If robots and human drivers can’t get along, is geofencing the answer?
There has been a huge amount written, said and promoted about autonomous cars. Read or listen to some and you’d think we should all be sitting in vehicles that require no input from us at all already, blasting from here to there while catching up on reading, sleeping, work or leisure.
The reality has proven somewhat different and is likely to remain that way for a very long time to come.
Will fully autonomous cars and human-driven vehicles ever be able to mix?
At present, we have Level 2 autonomy in most new cars. This is where systems such as adaptive cruise control, parking assist and emergency collision avoidance are all part of a suite of aids to make driving safer and less stressful. The key point about all of these functions is they always require the driver to be in charge of guiding the car and retain responsibility for its overall control.
There are a number of cars on the market already that are called Level 3 Ready. These include Tesla cars, the latest Audi A8, BMW X7, Nissan Leaf and some Volvos. What they allow is for the driver to remove his or her hands from the steering wheel for a period of time and let the car take care of guiding itself along the road.
Yet there are only certain circumstances where this Level 3 autonomy will function properly. The car needs to be able to ‘see’ clear road markings to know where the correct lane is so it can stay within those confines. It also requires the driver to give the steering wheel a regular input, usually every 10 seconds or so, to let the car know you are awake and alert. If it doesn’t receive that input, the car will gradually come to a safe stop.
What actually defines an automated vehicle?
However, this Level 3 autonomy is giving the auto industry some big headaches. After all, what’s the point of letting the car drive itself for very brief periods if the driver still has to constantly monitor what’s going on? From a consumer’s point of view, this seems like the worst of both worlds and there’s certainly a good deal of resistance from many drivers to use these systems even when they are fitted to a car they’re driving.
To create a car that can genuinely drive without much in the way of interaction or input from a human, vehicles need to achieve Level 4 autonomy, and that is still a long way off being applied to the entire roads network. Instead, this is where many within the car industry reckon Level 3 autonomy is merely a stepping stone and a big leap is needed to move swiftly to Level 4. This is where geofencing comes into the equation.
Rather than letting highly autonomous vehicles mix with normal traffic, a lot of engineers and automotive leaders believe geofencing is the answer. This creates a virtual boundary around an area where autonomous vehicles can move independently. A good example is a shopping mall car park, so rather than you having to spend ages driving around to find a parking space, you jump out to head off to the shops and the car will make its way to a pre-determined parking bay. That saves time and hassle.
Self-parking is an area where autonomy makes complete sense.
Crucially, this self-parking autonomy makes sense as it frees up time for the human side of the deal and makes parking a much quicker process for the vehicle. Also, with no human to worry about getting in and out of the car, more cars can be parked in the same given space to make it more efficient. Of course, this requires all vehicles in this type of scenario to use the same compatible autonomy.
Geofencing has other advantages for developing autonomous vehicles. For starters, a controlled environment is always needed for any experiment, and let’s not kid ourselves that this tech is anything but experimental at this stage. It’s coming on fast, but Volvo and Uber’s just announced autonomous taxi is still really an XC90 with a great, bug cumbersome box on its roof to sense what’s around it.
By keeping this kind of vehicle within pre-set boundaries, it’s easier to let it mix with similarly equipped vehicles. They can then be studied without the huge number of variables brought on by human-driven vehicles.
Even so, there comes a point when autonomous vehicles will have to deal with much more than a controlled environment if they are to work in a real-world fashion. Humans may be fallible, but they are also incredibly good at taking in vast amounts of information and acting on it at lightning speed. Just driving down any normal city street presents the human brain with huge possibilities for danger, resulting actions and outcomes.
For autonomous vehicles, all of this has to be learned, processed and then put into action. The only way to do all of that is to set these vehicles among other traffic to find out how they cope and behave. This is why geofencing has become so important as no car company wants the risk of their vehicle being involved in a collision due to the autonomous tech not being able to deal with the situation.
The shift in thinking and attitude of many within the car industry towards using higher level autonomy as part of the solution to improved driving safety is a result of this. In place of autonomous vehicles completely replacing the driver, there is a move towards seeing the human and machine being used to their key strengths. If the car can take car of many of the mundane tasks that a lot of drivers find boring, it frees up their attention to concentrate on more core functions.
Rather than being something to fear, autonomy presents drivers with an opportunity to embrace new tech that could make driving more enjoyable while also being safer. We will still see fully autonomous vehicles where the driver has not part of the driving at all, but these are set to be pods that will still be separated from most other traffic. For those of us who still find driving a pleasure, the future continues to hold exciting possibilities.