Quality Assessment and Accessibility Mapping in an Image-Based Geocrowdsourcing Testbed

Abstract

Geocrowdsourcing is a significant new focus area in mapping for people with disabilities. It utilizes public data contributions that are difficult to capture with traditional mapping workflows. Along with the benefits of geocrowdsourcing are critical drawbacks, including reliability and accuracy. A geocrowdsourcing testbed has been designed to explore the dynamics of geocrowdsourcing and quality assessment and produce temporally relevant navigation obstacle data. These reports are then used for route planning, obstacle avoidance, and spatial awareness. Recently, the geocrowdsourcing testbed has been modified to focus on the contribution of images and short descriptions, rather than the more lengthy previous reporting process. The quality assessment workflow of the geocrowdsourcing testbed is contrasted with a modified quality assessment workflow, implemented in the simpler and quicker image-based reporting paradigm. General quality assessment of data position and temporal characteristics is still possible, while general data attributes and detail are now supplied by a moderator from the contributed image. The derivation of obstacle location from multiple intersected image direction vectors does not produce reliable results, but an approach using buffered convex hulls works dependably. This simpler, quicker geocrowdsourcing workflow produces geocrowdsourced obstacle data and quality assessment estimates for location, time, and attribute accuracy.