How machine learning will transform your satnav - evo Connectivity

Before we start, for those not in the know, here’s a brief explanation of what ‘machine learning’ is. In simple terms, it is the use of algorithms that can iteratively learn from data, in turn enabling a computer to make predictions and adapt to new situations. This avoids the need for the computer to be explicitly programmed for each situation.

We come across machine learning processes all the time. Amazon’s ‘you might also like’ is one example, as is the technology used in fraud protection. Now machine learning is also making an impact on the automotive world.

Google’s self-driving car is a near-perfect execution of the process. Explicitly programming a car with details on how to tackle every possible permutation in a combination of areas including mapping, traffic avoidance and car control would be a mammoth task. Instead Google provides the car with a huge amount of data, all captured from the roads around its headquarters. This data is then studied by the car, and the patterns and behaviours within it logged, allowing the car to ‘learn’ how to react to scenarios such as cyclists or pedestrians.

This is at the absolute cutting edge of what machine learning is striving towards, but to the average consumer, the biggest change in the short-term will likely be experienced with satellite navigation.

Inrix, an infotainment tech-nology company, recently rolled out a redesigned version of its ‘Traffic’ app for iOS and Android. The application works by learning your driving routines and preferences and then creating a personalised predicted list of potential destinations and routes.

On top of this, the application uses a vast network of cloud-connected cars to intelligently map out traffic patterns, the end result being very accurate predicted arrival times and much better traffic-jam avoidance.

This kind of technology is what you can expect to see being integrated more and more into tech-focused cars in the future. BMW’s ultimate vision is to have a car understand and learn your daily routine, then self-drive itself to wherever you need it to.

The current crop of high-tech vehicles feature a plethora of touch inputs, menus and controls for you to communicate with them. Machine learning is all about removing these often complex barriers and the need to ‘inform’ a computer of what you want it to do. With time, machine learning, combined with accurate voice recognition, should transform in-car infotainment and navigation to something so slick and simple you barely need to think about it.