UW-Madison Augmented Reality Tour

In the fall of 2014 I worked in the Internet of Things Lab here at the University of Wisconsin-Madison. I first saw the group at their open house in the spring of 2014, where I got to try out some really cool tech, like the Oculus Rift and Google Glass. I knew I wanted to work with some of that, so when fall semester rolled around I submitted my project idea: an application for Google Glass that uses the users GPS and compass readings to give information on the building the user is currently looking at. I had three awesome group members, Nick Sielicki, Amy Slawson, and Nate Lading, and here’s what we made:

The app works like this. The user wears Glass and enters our app, and the users GPS position is sent to our server. There, it is taken by a small python webserver written with web.py that connects with a PostgreSQL database with a PostGIS (Geographic Information Systems) extension (For more info, see my post on setting it up). Then, it does a spatial query to get the information we want, and here’s how it does it.

First, we want to get all the buildings within some radius of the user. Once we have these, we want to look at each corner of each building. For this collection of points we have for each building, we then compute the angle that each point makes with the user and North. Then from these angles (for a single building), we get the minimum and maximum angles, and those are the left-most and right-most headings for that given building. Now what we have is a list of buildings in the users view, along with the range of degree headings that each building takes up in the users field of view.

Note there is an exception we had to worry about with this method of getting headings: if a building straddles North. In this case, getting the point with a max heading will return something close to 359, and the min will be close to 1, regardless of where the actual left- and right-most points are. To fix this, we need to use a different reference point when computing the angles, so we compute the angle between the user, the building corners, and South.

So now that we have each building and their heading ranges, we send this back in a nice JSON formatted response. The Glass app itself then can read this to draw the compass view with each building represented as a colored rectangle. Then, when the user moves their head, the Glass gets the current compass heading from the geomagnetic field sensor and rotates the compass view on the display. If the user taps on the side of the Glass when they are looking at a building, an information card pops up.

While I focused on the back-end with setting up the server and database, and figuring out the queries we needed, I helped with the Glass app in a pair programming setting. Coding for Glass is almost the exact same as making an Android app, as Google tried to make Glass a familiar environment for those used to Android programming. We didn’t start from scratch with the app, but instead forked a Google example project, a compass. We added code to connect with our server and give GPS coordinates, and then parse the response and draw the buildings on the compass view.

Check out the server and app code on github.

Posted on 03 Feb 2015