For decades cars have had ever more complex engine and other management systems fitted to them to help the driver by automatically completing tasks or setting the car up in a micro second to change the dynamics of the vehicle. I wrote about Lamborghini’s system on the Huracan Performante that uses processors and sensors to redirect the air around the car and to adjust the suspension according to the road conditions. In my technical articles that give a brief description of the car’s components, I often talk about how chips and sensors have been part of the evolution of a vehicle.
Now we are getting in to autonomous vehicle territory where the vehicle is crammed full of sensors, radar, LIDAR, cameras etc that are capturing data and the software running the core systems has been developed to deal with a known condition.
We are still in the early days of this and have seen high profile accidents with Tesla’s running in Auto Pilot mode hitting stationary vehicles – there was one recently where a Tesla smacked into the back of a fire truck attending an earlier accident.
What we are seeing now is the inter-breeding of the systems on the current evolutionary tree of the automobile. Initially, we saw the autonomous systems connected to the engine and suspension management systems such that the car can speed up or brake under certain conditions and the suspension soften or harden depending on where the car is.
The next iteration of these systems is where true artificial intelligence (AI) happens. Today, the software has been written to cope with the known knowns. In other words, the system knows that a particular shape is likely to be human, it knows from the GPS where on the map it is and it knows certain data points such as speed limit, weather conditions etc.
We are getting into the area of having the systems learn from that basic knowledge – to deal with the known unknowns of the transport systems across the globe. These unknowns could be related to the vehicle itself as it records where it has been and what the conditions are, thus turning them into future known knowns!
Researchers are now applying artificial intelligence algorithms (and coding) to what a car “sees” and does, with the data collected being stored alongside the core data in the systems that is initially uploaded into the vehicle. With every trip – and every mile/kilometre travelled – the car will be able to record many metrics about the journey. When the car next travels down a particular road, it will use the stored data to decide what to do and will fill up any gaps with new data.
We will soon see cars with very large flash drive “memory” that will hold this information and in theory could share that through a telecoms connection to other vehicles in the same fleet. Today there are companies that already share basic usage information with insurance companies this way – speed, engine information etc so that the driver is billed insurance premiums based on the usage and the risk assessed by the company.
As Donald Rumsfeld, the Secretary of Defence under George W Bush once said of the unknown unknowns, “…they are the ones we don’t know we don’t know. And if one looks throughout the history of our country and other free countries, it is the latter category that tend to be the difficult ones”. This will be the true evolution of AI as it tries to convert them into known knowns or known unknowns! If the AI is good enough, it should take all the data collected from vehicles and ask itself the important question: have I dealt with this situation before?
Could we see AI evolve faster in automobiles than in any other machine?