Project • Research & development

NavIndo

Vision

Make the indoor world navigable.


Tags
  • crowdsourcing
  • indoor mapping
  • indoor localization

INTRODUCTION

Indoor mapping is a growing need in our current urban environment. Indoor mapping is the most fundamental need in robotics, augmented reality, virtual reality, mobile AdHoc networks and Internet of Things. Even though we spend 80% of our time indoors and accessibility is listed among the human rights, there are no location-based services available indoors (e.g., routing, localization, location sharing, etc.). The main reasons are: first, because GPS cannot operate indoors since its signal cannot penetrate through solid objects, such as walls, and second, there is no availability of indoor maps.

Mapping the indoor environments is a resource demanding procedure with prices that exceed 10.000€ for mapping a building not larger than 6.000 m^2. However, even with the current prices, existing technologies are not scalable enough to cover the demand of indoor mapping since most of the existing approaches require very expensive equipment.

On the other side, outdoor mapping has been a great success because of the open source communities which voluntarily help to map places with the guidance of airborne and satellite images. Even though the open source community has come with solutions for voluntarily indoor mapping, those solutions are not widely applicable and they are largely incomplete, since they are not rich in semantics and they cannot enable localization.

Our Proposal

Our proposal can be summarized in two actions. The first action is to enable people to provide data streamed by their smartphones to the IOTA marketplace (i.e. mine landmarks). The second is to provide algorithms that will enable the dynamic generation of highly detailed indoor maps from the efficient use of these data. This will motivate people to purchase such data from the IOTA marketplace and hence other people to mine such data. We expect to reduce the cost of the generation of indoor maps through this approach by one order of magnitude, as well as the cost of maintaining them in a depth of a decade by two orders of magnitude. In this way, buildings with a high need of mapping and low resources (e.g., hospitals or public transportation providers) will be enabled to obtain such services, while empowering the IOTA community.

Background Information

We have conceived and developed tools that can enable the generation of detailed maps. These maps can be generated through crowdsourcing via a gamified approach, where users can collect tokens inside real environments as presented in Figure 1. Those tokens are of a certain value which is described by their color (silver has low value, while gold has a higher value). The value of each coin/landmark is assigned as a function of the maximum required visits per landmark and the aging of every collected landmark. This implies that less mapped places have a higher value while more frequently mapped places have a lower value, unless it is a place hasn't been recently visited. The user can start mapping the environment from known locations that have already been mapped or in case of yet unmapped buildings the user starts mapping from outdoors, where GPS localization is possible.

This data is later fused through our novel approach that combines state-of-the-art methods for data analysis, for generating a point cloud of the indoor place. Aggregated data is then used to reconstruct the 3-Dimensional geometry of those places. Those generated models can enable routing from an unknown location to location and provide localization during the entire path. Finally, people who use the localization approach can be anonymously monitored and useful analytics can be extracted from their data.

Our approach can operate in most recent phones and requires users to walk inside buildings and stream their sensor data to our server. The data anonymity is completely protected since no information about the user is collected, while the user privacy is completely ensured through geofences that forbid streaming data when users are not inside the desirable locations.