Accessible path routing for wheeled mobility is an important problem given the permanent and temporary obstacles in the built environment. Existing research works have focused on identifying several obstacles as well as facilities such as crosswalks with traffic signals using smartphone based sensing or crowd-sourcing and used those knowledge to generate accessible routes. While all those research works have identified the nature of the road or sidewalk surface (even/uneven) as a major accessibility concern, there is no in-depth empirical research regarding how the accessibility varies from surface to surface. In this work, we propose a hybrid system, called WheelShare which can clearly classify (in/)accessible surfaces.We have analyzed vibration (using an aide-propelled manual wheelchair) data from 11 different standard indoor and outdoor surfaces found in the built environment and classified them using WheelShare into five different categories across the accessibility spectrum with an accuracy of up to 95\%.Our objective knowledge of surfaces is then applied to generate an accessible route through the best possible surface depending on user and wheelchair requirements. WheelShare is 1) scalable, as it uses crowd-sensing to collect data, 2) dynamic, as the data gets constantly updated, and 3) objective, as it uses an objective and data-centric approach.
%0 Conference Paper
%1 EdRaHoWaBeKr2019-TSUM-wheelShareConcept
%A Edinger, Janick
%A Raychoudhury, Vaskar
%A Hofmann, Alexandra
%A Wachner, Anton
%A Becker, Christian
%A Krupitzer, Christian
%B Proceedings of the IEEE International Conference on Pervasive Computing and Communications Workshops (PerCom Workshops)
%C Kyoto, Japan
%D 2019
%K descartes Optimization Self-aware-computing myown Internet_of_Things Self-adaptive-systems
%T WheelShare: Crowd-sensed Surface Classification for Accessible Routing
%X Accessible path routing for wheeled mobility is an important problem given the permanent and temporary obstacles in the built environment. Existing research works have focused on identifying several obstacles as well as facilities such as crosswalks with traffic signals using smartphone based sensing or crowd-sourcing and used those knowledge to generate accessible routes. While all those research works have identified the nature of the road or sidewalk surface (even/uneven) as a major accessibility concern, there is no in-depth empirical research regarding how the accessibility varies from surface to surface. In this work, we propose a hybrid system, called WheelShare which can clearly classify (in/)accessible surfaces.We have analyzed vibration (using an aide-propelled manual wheelchair) data from 11 different standard indoor and outdoor surfaces found in the built environment and classified them using WheelShare into five different categories across the accessibility spectrum with an accuracy of up to 95\%.Our objective knowledge of surfaces is then applied to generate an accessible route through the best possible surface depending on user and wheelchair requirements. WheelShare is 1) scalable, as it uses crowd-sensing to collect data, 2) dynamic, as the data gets constantly updated, and 3) objective, as it uses an objective and data-centric approach.
@inproceedings{EdRaHoWaBeKr2019-TSUM-wheelShareConcept,
abstract = {Accessible path routing for wheeled mobility is an important problem given the permanent and temporary obstacles in the built environment. Existing research works have focused on identifying several obstacles as well as facilities such as crosswalks with traffic signals using smartphone based sensing or crowd-sourcing and used those knowledge to generate accessible routes. While all those research works have identified the nature of the road or sidewalk surface (even/uneven) as a major accessibility concern, there is no in-depth empirical research regarding how the accessibility varies from surface to surface. In this work, we propose a hybrid system, called WheelShare which can clearly classify (in/)accessible surfaces.We have analyzed vibration (using an aide-propelled manual wheelchair) data from 11 different standard indoor and outdoor surfaces found in the built environment and classified them using WheelShare into five different categories across the accessibility spectrum with an accuracy of up to 95\%.Our objective knowledge of surfaces is then applied to generate an accessible route through the best possible surface depending on user and wheelchair requirements. WheelShare is 1) scalable, as it uses crowd-sensing to collect data, 2) dynamic, as the data gets constantly updated, and 3) objective, as it uses an objective and data-centric approach.},
added-at = {2020-04-05T23:20:22.000+0200},
address = {Kyoto, Japan},
author = {Edinger, Janick and Raychoudhury, Vaskar and Hofmann, Alexandra and Wachner, Anton and Becker, Christian and Krupitzer, Christian},
biburl = {https://www.bibsonomy.org/bibtex/254c1e9fea6bf3b9280b83acc0c027d96/chris.krupitzer},
booktitle = {Proceedings of the IEEE International Conference on Pervasive Computing and Communications Workshops (PerCom Workshops)},
interhash = {f5e82f5026108eb536ef498c7b9ba0a5},
intrahash = {54c1e9fea6bf3b9280b83acc0c027d96},
keywords = {descartes Optimization Self-aware-computing myown Internet_of_Things Self-adaptive-systems},
timestamp = {2020-05-20T13:27:19.000+0200},
title = {{WheelShare: Crowd-sensed Surface Classification for Accessible Routing}},
year = 2019
}