilities Act Accessibility Guidelines related to surface roughness are subjective and not measurable which results in public pathways with many bumps and cracks which can lead to harmful whole-body vibrations for wheelchair users. Thresholds relating to surface roughness of pedestrian pathways have been proposed and new tools to measure pathway roughness are being developed. This paper evaluates the potential of using wheelchairs and a crowd sourcing approach to identify surfaces that comply or don’t comply with the proposed roughness thresholds. A Support Vector Machine learning analysis was conducted on previously collected vibration data that was obtained from wireless accelerometers attached to wheelchair frames as they were driven over various surfaces. The results show that using only average RMS acceleration data, 75% of the surfaces can be accurately grouped as compliant or non-compliant with the proposed roughness thresholds. The results show that it may be possible for wheelchair users to passively collect data with an accelerometer, such as the internal Inertial Measurement Units in smartphones, that will be able to predict whether the surfaces that they traveled over are compliant or non-compliant with the proposed roughness thresholds. This approach could be a quick and easy way for cities and municipalities to inventory their sidewalk systems in order to find which surfaces are in need of repair.
-
Supplemental Notes:
- This paper was sponsored by TRB committee ABE60 Accessible Transportation and Mobility. Alternate title: Characterizing Roughness of Pedestrian Pathways Using Crowdsourced Data and Support Vector Machine Analysis.
-
Corporate Authors:
500 Fifth Street, NW
Washington, DC United States 20001 -
Authors:
- Duvall, Jonathan
- Sinagra, Eric
- Cooper, Rory
- Pearlman, Jonathan
-
Conference:
- Transportation Research Board 94th Annual Meeting
- Location: Washington DC, United States
- Date: 2015-1-11 to 2015-1-15
- Date: 2015
Language
- English
Media Info
- Media Type: Digital/other
- Features: Figures; Photos; References; Tables;
- Pagination: 12p
- Monograph Title: TRB 94th Annual Meeting Compendium of Papers
Subject/Index Terms
- TRT Terms: Accelerometers; Crowdsourcing; Data collection; Measuring methods; Mobility; Pedestrian areas; Persons with disabilities; Roughness; Sidewalks; Vibration; Wheelchairs
- Identifier Terms: Americans with Disabilities Act
- Uncontrolled Terms: Support vector machines
- Subject Areas: Pedestrians and Bicyclists; Safety and Human Factors; Society; I72: Traffic and Transport Planning;
Filing Info
- Accession Number: 01556604
- Record Type: Publication
- Report/Paper Numbers: 15-5032
- Files: TRIS, TRB, ATRI
- Created Date: Mar 10 2015 7:50AM