Google Bike: using Google Earth as as a VR trainer

There's those among us that will just go out and buy a Tacx iMagic trainer if they want to cycle around in a virtual world. But what if you could knock something similar up for next to nothing, using open source software? That's exactly what one clever geek has done: say hello to Google Bike!

Okay, the first thing you might notice is that this system isn't quite as neat a solution as a shop-bought virtual reality trainer, but there's some clever tinkering going on underneath the mess of wires. Basically the Google Bike uses a standard wheel sensor from a cycle computer and a joystick hacked from a PS2 controller to send speed and direction information to a computer, which uses the data to navigate around Google Earth. The full build is available on Instructables; most of the kit is cheap or hacked parts and it uses an interface called Arduino to do the talking between the bike and the computer.

Okay, maybe not that exciting at the moment, except if you're into your home electronics, but we reckon there's legs in this. If the Instructables bike is proof of concept, then there are vids on YouTube that show that even now you can make a much more refined version of the system. The vid below is a student/teacher project from Tokyo University that uses a steering sensor instead of a joystick and also seems to make use of terrain data to affect how the bike behaves on screen; you can see it jumping off lips for a bit of virtual airtime.

So what's the next step? Well, Google Earth is only going to get better. Already there's plenty of 3D environments in there and the latest version, 6, incorporates all of the streetview data that you can also access from Google Maps. Since the Streetview car made a point of driving the whole Tour de France route last year that means there's already plenty of landmark climbs covered.

Switching between views in streetview will gradually get smoother, position sensing will improve so you don't accidentally ride off the Tourmalet and up the Pic du Midi, and suddenly you've got a fairly robust VR platform. All you need is some simple two-way controls, the bike feeding the computer with speed and steering data and the computer returning gradient information to a computer-controlled trainer so that it can mimic the difficulty of the climb. You could even skip that last bit and have on-screen gradient instructions for you to apply to your trainer. Who's up for sorting this out then?