Shared space intersections

[Underseer]

You've probably read about these things. More and more city planners in Europe are converting intersections to "shared space," which means stripping out all street signs, street lights, etc. Counterintuitively, many people freak out, slow down, and pay more attention, which results in fewer accidents while maintaining decent traffic throughput.

As the video mentions, there is some concern that these intersections make things worse for certain disabled people.

My concern is something else: what happens after people get used to intersections like this? At the moment, these intersections work because of fear. Fear causes people to become more cautious, slow down, and pay more attention. Right now these intersections are novel, but what happens after people get used to these things? Wouldn't they become less afraid? Couldn't familiarity reduce the safety advantage of shared space intersections?

Anyway, what do you guys think? Is this the wave of the future?

 

[Speakpigeon]

I think most people will remain weary and err on the safe side but a small minority will get used to these things and take advantage of the principle to zoom through traffic at speed, counting on their own superior reactions. And on occasions, two of these people will collide at full speed.

This is natural selection at work, which explains why most people will stay on the safe side.
EB

 

[Underseer]

I think most people will remain weary and err on the safe side but a small minority will get used to these things and take advantage of the principle to zoom through traffic at speed, counting on their own superior reactions. And on occasions, two of these people will collide at full speed.

This is natural selection at work, which explains why most people will stay on the safe side.
EB

I'm more worried about one of those idiots plowing into an old person or someone in a wheelchair or a baby in a stroller.

 

[RavenSky]

I think most people will remain weary and err on the safe side but a small minority will get used to these things and take advantage of the principle to zoom through traffic at speed, counting on their own superior reactions. And on occasions, two of these people will collide at full speed.

This is natural selection at work, which explains why most people will stay on the safe side.
EB

I'm more worried about one of those idiots plowing into an old person or someone in a wheelchair or a baby in a stroller.

Me too.

I understand the point that they are making, and I do like the concept of "pedestrian priority" streets, but the illustrated space seems like it puts the responsibility for crossing safely on the pedestrian.

The demonstration with the little boy doesn't tell us anything as most people will stop for a person in the street... IF they see the person in time.

 

[Shake]

I had an idea once — which with today's computation speed and power may be closer to realization — of making traffic signals more intelligent. Rather than lights being set for specific time intervals — even with adjustments for time of day and usual traffic patterns — things could be optimized to be more efficient for the most drivers, reducing travel times and increasing fuel efficiency.

 

[bilby]

None of this matters; All of our current methods for preventing cars from crashing into each other, or into pedestrians, or into other hazards, will be obsolete in a few years, when the weak link - the human driver - is taken out of the system.

A car with radar and lidar sensors, that is in constant two-way communication with all the other, similarly equipped, vehicles in the area, will have a far better baseline understanding of where other moving and stationary hazards are, where they are going, and how they plan to get there. Cars can approach intersections at speeds selected long before their arrival such as to maximize traffic flow while minimizing conflicts with other vehicles or with pedestrians. Road signs and traffic lights will be needless; vehicles will be able to flow freely, with priority determined by optimizing algorithms that ensure that every vehicle travels at the best possible speed conversant with comfort and safety, from departure to destination. (Whether that means fastest, most fuel efficient, or something in between can be determined by the passengers).

Almost all of the stuff we have now is designed to inflict large margins for safety on everything we do; But if the cars are talking to each other, and are highly aware of their surroundings and of each other's intentions, many of these margins can be significantly reduced.

[YOUTUBE]http://www.youtube.com/watch?v=kh7X-UKm9kw[/YOUTUBE]

 

[James Brown]

I agree that self-driving vehicles will soon be adequate in the majority of scenarios.

Then there are the edge cases.

For example, will a self-driving vehicle know to pull over when an emergency vehicle is behind it? The standard answer is that emergency vehicles will have some kind of broadcast sensor that alerts vehicles ahead of it to pull over. Now imagine that broadcast sensor in the wrong hands.

I've had police officers wave me through a red light when directing traffic around an accident. Will a self-driving vehicle understand the difference between stopping at a red light and proceeding through just because some random person is waving ahead of it? Will the vehicle know when the waving person is an official uniformed police officer and not just some guy in dark blue waving to a friend?

Presumably a self-driving vehicle will come to a stop if a pedestrian walks out into the road in front of it. So imagine a guy stands in front of an autonomous delivery truck, while his partners breaks into the rear of the truck to pilfer the contents.

I agree that autonomous vehicles will be here eventually, but I enjoy thinking of these kinds of scenarios.

 

[Jimmy Higgins]

You've probably read about these things. More and more city planners in Europe are converting intersections to "shared space," which means stripping out all street signs, street lights, etc. Counterintuitively, many people freak out, slow down, and pay more attention, which results in fewer accidents while maintaining decent traffic throughput.

As the video mentions, there is some concern that these intersections make things worse for certain disabled people.

My concern is something else: what happens after people get used to intersections like this? At the moment, these intersections work because of fear. Fear causes people to become more cautious, slow down, and pay more attention. Right now these intersections are novel, but what happens after people get used to these things? Wouldn't they become less afraid? Couldn't familiarity reduce the safety advantage of shared space intersections?

Anyway, what do you guys think? Is this the wave of the future?

Like all things in civil engineering, there is no one size fit all solution. This has its advantages, but there is trouble in plenty of areas and it is called Right-Of-Way. You can't make circles where the ROW doesn't exist. In general, at least in the midwest, roundabouts (rotaries to those in Massachusetts) have finally become popular as they allow for an on-demand access to the intersection which is much better than lights.

And while this no holds barred access sort of thing can work, I think it increases driver anxiety and the elderly most likely can freeze in these conditions.

 

[RavenSky]

I agree that self-driving vehicles will soon be adequate in the majority of scenarios.

Then there are the edge cases.

For example, will a self-driving vehicle know to pull over when an emergency vehicle is behind it? The standard answer is that emergency vehicles will have some kind of broadcast sensor that alerts vehicles ahead of it to pull over. Now imagine that broadcast sensor in the wrong hands.

I've had police officers wave me through a red light when directing traffic around an accident. Will a self-driving vehicle understand the difference between stopping at a red light and proceeding through just because some random person is waving ahead of it? Will the vehicle know when the waving person is an official uniformed police officer and not just some guy in dark blue waving to a friend?

Presumably a self-driving vehicle will come to a stop if a pedestrian walks out into the road in front of it. So imagine a guy stands in front of an autonomous delivery truck, while his partners breaks into the rear of the truck to pilfer the contents.

I agree that autonomous vehicles will be here eventually, but I enjoy thinking of these kinds of scenarios.

Are you a science fiction writer?

As in, those questions would make a great basis for a hard science fiction novel.

 

[Jimmy Higgins]

Yeah, traffic control in construction zones will likely require some sort of signal to indicate to the vehicle there is a lane closure or flagging. I wonder how the current cars handle it on the highway.

 

[Worldtraveller]

Yeah, traffic control in construction zones will likely require some sort of signal to indicate to the vehicle there is a lane closure or flagging. I wonder how the current cars handle it on the highway.

Unless something has changed recently (~3 months), most self driving cars currently on the road are limited to 25 - 30 mph, and do not go on the highways. I know the Tesla that killed its driver was an exception to that, but I don't know if others are.

 

[bilby]

I agree that self-driving vehicles will soon be adequate in the majority of scenarios.

Then there are the edge cases.

For example, will a self-driving vehicle know to pull over when an emergency vehicle is behind it? The standard answer is that emergency vehicles will have some kind of broadcast sensor that alerts vehicles ahead of it to pull over. Now imagine that broadcast sensor in the wrong hands.

There's no particular reason to imagine that people will abuse such systems any more than they abuse the current 'flashing lights and siren' system. Can you tell the difference between an un-marked police car heading to an emergency, and a citizen impersonating police in order to get through the traffic faster than everyone else? I doubt it - but it's not a common crime, because the risk/benefit is too small - impersonating emergency services is taken very seriously by the authorities, and doing so electronically would be treated no less seriously than doing so by fitting blue and red strobes behind your radiator grill would be.

I've had police officers wave me through a red light when directing traffic around an accident. Will a self-driving vehicle understand the difference between stopping at a red light and proceeding through just because some random person is waving ahead of it? Will the vehicle know when the waving person is an official uniformed police officer and not just some guy in dark blue waving to a friend?

If you can recognize a police officer, what makes you think that a computer cannot? Particularly if a transponder becomes part of the uniform. It could be built into police badges - which, as discussed above, it is a serious offense to imitate with intent to impersonate an officer.

Presumably a self-driving vehicle will come to a stop if a pedestrian walks out into the road in front of it. So imagine a guy stands in front of an autonomous delivery truck, while his partners breaks into the rear of the truck to pilfer the contents.

Imagine if one truck in ten, or one in twenty, has an armed guard on board. How long do you think such thieves will go before getting caught?

Human driven trucks can be stopped in exactly the same way - no truck driver is going to run someone down deliberately, and cargoes have been stolen by this technique in the past. It's not a new problem, nor one that is unique to autonomous vehicles; It's a lot cheaper to pay for a security guard in a small random selection of trucks than to pay for a driver in every truck, and the overall effect is much the same. Of course, autonomous vehicles also capture detailed footage of all the people and vehicles they encounter, and that could be streamed in real time to HQ, rendering any such hijacking very easy to investigate.

I agree that autonomous vehicles will be here eventually, but I enjoy thinking of these kinds of scenarios.

And it's important that we do. But so far I have yet to see any that are a realistic barrier to this technology.

 

[bilby]

Yeah, traffic control in construction zones will likely require some sort of signal to indicate to the vehicle there is a lane closure or flagging. I wonder how the current cars handle it on the highway.

Unless something has changed recently (~3 months), most self driving cars currently on the road are limited to 25 - 30 mph, and do not go on the highways. I know the Tesla that killed its driver was an exception to that, but I don't know if others are.

That's not a technical limitation though - it's just political. Freeway driving is an easier problem to resolve than urban low-speed driving, for an autonomous system, or for a human driver for that matter.

Tesla cars are legally prohibited from full autonomy, and the driver of that car killed himself by wanton disregard for the rules - 'autopilot' is an advanced cruise-control system, and is not intended to replace a human driver. If someone is dumb enough to set their cruise control and then stop concentrating on the road, the subsequent crash is entirely their own fault.

 

[Shake]

This seems highly relevant to this discussion (courtesy: xkcd):

 

[dendrast]

Yep. It'll all be moot when there are self driving cars. If the car you're in crushes a granny, it won't be your fault. Just some programmer's, far, far away.

 

[bilby]

Yep. It'll all be moot when there are self driving cars. If the car you're in crushes a granny, it won't be your fault. Just some programmer's, far, far away.

And, in excellent news for grannies everywhere, it will be an extraordinary and unusual event worthy of international headlines, rather than a daily occurrence that hardly makes the local news.