I used to think that all us driving instructors would soon be out of a job. Attending the last DIA conference and the Portsmouth University Artificial Intelligence Transport Hub Conference tells me something different. There are going to be three basic problems that need to be solved for autonomy to work.
- The Driver:
Nobody is giving much thought to the driver. Autonomous vehicles will have 5 levels:
- Feet off
- Hands off
- Eyes off
- Brain off
- Complete autonomy
All the way up to level 4 will require the driver to be ready to take over in uncontrolled or hazardous conditions.
How will our nervous and anxious pupils, who if regularly using these types vehicles will have had little driving experience, be able to cope when everything is going wrong. As instructors we are all aware of how unpredictable a learner or inexperienced driver can be.
Skills will need to be practiced to be of use. We are ourselves constantly reminded of the need to keep training. Our skills are assessed to ensure we remain up to scratch. Will insurance companies insure people to drive who have very limited experience and will have to do so in difficult circumstances. Will there be substantial reductions for the driver who can prove their skills are up to date.
As professional instructors, we all know that other motorists treat our learner cars differently. Nobody wants to get stuck behind a learner. Will not the selfish drivers will be able to take advantage of autonomously driven ones. Think about how other drivers cut your learners up. While someone might think twice about cutting a person up, how about cutting a machine up.
Nobody has given much thought to how the transition to autonomy will work. How will drivers maintain the skill level to deal with uncontrolled or hazardous conditions. What will be the challenges.
- The driving environment:
An autonomous vehicle will need to know where it is and where it’s going. It will have a series of sensors to measure the road in different ways. These sensors are Camera, Radar and Lidar (laser detection). The new driving test has put us all on to rural roads. Edges of these roads are either grass, granite kerb stones, or granite kerb stones covered in grass. What about when the pot hole has a nice bit of ice covering it. Road signs and road markings can all be obscured in different ways.
Radar, Lidar or cameras will all have to cope with this routinely. Besides their normal limitations what will be their capabilities in hazardous conditions.
Selby rail crash 2001 was caused by a Range Rover. The insurer sued the DoT for part of the £22 million pound settlement. This was on the basis of inadequate crash barriers. Will The Highways Agency and Local Authorities have to maintain roads and signs to better standards so the autonomous systems can cope. Think about how likely that is next time you go over a pothole.
Sat nav will be an essential part of how these cars will function. It’s only as good as the last update and the info in that update. When out on test the examiner will quite often correct the sat nav or at least make it clearer.
The manufactures will no doubt say that it will be learning all the time and getting better and more accurate. But sat nav been around for years and still gets things wrong. What about mini roundabouts, private roads and junctions that aren’t recognized by sat nav. Why isn’t that getting more accurate.
Knowing where you are and recognising the hazards around you are key components in the move to autonomous cars. The car will need multiple sensors and systems of different kinds to properly read what’s happening around it. These need to be maintained and calibrated. When sensors degrade, malfunction or fail at what point will the system stop controlling the car. How will that become apparent.
The car insurance Thatcham Research group have a document giving an insurance industry view. https://news.thatcham.org/documents/assisted-and-automated-definition-and-assessment-summary-79399
This states among other things that the cars must be law abiding and comply with the advice in The Highway Code.
Try driving round a busy urban environment without breaking some aspect of the code or the law. Minor transgressions to maintain traffic flow are not normally enforced by the police. A police officers sensible discretion will compliment a drivers careful and considerate driving.
For instance, what if hypothetically my nearside wheels went into a bus lane to get around the car in front which was turning right. Only for a second and only just the nearside wheels. All the cars in front had already done so.
The law is quite clear. You must not cross a solid white line. The guide lines from The CPS say that for this to result in my prosecution two things must be in place:
- Evidence of my offence
- A public interest in having me prosecuted
From an evidence point of view for me to get caught, either something must have had to have happened which would require a police presence. Or I had been observed doing this by the police. Or maybe my retractable confession.
Would it have been in the public interest for me to wait and hold up the traffic behind me. Not to have moved would have meant the junction behind me becoming blocked. Would it have been in the public interest to prosecute me. It was perfectly safe.
But will your autonomous vehicle be making that decision. If you want to safely maintain traffic flow you now have a car manufacturer deciding how to break the law and to what extent. Or will rigorously enforced laws lead to gridlock.
What about a busy zebra crossing, will you wait until the pedestrians are safely clear of the crossing and on the pavement. Or will you safely clear the pedestrian? What happens when the law conflicts with the advice.
Hears what the police say about zebra crossings:
Generally, if a pedestrian is on a pedestrian crossing then a driver must give precedence to them and allow them to cross. However, in the case of Kayser v London Passenger Transport Board  1 All ER 231, the court ruled that where a driver is satisfied that persons who are lawfully entitled to cross the road – whether they are on a pedestrian crossing or not – are well out of any danger from him if he goes on in the normal course, he is perfectly entitled to go on, but, of course, only at such a pace as will enable him to stop almost immediately in the unlikely happening of those persons doing something dangerous and negligent themselves.
Nonetheless, we would suggest that it is best for a driver to wait until the pedestrian has left the crossing before proceeding, so as to avoid any potential issues.
The law says one thing, police advice says another!
Go out on test and your learner will likely fail if they move on when a pedestrian is still on a crossing. At a busy zebra crossing and you could be waiting a long time if you only go following the advice given above. Who would want to get stuck behind an autonomous car.
Next time you are reading your Highway Code have a think about how some of the rules and advice might be applied by an autonomous vehicle. What might the problems be. How will that advice be interpreted.
Think about how autonomous vehicles will deal with bicycles. Rule 163 says give bicycles as much room as you would give a car when overtaking. The Safe Pass advice for going round a bicycle is 1.5 meters.
We now have advice from The Highway Code and the police which could be said to be saying when you overtake another vehicle you should give a gap of 1.5 meters. In congested or narrow roads will an autonomous vehicle be able to overtake anything.
So with solid white lines we have seen how the law can affect autonomous vehicles. With zebra crossings how police advice could be different from the law. And with overtaking how the advice from two different sources (police and The DVSA) could create some interesting problems.
Courts will make decisions about laws and advice. Those decisions will decide how autonomy will work. But they will lag behind the technology.
The big problems are how will:
- Our nervous and anxious pupils will cope when required to drive
- The autonomous vehicle cope with badly maintained and poorly mapped environments
- Law and advice affect the routine operation of autonomous vehicles
My own conclusions are everything will become safer and journey times will increase because everyone is following the rules. Bicycles will become more dominant on the road as the autonomous vehicles will treat them properly. There will be some very interesting court cases involving the vehicle manufacturers. Drivers will have to prove they are maintaining a skill level to obtain affordable insurance for the situations that will require them to drive.
One of the big reasons for autonomy is safety with 95% of incidents involve driver error. So what about putting some extra effort into improving the skill level of the driver. At the end of the day that’s what we as driving instructors are all about.