“Mapping the unmappable” – how we enabled positioning at the REDI shopping mall in less than 8 hours

Here in Finland the opening of a new shopping mall is always an event. Recently a new one called “REDI” was opened in Helsinki. There has been an unusual amount of publicity around this specific shopping mall – apparently people have great difficulty finding the way around the mall and even get lost there. Not surprising, then, that as we had our hackathon event a few weeks ago, one of the tasks was to enable the REDI shopping mall for positioning.

During our hackathons we normally divide ourselves into groups of two – although some tasks did get done by a single person. I joined the team of my colleague Ilari Vallivaara, and we chose the topic of mapping the REDI shopping mall, an idea of our CTO Otto Seiskari.

It is always good to eat one’s own dog food, giving a fresh customer perspective to one’s own products. In our case this was a very pleasant experience: roughly one year ago as a hackathon topic I participated the mapping of another shopping mall here in Helsinki, and the tools were somewhat rough around the edges at the time. Not so now – the project went almost too smoothly. The result was that we had a working indoor positioning solution in matter of about four hours! As there were two of us, the total time spent was approximately 8 hours.

The video demonstrates how you can simply in the application see where you are indoors. No more guesswork.

The process of enabling a space for indoor positioning with the IndoorAtlas technology involves a few simple steps. After completing these the space becomes positioning enabled, meaning that if one uses a smartphone app with our technology, the positioning works in the app. This process of enabling a space for positioning is called fingerprinting or mapping.

We followed our standard workflow: as prerequisites we first got the floorplan pictures, uploaded them to our cloud using our developer web tools, and then positioned the floorplan pictures properly on the world map. Since this was an internal hackathon project, we did this very quickly by just taking screenshots from the appropriately scaled indoor maps available at REDI’s public web site. Positioning of the floorplans in world coordinates was a little challenging: normally one can easily position the floorplans over satellite images available in Google maps or Mapbox maps (both available on our web tools). Since this was a new building, the satellite maps basically showed a construction zone, but we were able to use the locations of the nearby roads to fix the floor plans to world coordinates.

With the pre-requisites out of the way, it was time to go on-site to do the fingerprinting. This involves simply walking around the site in certain way, while running the IndoorAtlas MapCreator 2 app on the smartphone. As we were doing the fingerprinting we could experience first hand that the press was not wrong – a lot of people were asking directions and spending time looking at the maps. We even had an older gentleman approaching us and asking how he can escape the mall, apparently he had been stuck walking circles and could not find an exit. Not the kind of experience you want in any shopping mall!

Normally the first step in any installation is to make sure there are some BLE (Bluetooth Low Energy) beacons available, so that as the fingerprinting process is done, the recorded radio signal environment includes beacons too. (The radio environment is used to compute quickly what is known as the “first fix”, the first positioning estimate. After the first fix is computed, the heavy duty algorithms kick in and very accurate positioning estimates are calculated using the geomagnetic and other methods). As it turned out, the REDI shopping mall contained already enough beacons to facilitate good first fix performance!

For background information – our technology requires between 10% to 20% of the number BLE beacons compared to solely beacon based positioning solutions. In this case it meant that the positioning system could be installed completely un-intrusively, without installing any new infrastructure to facilitate positioning. Our guess is that these beacons were on-site for other reasons, such as supporting certain functions at the stores. We call these types of beacons “ambient beacons”, as they are there for other reasons than positioning, but our technology can use these for positioning purposes too.

Without the need to do a beacon installation, the fingerprinting became a simple matter of systematically walking around the shopping mall while using the MapCreator application. We completed the two main floors in a few hours, and the proceeded to do some further mapping to support positioning all the way from the exit of the subway to indoors. After doing the fingerprinting work, we did a bunch of testing with the built-in positioning features of the MapCreator app. The positioning was working really well! We also confirmed this later by looking at our analytics displays.

The remaining task now would be to integrate the positioning capability into an actual production app. In a place like REDI it seems that even one of the simple example applications we provide in our SDK would be sufficient to help customers find their way – and not get stuck and frustrated in walking circles. The technology has really matured and is here today.

Erik Piehl