Are we ready for driverless cars?
For some the thought of driverless cars is exciting and for others it’s scary.
Driverless car technology has been getting a lot of press recently and it’s easy to see why. When someone uses the phrase “driverless cars” or “self driving cars” it instantly conjures up thoughts of futuristic cars zooming around the public highways, taking care of all of the directions and boring safety stuff whilst you, the occupant, simply sit back and relax – read a book, catch up with work, watch TV?
Just as recently as 10 years ago this type of technology was reserved for the realms of science fiction, sure Google and a few other firms were exploring the technology but it seemed so far removed from the current capabilities that most people did not take it seriously.
Today it is a completely different story with Tesla, Mercedes Benz, BMW, VOLVO, Ford and Google all testing out driverless technology. Indeed, Tesla CEO Elon Musk has been quoted as stating that he believed the implementation of driverless cars could be as little as 2 years away and more strikingly that within 3 years the technological capabilities would exceed that of a human driver counterpart.
So, are we really that close to such a remarkable technological achievement? Are we ready for it?
The answer to both questions is kind of yes and no. Certainly recent years has seen a massive leap forward with the technology but as of today we are still a long way off such cars being common place on our roads. Why? Well to understand the implications and issues currently faced we have to look into how driverless cars work which is explained below:
LIDAR is used to bounce a laser beam of surfaces to create a 3D map of the cars surroundings to enable the car to “see” potential hazards.
Whilst LIDAR can accurately build up an image of the surrounding landscape it has one major flaw with the fact that it cannot accurately monitor the speed of surrounding vehicles in real time. This is where RADAR picks up the slack sending data to the internal computer to apply the brakes where required or manoeuvre to avoid a hazard.
GPS Positioning and High Powered Cameras:
The vehicle would also have to make use of high powered cameras which provide overlapping images to provide depth of field, peripheral vision etc. The vehicle would also have to utilise advanced mapping technology to effectively navigate its journey. The vehicle would have to know exactly where it was positioned down to within a few centimetres as well as precise driving speed.
The on board computer would have to be able to process an enormous amount of information in real time. The system would be based around complex learning algorithms which would allow the computer to find the appropriate action to take dependent on the scenario presented.
The problems faced are both practical and theoretical not to mention the legal and insurance ramifications.
From a technology standpoint a suitable mapping system would have to be developed. At present Google have mapped around 2,000 miles of road for it’s car to effectively work on. Clearly there is a long way to go before nationwide/worldwide maps have been created.
Furthermore the software currently has limitations on object recognition. For example it is currently unable to distinguish between a child in the road and a paper bag and in theory would react the same way by swerving to avoid the collision. For the system to operate effectively some sort of moral considerations would have to be factored in. The magnitude of potential scenarios is effectively infinite.
To further the question, the vehicle would have to calculate the right course of action when presented with a lose-lose situation. For example, when faced with running over a child or swerving but potentially hitting a bus stop of say 5 people would the system calculate that running over the child is the lesser of two evils? If not, what considerations make up to decision? Would it be different if it were faced with running over an adult? What age does the thought process change? What if it were faced with hitting a child or swerving off a bridge potentially killing the occupant? Would you really want that decision to be taken out of your hands?
There are many questions to be answered alongside the not so insignificant task of improving the technology and building up a suitable, usable, mapping system. Currently the vehicles also struggle with adverse weather conditions and sun glare.
The stages of driverless cars.
At present, driverless technology is divided into 5 categories. Stages 3 to 5 are generally thought of as the next stages of driverless technology.
Stage 3 will perform all usual driving activities without human intervention however would hand over to human control in certain situations meaning that alert human attention is still a requirement.
Stage 4 would be a step further meaning all driving situations would be handled autonomously. This is what is currently thought of driverless technology. It is important to note that even with this technology, not all driving scenarios would be covered. Finally stage 5 is considered fully autonomous meaning that every aspect of the journey is handled by the vehicle in a manner at least comparable to that of a human driver.
Stage 5 there would be no expectation for human intervention in any situation.
All stages still present significant hurdles before they become mainstream technology. For example, stage 3 may seem the simplest on paper but there are still issues surrounding the length of notification time given to the driver to deal with a potential hazard. Indeed, most large firms believe this is unrealistic and are aiming to move straight to stage 4 or 5. Stage 4 or 5 brings us back to the issues denoted above.
The insurance perspective.
From an insurance perspective we would be faced with one of the largest overhauls to motor insurance we have ever seen. Almost certainly, accident frequency would be much lower in self driving cars – indeed, most accidents these days are attributed to driver error but what if there was to be an accident that can’t be attributed to driver error, who is to blame?
If the software fails or malfunctions would the liability for the accident lie with the manufacturer? If so, would motor insurance premiums drop and a new kind of manufacturer liability premium be introduced? Furthermore, it is likely that it will take years for the full effects of the technology to be realised.
Driverless vehicles are coming and it is no doubt an exciting leap forward for automotive technology but there are still many hurdles to overcome before we see it becoming mainstream.
As for 2 years? It is possible but I can’t see it myself outside of further test scenarios.
What are your thoughts?