Autonomous cars #2: Speed Limits and other ignored-laws

Previous post #1: If an autonomous car break the speed limit -who takes the blame?

Cycling journalist Carlton Reid, followed up by referencing an article by "Robocar" engineer and evangelist, Brad Templeton, who argues that fears of robocar-related crashes are overrated. He may well have a point given the willingness of todays drivers to make decisions that endanger "overtake cyclists before a corner" with followon decisions that compound the first mistake "I continued ever after seeing the oncoming cyclists".

Robocars are unlikely to suffer from it is the unwillingness to reverse a decision once the situation made it clear it was not the right one. Yet we see that on the road every day: the overtaker who keeps going, the left-hook van driver who, even though they have underestimated the cyclists' speeds, continues to pass and turn. Removing egotism from the decision making process may be progress.

However, the article wasn't actually on the reliability of robocars, but on the moral issue: should robocars break speed limits -and if so, who is at at fault.

Brad actually picks this issue up at one point:

More complex are the situations where breaking the vehicle code is both normal and even necessary, particularly for unmanned vehicles. One must be assertive on some roads in order to get through at all, and everybody does it and nobody is ticketed except in an accident or during zero-tolerance enforcement days.

He has acknowledged that breaking traffic laws is a social norm in many cities, and  rather than saying "it will be impossible for robocars to break the laws", he says "it's complex".  Yet he hints at why the laws get disregarded: because you can. If you only get tickets on "zero-tolerance" days then you can discount the law.

Which is of course precisely why the Daily Mail Fuckwits are so strongly against speed cameras, RLJ cameras and CCTV enforcement of school keep clear zones and bus-lanes. We can automate enforcement of traffic laws today -yet to do so generates a backlash from those who consider being able to break them a human right.

Brad gets into detail on the speed limit issue in another article, where he argues that it should be the right of a robocar to decide for itself what a safe speed is, looking at the two options of speed limits

1. As with its ancestor, the cruise control, the operator of a robocar can set the car to operate at any speed within its general limits, regardless of the road speed limit. The moral and safety decisions rest with this person.
2. The vehicle must be programmed to not break the speed limit, nor allow its operator to do so. It must be aware of all limits and obey them. 

I believe the first choice is both better and more likely. It's more likely because the public has a strong love for having control of their car, even if it is automated. Attempts to put in speed limiters by law have all been rejected, and cars are routinely sold able to go much faster than any allowed speed limit. 

That is, he believes that the driver has the right to make the decisions, even though they endanger others more than the driver -and society has  a set of laws because we recognise that the drivers aren't the best people to make those decisions.

Key claims here:

A vehicle limited to the speed limit will be going much more slowly than traffic on most US freeways, and be forced to drive in the right lane. (& arguments against this)

--- Only if the speed limits were not being enforced. If they were, the robocar would be integrated with the normal traffic flow.

"On many roads all lanes are moving faster than the limit. The limited car would become an obstruction to traffic."

-- Only because all lanes are full of drivers breaking the law. I they were all driving legally this would not be an issue

"This less comfortable ride, plus the longer travel time, will create a great temptation to manually take the wheel on many highways." (& increased risk follows)

Again, this only holds if the driver could break the speed limit with impunity. If speed limits were enforced there would be no benefit of manual driving, hence no increased risk. Anyway, surely robocars are about the ability to work while you drive, to not get tired on long journeys (where the speed difference between 70 and 80 mph may actually alter journey times)

Eventually he comes out and  makes his stance clear -and this is a stance of someone who works on robocars at at google-

I believe the math and other arguments clearly show that robocars should be allowed to move faster than the speed limit so long as they are rated suitably safe in the particular conditions, and the bulk of other traffic is also doing this.

This is just the safe speed and 85% percentile bollocks restated. Yet it ignores how the defacto speed limits of a road and a country have evolved: through a failure to enforce the limits and the gradual acquisition of a motor vehicle fleet designed to break those limits with impunity. Now that a majority of British cars are stable and quiet above 80 mph, the drivers and passengers believe that it is not only safe to do so -but their right- and that any attempt to rigorously enforce the speed limits becomes a political minefield.

Finally, Brad comes out and admits the situation we have in cities today: breaking the law is sometimes necessary because everyone else does: Speeding is just one of code violations almost everybody does.
 

In cities today, "aggressive driving" is often viewed as necessary. Examples: squeezing past bicycles, pulling out half-way across a road to complete a right turn, pulling out in front of oncoming traffic at a roundabout, turning right across the front of an oncoming car.

These are actions that are commonplace in a city, viewed as acceptable to drivers -yet which are some of the actions which are utterly terrifying to anyone trying to cycle in the area.

The "rules of the road" which have evolved in our cities are rules built around  800 Kg vehicles with ABS brakes and fast acceleration -and an assumption that they have right of way over all others. 

In the UK, the law that gets broken the most must be the "give way to pedestrians when turning" law. Anyone walking would die if they believed this would hold: so what will the robocar story be here? To actually give way to pedestrians? Or to follow the actions of the rest of the motor fleet?

The choice of which driving policies to implement in a robocar is something we cyclists cannot naively hope will be made so as to suit us. It will be made to suit the paying customers, within the constraints of the legal system. If robocar sales are suppressed because they don't drive aggressively enough, because they don't speed on the motorways, then the manufacturers will keep pushing this "we need to keep up with traffic" bollocks, this "everyone does it" rule.

But what will happen over time? Even if the first generation of robocars broke the speed limit, cut up pedestrians and squeezed past cyclists "because everyone did it", will there be some threshold -say when 51% of the motor fleet is automated, when suddenly the "everyone does it" argument becomes obsolete? As at that point vehicles will be driven dangerously not because people choose to drive that way, not because robocars do that way to blend in, but because it is what robocars have been programmed to do -and even when they are in the majority, it is how they will behave.

This is why it is critical that autonomous cars are not allowed to break speed limits in autonomous modes from day 1. If not, a precedent will be set that will remain unchallenged even if the cars form the vast majority of the fleet. There won't be a sudden "75% day" when there are deemed to be enough of them that they have to stop speeding, to stop blocking junctions to make progress, or to start giving way to pedestrians. These have to be done from the outset, so that the robots can set an example for the humans.