Evaluating Crowd Sourced Navigation for the Visually Impaired in a Virtual Environment

Crowdsourcing has been shown to be a powerful method for solving a variety of problems. In this paper, we introduce an approach for allowing a crowd to help navigate a visually impaired user to their destination in real-time. Furthermore, we experiment with several approaches in aggregating and feeding back crowd data to determine the optimal method. Our approach streams live video from the visually impaired user's mobile device to a crowd of sighted volunteers. Each crowd member is able to provide their opinion on how the user should proceed and our algorithm aggregates this into a single opinion that is sent as feedback to the user. In this paper, we first present the design and implementation of our crowd sourced navigation system, including webapp design and two aggregation algorithms: averaging and league leader approaches. A virtualized user (avatar) is also developed for more controlled and repeatable testing of aggregation approaches with multiple crowd sourced volunteers. We also tested two navigation modes: a real-user controlled avatar, and a program controlled avatar. Experimental results are provided with the two aggregation methods and the two navigation modes. Our results show that we do not have significant difference between the aggregation methods and the program controlled avatar performed better than a real-user controlled avatar.