Car 'autopilot' concerns?

kingstongraham

PBlakeney wrote:

KingstonGraham wrote:

We're all assuming this will be positive for cyclists - the only people still in control of vehicles on the road. How long is that going to be allowed?

If we stick to the centre of the road, we become virtually impossible to overtake don't we? Is that a good thing?

Justification - Cyclists don't kill people. Drivers do.
(Mostly)

I really don't understand. How will drivers kill people once there aren't any drivers?

wiznaeme

HaydenM wrote:

My dad is convinced cars with autopilot will mean he can get home from the pub when he is pissed. Personally I can't see that being the case until human controls are removed completely as the driver will (in the eyes of the law) still technically be in control of the car. This guy watching a film does call that into question a bit...

If the car doesn't have a steering wheel, and three drunk people are in it, no one behind a wheel, then I would think no one could be identified as the driver and charged with drunk in charge. This may sound strange but driverless cars may be more like delivery pods or a theme park train, eh, or something. Interesting times ahead.

capt_slog

Giraffoto wrote:

If we want to make ourselves more visible to radar we need a miniature version of the radar reflectors you get on boats, like this:


If you made it about the size of a Rubik cube you could fit one under the saddle and one under the stem. Anyone got a 3D printer that does aluminum, and a kickstarter account? Make it a bit smaller and it could be fitted inside the downtube of a carbon frame - where it would still be nicely visible to radar.

Not sure about that,

I get the feeling that they have to be a certain size to work, and that might depend upon some ratio of the radar's (radio) wave

kingstongraham

WiznaeMe wrote:

HaydenM wrote:

My dad is convinced cars with autopilot will mean he can get home from the pub when he is pissed. Personally I can't see that being the case until human controls are removed completely as the driver will (in the eyes of the law) still technically be in control of the car. This guy watching a film does call that into question a bit...

If the car doesn't have a steering wheel, and three drunk people are in it, no one behind a wheel, then I would think no one could be identified as the driver and charged with drunk in charge. This may sound strange but driverless cars may be more like delivery pods or a theme park train, eh, or something. Interesting times ahead.

Like Johnnycabs

pblakeney

KingstonGraham wrote:

PBlakeney wrote:

KingstonGraham wrote:

We're all assuming this will be positive for cyclists - the only people still in control of vehicles on the road. How long is that going to be allowed?

If we stick to the centre of the road, we become virtually impossible to overtake don't we? Is that a good thing?

Justification - Cyclists don't kill people. Drivers do.
(Mostly)

I really don't understand. How will drivers kill people once there aren't any drivers?

Nobody dies. Everybody is happy.
PS - Mopeds.

veronese68

finchy wrote:

That would be good, but how do you fit them to hedgehogs?

Fit some spikes to them and they will attach like Sticklebricks.

Giraffoto

Capt Slog wrote:

Giraffoto wrote:

If we want to make ourselves more visible to radar we need a miniature version of the radar reflectors you get on boats, like this:


If you made it about the size of a Rubik cube you could fit one under the saddle and one under the stem. Anyone got a 3D printer that does aluminum, and a kickstarter account? Make it a bit smaller and it could be fitted inside the downtube of a carbon frame - where it would still be nicely visible to radar.

Not sure about that,

I get the feeling that they have to be a certain size to work, and that might depend upon some ratio of the radar's (radio) wave

I believe (possibly wrongly, where are the people who know what they're talking about?) that an ideal reflector has to be at least half the wavelength long, and that shorter wavelengths are used for more accurate tracking. So if the wavelength is short enough, it may be possible.

drlodge

Today, you can be charged if you don't have your hands on the steering wheel. In some years time once driverless cars are proven to be safe (or at least as safe as cars with drivers) I expect laws will be changed so you can be charged if you *DO* have your hands on the steering wheel (assuming there still is one).

And then we'll all look back at the medieval way we used to try and drive cars safely and wonder how more accidents didn't happen.

kingstongraham

PBlakeney wrote:

KingstonGraham wrote:

PBlakeney wrote:

KingstonGraham wrote:

We're all assuming this will be positive for cyclists - the only people still in control of vehicles on the road. How long is that going to be allowed?

If we stick to the centre of the road, we become virtually impossible to overtake don't we? Is that a good thing?

Justification - Cyclists don't kill people. Drivers do.
(Mostly)

I really don't understand. How will drivers kill people once there aren't any drivers?

Nobody dies. Everybody is happy.
PS - Mopeds.

So you expect there'll be a flow of cars that any pedestrian/cyclist/moped rider can cause to slow down at virtually no risk to themselves. That isn't going to happen - there will be more segregation to keep things moving, away from residential streets where that would be regarded as a positive.

Be more off road cycling I reckon. I'll be dead by the time it's all sorted anyway.

fenix

I don't like your view of the future Graham.

Cars could be narrower - they will be safer with less collisions due to people driving - so they can give us adequate space on the road.

No boy racers mean no head on collisions - they'll probably have some kind of radar that can see what is round the next bend and make overtaking safer for everyone.

I dont think we need worry about this just yet anyway.

kingstongraham

Fenix wrote:

I don't like your view of the future Graham.

Cars could be narrower - they will be safer with less collisions due to people driving - so they can give us adequate space on the road.

No boy racers mean no head on collisions - they'll probably have some kind of radar that can see what is round the next bend and make overtaking safer for everyone.

I dont think we need worry about this just yet anyway.

I think the future with driverless cars sounds great. If I can get from A to B without having to drive, I'm happy.

Cars can then drive closer together and they don't really need to overtake as much unless you include in the program a ludicrous need to feel superior.

pblakeney

KingstonGraham wrote:

PBlakeney wrote:

KingstonGraham wrote:

PBlakeney wrote:

KingstonGraham wrote:

We're all assuming this will be positive for cyclists - the only people still in control of vehicles on the road. How long is that going to be allowed?

If we stick to the centre of the road, we become virtually impossible to overtake don't we? Is that a good thing?

Justification - Cyclists don't kill people. Drivers do.
(Mostly)

I really don't understand. How will drivers kill people once there aren't any drivers?

Nobody dies. Everybody is happy.
PS - Mopeds.

So you expect there'll be a flow of cars that any pedestrian/cyclist/moped rider can cause to slow down at virtually no risk to themselves. That isn't going to happen - there will be more segregation to keep things moving, away from residential streets where that would be regarded as a positive.

Be more off road cycling I reckon. I'll be dead by the time it's all sorted anyway.

Much in the same way as it is supposed to work today.
Traffic gets held up for a short while and passes when safe to do so. Or, do you advocate dangerous passing?

Tashman

Im obviously in the minority here who actually enjoy driving a car too. Personally i hope they don't come in during my lifetime to any mass degree.
I also can't see them becoming normalised until they are able to "dock" and refuel somehow themselves at their journey's end otherwise i envisage people ending up stranded as no-one took responsibility to "fill er up"

kingstongraham

PBlakeney wrote:

KingstonGraham wrote:

PBlakeney wrote:

KingstonGraham wrote:

PBlakeney wrote:

KingstonGraham wrote:

We're all assuming this will be positive for cyclists - the only people still in control of vehicles on the road. How long is that going to be allowed?

If we stick to the centre of the road, we become virtually impossible to overtake don't we? Is that a good thing?

Justification - Cyclists don't kill people. Drivers do.
(Mostly)

I really don't understand. How will drivers kill people once there aren't any drivers?

Nobody dies. Everybody is happy.
PS - Mopeds.

So you expect there'll be a flow of cars that any pedestrian/cyclist/moped rider can cause to slow down at virtually no risk to themselves. That isn't going to happen - there will be more segregation to keep things moving, away from residential streets where that would be regarded as a positive.

Be more off road cycling I reckon. I'll be dead by the time it's all sorted anyway.

Much in the same way as it is supposed to work today.
Traffic gets held up for a short while and passes when safe to do so. Or, do you advocate dangerous passing?

Yes, clearly that's what I advocate. Love it. The adrenaline makes me feel alive.

Alex99

KingstonGraham wrote:

Fenix wrote:

I don't like your view of the future Graham.

Cars could be narrower - they will be safer with less collisions due to people driving - so they can give us adequate space on the road.

No boy racers mean no head on collisions - they'll probably have some kind of radar that can see what is round the next bend and make overtaking safer for everyone.

I dont think we need worry about this just yet anyway.

I think the future with driverless cars sounds great. If I can get from A to B without having to drive, I'm happy.

Cars can then drive closer together and they don't really need to overtake as much unless you include in the program a ludicrous need to feel superior.

I think the possibilities are overwhelmingly positive. Can't wait to see the changes the next 20 years brings. If it means that I don't have to own a car to use a car at my convenience, then great. It's just uber+

pilot_pete

The problem in my mind is designing a system that can overcome the many, many variables and 'make decisions' with a minimal failure rate.

I am an airline pilot by trade and we have had autopilots for years and years. Indeed the first aeroplane that could land itself (an auto land) was the Trident. However, even today an auto land takes significantly more input and system monitoring by the pilots than a straight forward manual landing. The autopilots cannot make any decisions and any system failures must be acted upon immediately by the pilots to decide if the aeroplane can continue to auto land or is degraded to the point that an auto land will not be successful.

The problem with such automated systems, including the sophisticated autopilots that we have today is that we take the human away from 'doing', which they can become very highly skilled at and make them 'monitor' a computer which they are relatively poor at (if you think about constant monitoring without ANY breaks in monitoring). Incidentally, computers are brilliant at monitoring and throwing out a warning but the designers have decided to go the other way.

Another problem is that autopilots don't make decisions. They just follow commands. Everyone thinks airline pilots press one button and the aeroplane takes them from A to B with no further intervention. It is really laughable, what an autopilot does is free up brain capacity for decision making by pilots who manage the whole operation, not just the flight path of the aeroplane. The aeroplane can follow a course that is pre-programmed, but cannot update this route in response to air traffic control or for avoiding action etc. It cannot climb or descend without intervention, so actually the pilot is still flying the aeroplane, just using different controls to do it.

The difference with an automated car is the complexity of the decision making which must occur - something a trained human is pretty damn good at doing, but computers aren't yet as good at. It has to make the decisions and manipulate the controls (the easy bit). It seems funny that I read in the aftermath of the fatal accident that the driver was not operating the system correctly - he was supposed to be ready to take control from the 'autopilot' instantly. What a surprise, make the human do the monitoring (which remember we are pretty bad at as we are easily distracted), make that human a completely untrained member of the public who is watching a film, playing in his phone etc etc and an accident ensues.

There is concern in the aviation industry about pilots losing hand flying skills due to a new generation who rely more and more on the automatics and airlines, some of which dictate that the autopilot should be in control for huge chunks of the flight and discourage hand flying.

Funny how pilot monitoring of automated systems has become a huge risk and causal factor in a number of accidents - yet again, make the human monitor the computer and then wonder why his performance degrades. Why not make the computer do the monitoring, which they are brilliant at?

I see the latest accident was due to the car being driven on a twisty single carriageway by the autopilot, which apparently they aren't as good at doing and Tesla have said the autopilot is better suited to highways with dividers...still some way to go then before the systems are robust enough? I think so...

The problem is the complexity of the decision making and the huge variables that the real world throws up. It is strange that they are advocating the car doing the driving and the 'driver' does the monitoring of the systems! In my opinion Joe Public is poorly equipped to act in such a capacity, yet with driving experience most are pretty damn good at driving...

Remember, airline pilots are highly trained and constantly tested professionals and the industry has problems with 'pilot monitoring'. The future of car automation must take the driver completely out of the equation as their performance will degrade more and more as far as intervention skills go. This is obviously what the industry is aiming at but it would appear we are still some significant way off. The other issue is perception - statistically the automated cars may well end up safer but for many years to come the crashes are going to attract a disproportionate amount of bad press and negative sentiment. They will therefore have to be nigh on perfect to be successful!

PP