NaviBlind Version 2.0 – Independent Navigation for Blind People

Last Updated: 09/04/2024 09:55     Created at: 09/04/2024 09:55

Final Presentation of NAVISP Project EL2-150 now available:

On Thursday, March 28th, 2024 NaviBlind ApS presented the results of the NAVISP EL2-150 project EL2-150 "NaviBlind 2.0 – Independent navigation for blind people". 

The project NaviBlind represents a leap forward in aiding the mobility of blind and visually impaired individuals by using a wearable GNSS device, NaVICap, and a related mobile app.  The wearable device, NAVICap, integrates an antenna and an external GNSS device to process signals from GNSS constellations, providing centimetre-level positioning accuracy to the mobile app. The NaviBlind app allows users to request, store, and navigate both private and public routes. Upon initiating navigation, users' positions are accurately determined, and they are guided to their destination in real-time. 

One important step was the development of a protocol for micro-taskers, referring to individuals engaged in executing precise, minor tasks on crowdsourcing platforms. This entailed furnishing them with explicit directives for labelling aerial images to identify pedestrian crossings, Audible Pedestrian Signals (APS), and sidewalks. Micro-taskers are essential to get this information quickly and efficiently. This protocol ensures consistent data collection for streamlined analysis. Here's how it works: Utilizing map segmentation, it generates a dataset containing geographic coordinates of sidewalks, crosswalks, and pertinent features extracted from aerial imagery. Employing machine learning algorithms, it detects these features in images, ensuring dataset reliability for navigation applications. 

The project trialled the protocol. Despite difficulties, especially in finding sidewalks due to their complex patterns, a semi-automated approach was used. This method combined open-access datasets with manual verification to produce a complete reference dataset covering large urban areas of reference, improving the accuracy and coverage of NaviBlind's navigation capabilities. In terms of route optimization and algorithm development the project successfully transitioned data from polygonal representations to mid lines streamlining route calculations, useful for the NaviBlind community assess risk efficiently. Led by Artificial Intelligence (AI) developers and guide designers, a route algorithm was devised to determine the shortest and least complex paths between points, enhancing route planning efficiency.  

The project has yielded satisfactory results, with the solution now accessible to users in Denmark. To enhance functionality and broaden market reach, additional research into route calculation algorithms, pedestrian map creation, and related tools will be essential for enriching navigation capabilities. Expansion strategies involve conducting hardware and software testing in other countries and collaborating with blind influencers to enhance product visibility and outreach. Furthermore, the exploration of automated vocal notifications for critical navigation is planned, with the aim of enhancing both user experience and safety. 

The project was part of NAVISP Element 2, which aims to maintain and improve the capabilities and competitiveness of the participating States' industry in the global market for satellite navigation and, more broadly, PNT technologies and services.

More detailed information can be found in the slides of the Final Presentation.