For some reason this augmented reality video from a couple of years ago is making the rounds again on various blogs and even Gizmondo is reporting it as fresh news. Since its resurfaced, as a fellow car engineer for Toyota, I’ve decided to address some of the challenges to making a system like this really work.
1. AR Glasses
The first and obvious challenge is that the glasses aren’t yet a reality though they will be this year or next. While a mechanic doesn’t care about style, the glasses still have to function well. How fast of a refresh rate will the glasses need to not create “head lag” as the mechanic looks around in the engine. The controls for the unit will also matter, since they’ll need to stop, start and rewind.
Without a paper marker, the glasses will need to orient the graphics onto physical locations on the car. How well will they respond to dirty motors, poor lighting or the mechanic getting in the way of the camera?
3. Spatial Issues
When doing maintenance on parts that are easily accessible, augmented reality works well, but what about when the water pump under the wheel well needs replacement? How well will augmented reality project the proper maintenance steps deep into the car engine space?
4. Filling the Database
In the early nineties, our plant purchased a “wonderful” maintenance system that would house all of our PMs, keep track of all parts and what machines they went to and make sure we kept our machines in tip-top shape. The flaw in this wonderful idea was that someone had to actually put all this data into the system and maintain its integrity as we upgraded machines and moved equipment around. For high-frequency work, these AR maintenance glasses would be easy to develop, but who is going to make the other two hundred infrequent displays needed?
Overall, the promise of using augmented reality in industrial situations is quite large. Given an improvement in the technology, I could find dozens of applications in my own plant. Visualizing difficult tasks for team members using 3D data has huge potential, but it has to work right and be cheap enough to make it to the plant floor. While I think AR will make its way into the production of cars (in fact, I’m working with Metaio on one such application) many issues need to be solved to make it a widespread usage.
nice to get this discussion back into the focus, where everyone considers building AR glasses. 😉
But I don’t see this problem that big if you are talking about a controlled environment, let’s say in a repair shop, etc.: you can always just track the person’s glasses plus the car’s position and then calculate the appropriate overlay from outside-in. You don’t need markers or features attached to car parts.
But if you want to track through your HMD’s cam without setup time,…. I guess you are right and it’s still tough. 🙁
@ toby – I’m also thinking about enviroments more challenging than a clean maintenance shop. There are lots of applications to use this type of system on our plant floor, creating visual cues for difficult jobs, but there’s a lot of variables to keep track of (we also have a heat problem, since its a foundry.)