Monday Evening Commute - Bionic Segways

Good evening

• Arlington offers goodies to get people to join CaBi and wear helmets.
• Technically, Capital Bikeshare isn't a business.  
• One vote against moving the Capital Crescent Trail out of the tunnel. 
• Man who hit teenager walking bike across Alabama Avenue SE and then fled, pleads guilty and is sentenced to 3 years. 
• A majority want a new bridge across the Potomac. It better have a bike/ped lane (of course, there won't be a new bridge, because no one wants to pay for it).
• "Driverless cars, therefore, are poised to trigger a whole new round of pressure to further redesign intersections for the throughput of vehicles above all else."

Comments

[Purple Eagle]

It's a rather useless poll - no particular location is indicated for this "new crossing". I have seen two proposals over the years - one would be on a road connecting the ICC to Dulles, and the other would be somewhere south of Mt. Vernon. Don't think either idea has a strong chance of happening soon.

[contrarian]

I was disappointed at the lack of vision displayed by the GGW commenters. Autonomous cars are coming, probably within a couple of years. They will transform transportation, settlement and land use patterns as much as the advent of the automobile 100 years ago.

What most people don't seem to get is how bad the average driver is. Autonomous cars will have to be hundreds of times more reliable than a human driver before they are accepted -- and they will be. In 20 years the notion that 30,000 people a year are today killed in car crashes will seem as quaint as dying of scurvy.

[JeffB]

Yah, I'm under the impression that after building the ICC and widening 270 Maryland is pretty much going to be prostate for the next 30 years paying off the bonds.

[washcycle]

contrarian, when we have driverless trains, then you can start the clock on driverless cars. But there are a lot of technical and legal issues to overcome before we get there. I'd say we're much more than 20 years away.

[Kolohe]

"contrarian, when we have driverless trains,"

you mean like the MARTA people mover system?

The legal and liability (and frankly, union) issues will indeed keep human operators in fixed guideway transport well after the tech issues are resolved (which is basically now)

[Kolohe]

And, honest question, if CaBi isn't a 'business' what is it, then?

[washcycle]

A publicly owned transit system. Like Metro.

[Kolohe]

Sorry for the trip post, that should be Atl airport, not 'MARTA'.

[washcycle]

you mean like the MARTA people mover system?

No. I don't even know what that is.

I mean like CSX and BNSF trains that go without drivers. That's significantly easier to do.

[washcycle]

OK, an airport people mover is not that complex. It doesn't have to drive around anything. it's basically an elevator that moves horizontally. So, no, that is not what I mean.

[Kolohe]

There is no counterpart to WMATA in the range and scope of functions to what Alta Bike Share, Inc fulfills for Capitol Bikeshare. Would that there were, we might have decent elevators, air conditioning, and track switches among other things, on the Metro system.

[Kolohe]

But I grok the concept of an 'authority' which I would agree CaBi is. Cheers

[Greenbelt]

"prostrate" I think, not "prostate." Although comparing Maryland's transportation spending to a male-only gland than can swell uncomfortably as it ages to the point that simply daily functions get uncomfortable might have some merit.

[contrarian]

People were driving cars for decades before the significant legal and technical issues were overcome. It took about 30 years for the stop sign to be invented and about fifty for the traffic signal. AAA was founded in large part because early motorists had a lot of run-ins with the legal system. The technology was useful, people decided to use it and think about the consequences later.

From what I've seen Google already has an autonomous car that operates better than an average driver. Sure, it's not perfect, just better than an average human. If I were too old to drive and lived in a place where I needed a car, I'd buy one in a heartbeat, and let my estate deal with the consequences. I suspect there's a not-insignificant number of people in that situation, and we're going to start seeing a trickle and then a stream of these cars.

What I wonder is how these cars will be programmed: will they be programmed to follow the law, or to follow the way that people actually drive? Will they obey speed limits even if no one else does? Stop completely at stop signs? Yield to pedestrians at crosswalks? Will it be something the driver will be able to configure?

Will the government regulate the behavior programmed into these things? It might be that, for the first time since the dawn of the automobile, traffic engineers will actually be able to control traffic, rather than merely try to influence it. What will that look like?

[Jim T]

I should think that we will have something akin to computer-assisted driving (e.g. deceleration to avoid a hazard; speed governors; warnings against lane changes) long before we have driverless cars. Insurance companies or toll prices might encourage people in that direction.

Until we have driverless trains and decades of computer-assisted driving, I doubt we will see driverless cars on public roads. They would be an intersting alternative to valet parking.

[Mike]

People who believe things like autonomous cars are a near-future thing generally aren't the technologists actually working on them. Artificial intelligence has been less than a generation away for 60 years, and we still simply don't know how to do it. There have been some cool technology demonstrations in constrained environments, but the public roads are far from constrained. Sure, there are just a few more tweaks needed to go from the tech demo to widespread public use--but those tweaks are the same sort that have been in the works for 60 years...

[Froggie]

Driverless trains already exist. Some airport people-mover systems are fully automated (the only one I've personally been on is the MSP connection to the Hiawatha LRT), as is the LRT portion of Singapore's rail system and SkyTrain in Vancouver.

[washcycle]

Again, I'm not talking about horizontal elevators. I'm talking about trains out on the rail system where they interact with other trains and switch between rails etc...

[contrarian]

"People who believe things like autonomous cars are a near-future thing generally aren't the technologists actually working on them."

According to Google, their autonomous car has already driven over 200,000 miles in the real world without an accident. How many human drivers can say that? Ironically, the one accident their car had was while it was under human control.

The technology is here. I suspect that cost is still an issue -- it's probably still cheaper to hire a human driver than it would cost to outfit a car with the Google driver (if they were selling it). But the cost is bound to come down.

Here's what I consider an interesting question: how would you know if an autonomous car was better or worse than a human? Currently the death rate is about 1.3 deaths per 100 million miles driven. Let's say you put 10,000 of these cars out on the road for a year, they averaged 10,000 miles each for a total of 100 million miles. What number of deaths would convince you that they were significantly different from human drivers? Zero? That could be just luck. One or two? No different. Three? Four? Five? Maybe. What if it was eight, but it was a single accident with a minivan? We're going to need billions of miles of experience to answer that question. People aren't going to wait, they're going to plunge in.

[antibozo]

The Metro trains used to be largely autonomous. The recent fatal accidents forced Metro to put the drivers back in control.

[Mike]

I'm not talking about autonomous systems on limited access tracks, I'm talking about systems that interact with humans. Note that the vast majority of the google car driving is done either on a closed or carefully designated course, or "with occasional human intervention". That means that when they hit the limits of what the car can figure out, a human is responsible for averting disaster. Do you think that in the real world a model of "the car will drive itself but you have to remain just as attentive in case there is a problem" would actually result in anyone looking up from their phone while their car runs over someone?

[antibozo]

Mike, i'm not sure whom you're responding to. My underlying point is that if, even on limited access tracks, an autonomous system could fail catastrophically, leading Metro to largely abandon it, how much more so should we expect general autonomous cars to have a lot of hurdles to surmount?

[Crickey7]

Has no one considered that operator liability lies with the owner now? With robocars the liability will lie with the manufacturer (assuming a properly maintained car). And there will inevitably be accidents. They'll have to price that risk in.

[antibozo]

Crickey7, no doubt they'll include a EULA that indemnifies them, putting all liability back on Captain Dunsel.

[JeffB]

How well will automated vehicles do when let loose in the real world? Well consider this …

Some years ago, while visiting the call center of the company I work for, I became intrigued with an automated mail cart that was used to collect and delivery paperwork around the cubicle farm. It so happened that a technician for the company was there that day laying new "track".

He explained that the cart simply followed an infrared path that they sprayed onto the floor. A particular spray pattern commanded the cart to stop at various locations so the employees could check it.

Being ever so curious I asked if the cart could travel between floors. Not this model he told me. But they did have a model with an additional IR signaling device that could talk to an IR receiver they placed on the elevator bays. This model, he said, could summon the elevator to travel between floors.

Then he laughed and said there were still a few kinks in that one. One day at a company where they were using that model the elevator was placed out of service. I guess part of placing an elevator out of service is turning off the control panel used to summon the elevator. But they didn't know about the mail cart's special IR control panel.

So on it's appointed rounds that day the cart soon came to the elevator. It stopped and signaled for the elevator. Surprisingly the doors actually opened. And our hero merrily rolled itself into the empty shaft and plunged to its death below.

The moral is that its a complex world out there and surprising situations can come up. You need a little judgment sometimes :).

[antibozo]

JeffB, while amusing, that story is prolly embellished. In most elevator systems, the outer doors are opened by the the inner doors on the elevator car, so if the car isn't present, there's no mechanism to open them.

[JeffB]

@antibozo,

Yes I personally don't know if this is true. But it is not unheard of for people to walk into open elevator shafts so it can happen.

[Crickey7]

Yep. Happened in Fairfax about 10 years ago. Modern office building. Tragic.

There is no such thing as a truly fail-safe system.

[washcycle]

What autonomous driving looks like.

Click here

[antibozo]

Crickey7, i remember the story in Fairfax, and it was also fishy. Never heard what really happened, but office building elevators simply cannot open without a car being behind them. Someone has to force them open. There was more to that story than what was reported.