Microsoft's Predestination Makes Your Car Like Knight Rider (Video)


Photo via realSMILEY via Flickr CC

We've heard of Google tracking your movement in order to provide traffic flow data, and we've heard of algorithms that can make traffic flow more smoothly and efficientlyfor data centers. Now Microsoft is taking these concepts and going a step farther. The purpose is cool - predicting how people move in a city so that traffic flow can be improved, or even perfected - but still, Microsoft's Predestination experiment has a decidedly 1984 vibe. The system can take into account your driving habits, predict where you're headed, and give you better directions for how to get to your destination before you even realize that's where you're headed or that you need the directions. Cool, but creepy, right? Even if it's good for building smart cities?

Under development for at least the last four years, Microsoft's Predestination is an experiment to create a driving directions recommendation program in the car, based on the data collected from your driving habits and the habits of others, urban driving data, traffic patterns and other factors in getting from one place to another. The program can use all the information at its disposal to predict where you're headed and give you the best route at that moment for where it thinks you're heading, even if you yourself haven't decided your destination.

A research paper from Microsoft states, "Driving behaviors include types of destinations, driving efficiency, and trip times. Beyond considering previously visited destinations, Predestination leverages an open-world modeling methodology that considers the likelihood of users visiting previously unobserved locations based on trends in the data and on the background properties of locations. This allows our algorithm to smoothly transition between "out of the box" with no training data to more fully trained with increasing numbers of observations."

Fast Company reports, "Microsoft created its recommendation algorithm by tracking GPS data on 169 different subjects who drove 7,335 different trips. Microsoft's algorithm takes into account different types of destinations, driving efficiency, trip times, and "an open-world modeling methodology that considers the likelihood of users visiting previously unobserved locations based on trends in the data and on the background properties of locations." That means the algorithm can predict where you're going before you even decide to go there."

It's a little disconcerting in terms of feeling like you're an independent being, but the application itself is appealing - imagine using this to ease traffic congestion during rush hours or locations where accidents have occurred, or figuring out the best way to get 5 errands done, without actually having to input information and ask a computer program beforehand.

Even better, the algorithm can be used beyond road traffic, making information systems from data centers to building management more efficient. Here is Rick Rashid, the senior vice president of Microsoft Research, explaining Predestination.

We are more predictable than we think in our habits, so while there will probably be some pitfalls, Predestination seems like a system that could really work. And if it means efficiency across all types of systems, from roads to buildings, it could hold the potential to revolutionize how we get people or data from here to there.

Follow Jaymi on Twitter
More on Perfecting Driving
Save Money by Driving Better: BP Doing Something Right (Gasp!)
MyFord Touch and Eco-Route Nav Feature Will Help Drivers Save Gas
GPS Systems with Real-Time Traffic Data Can Save Drivers Four Days Per Year (and Lots of Carbon Emissions, too)

Tags: Computing | Driving | Energy Efficiency

Best of TreeHugger