Google Maps, Waze, and Yelp have transformed the way people travel and access information about the physical world. While these systems contain terabytes of data about roads and points of interest (POIs), their information about physical accessibility is commensurately poor. New websites such as Axsmap.com, Wheelmap.org, and AccessTogether.org aim to address this problem by collecting location-based accessibility information provided by volunteers (i.e., crowdsourcing). While these efforts are important and commendable, their value propositions are intrinsically tied to the amount and quality of provided data. In a survey, for example, Ding et al. found that most accessibility-oriented mapping sites suffered from serious data sparseness issues (e.g., only 1.6% of the POIs in Wheelmap had data entered on accessibility). One key problem is that these sites rely on local populations with physical experience of a place to report issues, which dramatically limits who can supply data.
In contrast, our research team at the University of Maryland and the University of Washington is exploring a different approach embodied in a new tool called Project Sidewalk (http://projectsidewalk.io). In Project Sidewalk, volunteer users *virtually* walk through cities to label and assess sidewalk accessibility in Google Street View--a bit like a first-person video game. So, rather than pulling solely from a local population, our potential pool of users scales to anyone with an Internet connection and a web browser. As Project Sidewalk is designed to engage and train non-expert volunteers in a complex task—that is, accessibility audits—our work is strongly linked to work in citizen science, like Zooniverse.org, which similarly seeks to engage non-experts in collecting and processing data through novel interactive tools. We are also actively working on incorporating game design principles to help engage, train, and sustain users.
Since our beta launch in Washington DC in Fall 2016, over 690 users have contributed 82,000 accessibility labels across 558 miles of DC streets--more than the distance from DC to Detroit! You can see some preliminary results here: http://projectsidewalk.io/results. The collected data is shared with city governments and used to develop new accessibility-friendly mapping tools (e.g., route planners, map visualizations), to train machine learning algorithms to automatically assess accessibility, and to create better transparency about accessible infrastructure (imagine a walkscore.com for sidewalk accessibility!).
But we are not done! Our goal is 100% DC coverage and then to expand into five more cities by the end of 2017. To get there, we need your help. Please visit http://projectsidewalk.io and complete a few missions--it only takes ~10 minutes to start contributing valuable data. If you like our project, please also help us get the word out via blogs and social media. As a start, you could tweet out or email this article to potential friends or family members who care about accessible cities (as we do!). We are also happy to receive comments on how to improve our tool. Feel free to leave them below.
Our overarching research vision is to transform the way in which city accessibility information is collected and visualized. Let’s make the world more accessible together!
- The Project Sidewalk Team
Ding, C., Wald, M. and Wills, G. 2014. A Survey of Open Accessibility Data. Proc. of W4A (NY, USA, 2014), 37:1--37:4.