At Accuware we’ve been hard at work developing new products and new features for our existing products. Our goal is to increase indoor location accuracy and improve performance without additional effort and cost. What is there not to like?
Months ago we began working on video signal processing from mobile devices’ cameras and confirmed that there are substantial improvements we can make in the indoor location arena by harnessing these technologies. Think about it as adding yet another sensor that contributes to determining a device’s location.
Multiple approaches are possible. An interesting one was to extend the notion of a site survey to include camera images that complement the collected ambient WiFi and BLE signals.
How it works
Locating indoors using ambient signals is a well-known technique based on the notion that, in today’s modern environments, awash in WiFi and BLE signals, the mix of ambient signals in different parts of a building vary greatly.
This enables correlating a given set of signals and their strengths to a specific geographic location, such as “room 305 on the third floor, by the window”. And makes it possible for a mobile device to determine its location by taking a snapshot of the ambient signals that surround it, and then interpolating that data into a database of ambient signal snapshots taken across the entire venue. The snapshots database is built by the process known as a “site survey”.
What is a site survey?
The figure shows a person walking a predefined route from start (green dot) to finish (red dot), at times changing direction (blue dot), periodically collecting along the way lists of SSID (Service Set Identifiers) emitted by access points (AP) nearby.The collected data is uploaded to the server and stored in the site’s database.
Similarly, snapshots taken by the device’s camera can be collected along the route and added to the database, to be used for sensor fusion.
What is sensor fusion?
Sensor fusion is the technique by which measurements from multiple sensors are aggregated to reduce uncertainty and increase accuracy. This has proven very helpful to improve the performance of indoor location systems.
Typically, signals from sensors such as accelerometer and compass are “fused” with ambient signals to help smooth out jumpiness from successive location estimates. In our case, the “sensor” in question is the device’s camera, whose images are analyzed and interpolated into the venue’s database to estimate location. A clear advantage of using images with ambient signals is that it can help determine location in areas where ambient signals are weak or non-existent.
How does all this work together? It’s time to find out.
Show and Tell
We convinced Cyril Houri, Accuware’s CEO, to star in a demonstration video that shows how things really work. Here he is:
In the video Cyril performs the site survey or “fingerprinting” in the hallway outside of our Miami office. Note the following:
- The mobile app used is in Training mode, set to perform the site survey
- The screen displays the route to be used, which was created just before the video, indicated by lines and dots: Green (start), Blue (turn) and Red (end).
- In the video he walks the route to collect ambient signals and images data, all the way down the hallway
- The upper right-hand corner of the phone screen shows the current image from the camera (red circle and arrow)
A small mobile device, like the “tag” Cyril shows in the video, is capable of taking snapshots of ambient signals and camera images. The result is sub-meter location accuracy.
As Cyril teasingly says, locating by video is like “déjà vu location”, meaning that locations will be estimated based on images the system has previously seen, somehow closer to how human beings do it. Only much more geographically precise.
We are excited about these developments. Our Wearabouts product, intended for people and asset tracking just learned a new trick. And in the process, we increased its location accuracy down from 10 feet/3 meters to 3 feet/1 meter or better.
Yes, a well-planned site survey of the target venue is required. But no additional effort is needed to collect the images that will improve system performance. The enhanced “training” app does all the heavy lifting. And if your application does not need the new feature, you can simply use regular ambient-signal-based positioning without the camera.
Quick note about pros and cons of the new technology:
- You can get increased location accuracy with no additional effort
- Expanded system works where ambient signals are few or simply not present
- Your mobile device will use a bit more of its battery power
Ready to try the new, improved Accuware Wearabouts? Contact us.