Informatics in Autonomous driving

Autonomous cars are obviously vehicles. But they are also networked computers on wheels. The advantages are many. Consider for instance:

Regular updates

There is a huge potential associated with the ability to update cars on the fly. Security patches, new functionality, enhanced experience will all follow the route of software development as we know it from the web.

 

But please: Let the cars update themselves. A “Java Solution” will be unbearable: In order to turn left, please enter and exit the car and honk the horn 15 times when prompted …

 

The ability to update software will also pave the road for 3rd party applications serving cars and car owners.

 

Accessing previous experience

Because autonomous cars are networked they can share “experiences” and learn from each other. Try to visualize a map. For each trip, an autonomous car draws a thin trail of data behind it. And as more cars travel the same distance, the trail becomes thicker and thicker.

 

This will lead to an abundance of information for continuous use by the car as it traverses the landscape. In only 3 years, a car will be able to process 10.000 data packages of “experience” from previous trips per running meter at 80 km/h! In 4 years much more.

 

Infrastructure to vehicle communication

Continuous data transfer also has other advantages. A context where the road continuously signals the car is now within reach. That scenario calls for a dramatic rethinking of speed limits. We will no longer be bound by fixed limits that apply at all times.

 

Consider for instance roads in close proximity to schools. Adaptive speed limits may signal the cars to drive very slowly in peak hours, and much faster at other times, e.g., holidays. The same principle applies to construction work, temporary solutions such as town festivals and much more.

 

Protecting the Data

I urge legislators to take the data aspect of autonomous driving seriously. Please thoroughly think through the implications of mobile data!

 

In order for these prospects to be realized, big data algorithms require us to share data. Certainly it becomes critical to consider: who has access to data generated by autonomous cars? Operators, car manufacturers, end users, insurance companies, law enforcement, and road maintenance crews will all have different claims to the data.

 

We need a strong legislation that these matters into account. And we need it now.

 

/Henrik Schärfe