Ride-hailing in Metro Vancouver: Uber, Lyft, Sidecar - emerging technologies

[VancouverOfTheFuture]

Originally Posted by Changing City

Oh, absolutely it was a full time job initially. The first drivers were in limos that charged more than taxis. In the past the company had to pay a 20 million dollar fine for telling drivers they could earn 100k per year driving for Uber, which implies it was a full-time job. Apparently, if you drive for Uber in many places the only way to get the bonuses the company offer is to drive full time.

i remember it being marketed as "ridesharing" at one point. the idea is people sharing a ride to similar places. unless i am mixing them up with another company that does that. either way, i dont see uber surviving if they need to start paying employees. same with things like skip the dishes.

Originally Posted by Changing City

The company... are hoping that driverless vehicles finally become reality. Then they can replace their drivers, and taxis too.

yeah, thatll never happen.

[Migrant_Coconut]

As long as AVs keep doing this, there isn't a business in the world which'd let thousands of them out on the streets at 50+ km/h.

[WarrenC12]

Originally Posted by Migrant_Coconut

As long as AVs keep doing this, there isn't a business in the world which'd let thousands of them out on the streets at 50+ km/h.

Wait until you see the mistakes meatbag drivers make.

[Migrant_Coconut]

Originally Posted by WarrenC12

Wait until you see the mistakes meatbag drivers make.

If a meatbag driver screws up, the meatbag goes to court. They pay a fine, have their ICBC rates go up, lose their license and/or serve up to fourteen years, the end.

If a robot driver screws up, the rideshare company, manufacturers and programmers are all in hot water; even if they all point fingers at each other and walk away unscathed, that's a lot of legal fees. Seems safer for them to cap the top speed at 30-40 km/h.

[Changing City]

Driverless taxi services are already operating in cities in the US and China. It's not going so well in San Francisco though.

[Migrant_Coconut]

And just last year, Cruise recalled 950 cars after two accidents. Again, no way in hell any of them are going to push for 50+ km/h.

[VancouverOfTheFuture]

i think the legal ramifications and insurance will be what prevents them from having mass adoption.

though what happened in san fransisco poses another interesting thought. if theyre that easy to disable, then vandalize, would they even be safe to be a passenger? someone puts a cone on the hood while stopped at a traffic signal. now they can mug you and you cant drive off. then there is also just the general vandalism side of things. easier to justify hurting a company like google by breaking its car, than some random average persons car.

[WarrenC12]

Originally Posted by Migrant_Coconut

If a meatbag driver screws up, the meatbag goes to court. They pay a fine, have their ICBC rates go up, lose their license and/or serve up to fourteen years, the end.

If a robot driver screws up, the rideshare company, manufacturers and programmers are all in hot water; even if they all point fingers at each other and walk away unscathed, that's a lot of legal fees. Seems safer for them to cap the top speed at 30-40 km/h.

Meatbag drivers kill pedestrians all the time and get a slap on the wrist.

[Migrant_Coconut]

Originally Posted by WarrenC12

Meatbag drivers kill pedestrians all the time and get a slap on the wrist.

Meatbags are a lot less forgiving to "blind idiot" algorithms than to other meatbags.

[whatnext]

Originally Posted by WarrenC12

Meatbag drivers kill pedestrians all the time and get a slap on the wrist.

It would be interesting to compare the accident rate between human taxi/Uber drivers and the autonomous ones.

[Changing City]

Originally Posted by whatnext

It would be interesting to compare the accident rate between human taxi/Uber drivers and the autonomous ones.

It's been done. Somebody should teach you how to use a search engine.

[Migrant_Coconut]

As of that article, Waymo'd been capped at city limits of 25-30 mph (40-48 km/h). Now they've been cleared to go up to 65 mph (100 km/h), so we'll see if it's AI or just driving slower that's behind their safety record.

[Changing City]

Originally Posted by Migrant_Coconut

As of that article, Waymo'd been capped at city limits of 25-30 mph (40-48 km/h). Now they've been cleared to go up to 65 mph (100 km/h), so we'll see if it's AI or just driving slower that's behind their safety record.

In San Francisco most streets have a speed limit at, or below 30 mph. It's only the freeways and a few other major streets that have a higher limit. As Waymo point out, in San Francisco 33% of drivers exceed the posted limit, while their vehicles don't do that. (It's an even greater proportion of drivers in Phoenix, the other Waymo trial city).

In Greater Vancouver most roads have a limit at or below 50kph, and traffic often moves slower than that. The worst accident spots aren't on the higher speed roads, they're at busy intersections, often when the lower speed limit kicks in. Self-driving cars will still end up in some accidents because they really can't do much about somebody else driving into the back of a slow, or stationary vehicle, for example, but they're much less likely to mis-judge a turn (or the speed of an approaching vehicle), or get distracted when they get a text at the wrong moment, so statistically they're still likely to drive better than humans - whatever the speed limit.

[Spr0ckets]

Interesting discussion.

On an unrelated (but not entirely unrelated) note, it's worth noting that if any of you do any flying nowadays on commercial airlines, - particularly for long-haul and trans-oceanic flights - you'd be surprised (or not) to know that most of the flight is actually done through autopilot.
Particularly when the plane gets to cruising altitude.

In fact, flight systems are so advanced nowadays that planes can literally take off and land themselves on autopilot and fly the entire flight (once the route is entered in), but it's just human psychology, as well as flight safety protocols that demand a human pilot not only be in the cockpit, but also in control during critical portions of the flight.
(like in case they run into unforeseen circumstances like atmospheric turbulence or if the flight needs to be unexpectedly diverted (like say for a medical emergency on board)).

Other than that, your plane can fly you to your destination entirely on the same (or similar) type of AI systems that everyone is having conniptions about with self-driving cars.
Actually more advanced.
Granted, the traffic is much less up there, and there's less chance of errors (again,....given the lower likelihood of running into error-prone humans).

A final pin on this is the fact that majority of the major flight crashes we've had since we began using automated flight system in planes have been with the human pilot in control of the plane and not the AI system (or autopilot).
Yes, the autopilot systems have a much higher safety record than their human counterparts*.

(*The Boeing Air Max crashes of a couple of years ago were not because of flawed automated computer flight systems, but rather due to poor training of the pilots to fly using the new system, which resulted in them "fighting" the system while it was trying to react as it was programmed to, rather than the system not responding as it should have. That's the closest I can think of where the flight systems could have been blamed for being the cause of the crash, but ultimately it was still human (pilot) error and training.)

I just thought that was something interesting worth noting in this discussion.

One day self-driving cars will be ubiquitous and more common than they are now, and chances are that they will indeed be better drivers than humans because at the end of the day the biggest cause of car and auto accidents nowadays usually can be found between the steering wheel and the driver's seat.

[aberdeen5698]

Originally Posted by Spr0ckets

(*The Boeing Air Max crashes of a couple of years ago were not because of flawed automated computer flight systems, but rather due to poor training of the pilots to fly using the new system, which resulted in them "fighting" the system while it was trying to react as it was programmed to, rather than the system not responding as it should have. That's the closest I can think of where the flight systems could have been blamed for being the cause of the crash, but ultimately it was still human (pilot) error and training.)

I take strong exception to this. The MCAS software that caused the complete loss of two 737 MAX aircraft and all aboard had IMHO two major flaws that were not the fault of the hapless pilots who fell victim to it.

The first is that the software inexplicably relied on the inputs of only one of two redundant angle of attack sensors. If the active sensor suffered a failure that caused erroneous readings, the software would wrest elevator control from the pilots even though the other sensor was still providing valid data. This resulted in a single point of failure that could doom the aircraft. It's a patently absurd design decision to have redundant sensors upon which the safety of the aircraft relies and then to completely fail to check to see if those sensors agree with each other or not.

The second is that in it's eagerness to certify a new version of the 737 without having to update the aircraft's type certification, which would have required costly retraining of pilots, the very existence of the newly-introduced MCAS software and its behaviour was hidden from pilots. They literally were unaware of its presence and what it could do. Pilots had no training for scenarios where faulty angle of attack indications could cause MCAS to pitch the aircraft's nose down using a force stronger than they were able to counteract by pulling back on the control stick. Post-accident investigations have concluded that pilots would likely not have had enough time to understand the what was happening to the aircraft and recover from it before it was too late.

There's a good reason that the 737 MAX was grounded for the better part of two years - it has nothing to do with pilot error and everything to do with the design and implementation of the aircraft's flight control system.

[VancouverOfTheFuture]

Originally Posted by Spr0ckets

(*The Boeing Air Max crashes of a couple of years ago were not because of flawed automated computer flight systems, but rather due to poor training of the pilots to fly using the new system, which resulted in them "fighting" the system while it was trying to react as it was programmed to, rather than the system not responding as it should have. That's the closest I can think of where the flight systems could have been blamed for being the cause of the crash, but ultimately it was still human (pilot) error and training.)

nope, thats not true. MCAS was 100% at fault. boeing put in a single point failure system, with an unreliable sensor, their own testing showed issues, they lied about it, they didnt even include the information in the check lists, didnt tell the pilots about this new system, and they were also AGAINST training crews on the new plane. they talked airlines out of training, even when specifically requested.

even simulator pilots at boeing themselves raised alarms on MCAS being unreliable.

the overall plane design is flawed, and the software they made to overcome this flawed design, is what killed those people. it was NOT pilot error.

[Migrant_Coconut]

The whole reason why MCAS even exists in the first place? Because Boeing's engineers had accidentally redesigned the 737 Max to pitch up... so instead of fixing the weight balance like they should and stabilizing the plane... they had the software automatically pitch it down instead. No amount of training will 100% compensate for a design flaw.

Ditto AVs. More people drive than fly, and one lazy team of car engineers can (and have) hurt a lot more people at once than Boeing ever could. Did you know some Teslas don't have a manual door release?

[aberdeen5698]

Originally Posted by Migrant_Coconut

Did you know some Teslas don't have a manual door release?

Is that true? I thought the issue was that the manual door release isn't in an obvious position and in normal use people don't need to activate it and therefore are often not aware of its existence. That's what I've read about the Models 3 and Y, anyway.

See: https://lucidowners.com/attachments/screenshot_20220521-202236_rif-is-fun-jpg.2279/

Edit: Or maybe you're talking about the rear doors?

[Migrant_Coconut]

Originally Posted by aberdeen5698

Edit: Or maybe you're talking about the rear doors?

Yup, that's the one - sure won't be fun crawling to the front doors in an emergency (if you can at all), that's for sure. And don't get me started on Ford's plan to make an AV which can repo itself...

[WarrenC12]

Originally Posted by Migrant_Coconut

Meatbags are a lot less forgiving to "blind idiot" algorithms than to other meatbags.

I'm not even sure what that means. The data is already in that self driving cars are less dangerous overall than humans. But when it makes a dumb mistake a "human would never make", people lose their ability to understand statistics.