We almost take it for granted now that our locations are an easy thing to solve for. In our cars and on our phones, it seems that a location is ubiquitous. In fact, you have to be careful if you don’t want every photo you take and every Facebook post you type to broadcast your whereabouts to everyone in your network. However, with wayfinding, the challenge is accuracy. To be useful in our hospital case, to take one example, you need to narrow it down confidently to a room or a section of corridor. In my mind, this is the tricky part. GPS doesn’t work through walls, cell-tower triangulation isn’t accurate enough to be useful, and nav-grade inertial is too heavy and expensive. Everything else we have requires extensive infrastructure, which is costly both in hardware and in labour. The game here becomes to reduce the cost by finding the cheapest thing to roll out or, even better, taking advantage of something already there.
Cheap is RFID. These are the security cards and transponders (tags) that get you into your office or onto the toll highway without video charges. An RFID reader sends a radio query and gets a unique, very low power response from the tag. A tag that responds to multiple, networked readers should allow GPS-like trilateration (note: many use the term triangulation, which is a misconception). This is already being done where personnel safety around big machines is a concern, such as what Caterpillar is doing with the Detect Personnel system in underground mines. Of course, networked readers are expensive and the resulting position is in the network and not on the person, which now introduces privacy concerns and additional layers of complexity. Flip the equation around though, put the reader in the individual’s hands and the tags on the wall and it becomes more interesting. Then realize that the NFC function in your phone is really just a 2-way RFID tag/reader and you’ve got most of a solution already (great RFID/NFC comparison here). Walk into a building, an app on your phone updates its database containing a map and the position of all the tags and you now have a fully functional indoor mapping system. Brilliant right? Of course, reality isn’t that simple. Apart from the deliberate range limitations of NFC, you still have the problem of cheaply installing and surveying in the numerous tags needed, an effort which is keeping this kind of technology from getting the kind of mass-market acceptance it needs to be useful. Maybe someday construction companies will put RFID tags on every major component, as some are already experimenting with for automated progress monitoring and inventory tracking, but that’s not going to really help us any time soon.
What we do have currently are some interesting interim products. Apple’s iBeacon uses Bluetooth Low-Energy to alert an Apple device of it’s proximity to a specific location where the iBeacon transmitter is installed. This is mostly being used in retail applications where the store owner may want to target messages (yes, ads) to the shopper when they’ve approached a spot of interest in the store. There’s some push to use them as RFID is currently being used, to track assets through a warehouse, or to act as a security badge and track people’s whereabouts. The maker community is doing interesting things with home automation, turning on lights and unlocking doors, that sort of thing.
Wi-fi positioning is another way to go, utilizing the multiple overlapping presence of wi-fi routers around us at all times. Google infamously made headlines when it was revealed that its attempt to map wi-fi networks at the street level included archiving content being streamed over unprotected networks including emails and other sensitive information. As an indoor positioning system, this is not quite the solution we’re looking for, though. Wi-fi signals, like cellphone signals, tend to bounce around a lot making the distance from router to user difficult to narrow down in complicated spaces. It currently works well enough to tell you what building you are in but isn’t likely going to ever get you down to the accuracy you need.
Of course, the eventual solution will be a combination of all the above, plus inertial navigation using the gyros and accels in your phone, maybe even utilizing a vision system and SLAM algorithm using the camera or an RGBD camera like the Google Tango. We’ll get there eventually, it’s just a matter of time.
Stay tuned for the real fun, indoor mapping. Positioning is only part of the problem.