[GTO]

Finally, a sensible article on self-driving cars.

While it's *at least* 10 years away in the USA, you can add another decade or two to that timeline for India, what with our unpredictable roads + driving conditions.

Great news for those of us who live to drive . Things might not be so bright for the newly born though.

Link to full AutoNews Article

Notable excerpts:

But experts say a fully automated vehicle that is 100 percent safe 100 percent of the time and can operate on any street in any weather condition in the U.S. is not right around the corner.

It's a decade or more down the road.

And this software must be able to anticipate nearly every scenario a vehicle can encounter, from inclement weather to a traffic cop's hand signals to a pedestrian darting into traffic.

Infrastructure needs to be improved, from lane markings to traffic signals to bridges -- as well as the vehicle-to-vehicle and vehicle-to-infrastructure communication systems.

^ Wishful thinking if you think we're going to have that in India.

"We as humans have common sense and reasoning powers that we apply, and most of the time, if not always, we do the right thing," said Rajkumar. "Computers, though very powerful, are unfortunately lacking in common sense.

"Self-driving cars can only do what programmers tell them to do. They can't anticipate everything that can happen on the road."

Toyota President Akio Toyoda said fully autonomous driving, which the company calls chauffeur mode, will require lengthy validation.

In order to accomplish this safely, it is estimated that some 14.2 billion kilometers [8.8 billion miles] of testing, including simulation, are required

"With Ferrari, it's almost an obscene concept."
-- FCA CEO Sergio Marchionne

But to do it without a steering wheel and pedals would be "a hard nut to crack." A car with no steering wheel or pedals would make it a Level 5 automated vehicle. That's at least 10 years away

"Robots always work best when they have rules," Peng says. "What if other people are not following the rules? In the truest sense, Level 5 autonomous vehicles will take a very long time."

[desdemona]

If you cannot or do not enjoy driving or are too multi-tasked to do so properly, well the best answer remains, as it has for over a hundred years: public mass transit, or cabs or BRT buses or ride-shares and shuttles, etc, as the city-case may be.

This autonomous-driving technology is awful even in theory, in the ideal: it will just create further incentives for publics and governments to not do the obvious win-win+ win-city-ecologically: public multi-modal transport of a high-enough order. It will make us all happy and reconciled to jams and dead-slowness, that's all, and further captivated by our smart devices and their apps/data plans etc while the automated car potters and plods. Besides killing off drivers', cabbies' and chauffeurs livelihoods.

Just another planned-obsolescence conspiracy by some only-ostensibly futurist tech+car firms, afai am concerned. With many governments in on it, irrationally from a public-interest POV, as well?

Happy then that it will likely NEVER arrive in India viably! Except that, ex-Bombay/Mumbai, we do not have a remotely-adequate public transport system even on the horizon, either! And all the jams and slow-commutes and dirty-air and lawless-roads, etc, etc too. Sigh.

[SmartCat]

Some 80 years ago, people thought we'd be "driving" cars that can also fly. 40 years ago, people thought all commercial flights will be supersonic. 20 years ago, it was about hydrogen fuel cells. These future predictions about fully autonomous cars falls in pretty much the same category.

It's not surprising that fully autonomous cars are mostly being pursued by high-tech companies (Tesla, Google, Apple). The problem is - they see cars as an utility - like a smartphone or a washing machine. They see cars as a solution to a need (transport), and nothing else. Even if they do end up making a fully autonomous vehicle, it will primarily be used by people who take public transport. At best, it will attract people who see a car as an utility.

Anyway, over a period of time, various 'components' of autonomous driving technology will trickle into all cars - but as a primary/secondary safety feature (eg: like how we have radar braking).

[Nav-i-gator]

One signalling and tracking device fitted in every car being sold (maybe, through some mandatory legislation - like we had with catalytic converters in India), can make it possible. An ECU type module which will transmit signals with realtime info on vehicle inputs and possible changes (like braking inputs, clutch press, gear change, steering input or throttle press) which will be received and processed by all cars in the vicinity and allow the cars to manipulate driving accordingly. Maybe even bikes can have such devices fitted too.
However, this will not be fool-proof. Any miscreant can disable the device and go on to create mayhem on roads. What about pedestrians? or sudden events like a falling tree?

Considering all dynamics at play, truly automated self-driven cars will have to cross into AI world, which is, as the article says, very much possible but still a long time to get there.

But there will be numerous technologies coming gradually into cars, coming through researches and tech development related to self-driving cars. Whenever it happens, self driving cars will be a slow, gradual and progressive evolution rather than a sudden introduction. We already have AMT, radar braking, proximity sensors, even smaller and simpler technologies like ACC, rain sensing wipers, park assist, auto headlights, cruise control, voice commands for infotainment etc in mass market cars today. Already a driver is limited to just acceleration, braking and steering the car (the most complex triad).

[GTO]

Originally Posted by desdemona

well the best answer remains, as it has for over a hundred years: public mass transit, or cabs or BRT buses or ride-shares and shuttles, etc, as the city-case may be.

Well said!

Happy then that it will likely NEVER arrive in India viably!

India will probably see flying cars before it sees self-driving ones . Lack of proper roads, driving etiquette & road manners just won't cut it for autonomous cars. Imagine what the software will go through when it sees a truck coming toward it on the wrong side of the highway!!

Originally Posted by smartcat

Even if they do end up making a fully autonomous vehicle, it will primarily be used by people who take public transport. At best, it will attract people who see a car as an utility.

One thing we've missed out on is the cost. If & when self-driving cars do come, how much more will they cost over a regular car? 30 - 40%? How many will be willing to pay that kind of premium?

[abeerbagul]

I have a different opinion.

Technology has shorter adoption cycles nowadays. Also India is not "unique" as many of us like to believe, just that we are playing catch up.

I remember lot of people laughing in the same manner at google maps and navigation systems, saying that in the hodge podge of Indian roads and traffic conditions, navigation will never work.
It is best to ask someone on the road side for directions rather than rely on GPS.

But now almost everyone relies on google maps even in small towns in India. Navigation is not perfect and optimal, but it works 80% of the times, and now it even predicts traffic!

Cruise control was another feature which was regarded as useless just 10 years ago in India, but now so many drivers are demanding and using this on long highway trips, especially combined with AMT and automatic gearboxes.

We have expressways now where autonomous driving can be used on most of the stretches.
E.g. On Mumbai Pune expressway, barring the ghat section, I would like to use autonomous driving because it is really boring. The ghat section is just 15 km which I can easily drive.

I guess a similar situation will be on the Yamuna Expressway also.

Lets not look at past state of infrastructure and assume the same will prevail going forward

[theMandarin]

The referenced article addresses an interesting point that humans react based on common sense which is something difficult to program if not impossible.

Even if it can be programmed as an algorithm, the designer will need to address an exhaustive list of scenarios which will only increase as the system is ported over to different operating regions E.g. from US to India.

I came across an interesting social experiment designed by MIT that gives an idea of these challenges for the people designing these cars and the people “Choosing” to risk travelling in them.

Moral Machine - MIT

The key takeaway for me was that none of the decisions were absolutely right/wrong, but merely had a varying degree of favorability.

[Nav-i-gator]

One key difference with humans and "logic gates" driven cars - we can anticipate dangerous/potentially dangerous situations beforehand and this helps in increasing the time available to respond, while programmed machines would need a "stimulus", an input information in order to calculate and prepare an adequate response. Programmed self driven cars will do well (maybe even better) than humans in the event of an accident when it has just started (~0 second into a collision, when there is a need to limit the impact, prevent car from toppling, minimizing the drag, stabilizing the out of control car and so on) than preventing the accident altogether.

Self driven cars will be sussessful in reality only when all vehicles on road are self-driven only. Maybe designated roads/areas where only self driven cars are allowed to ply. Throw in one driver in a "manual" car with steering in one hand and gear lever in other, and all self-driven cars in the vicinity will have to resort to extremely defensive driving, and whole road ecosystem (harmony, traffic flow) will get disrupted.

[Pratzgh1]

While we can laugh about the predictions done a decade ago, we must embrace the reality that technology has evolved many folds since then. I believe we will have a choice - Automated cars aimed at utility (Driver-less cars) and cars that can be driven without the fear of accidents. Be glad to live in an era where we can rev traditional fuel based engines. The future will witness vehicles that are fast and extremely safe.

[capslock]

Sharing a web site related to this topic called The Moral Machine where a driver-less car must choose the lesser of two evils, such as killing two passengers or five pedestrians.

It is a platform for gathering human perspective on moral decisions made by machine intelligence like self driving cars.

You can try following on this website:

1. Start Judging: What should a self driving car do in case of various scenarios.
2. Design: Be creative and design your own scenarios.

[Skyline_GT]

Completely agree with the article above!
While self driving autonomous cars may arrive eventually in the future they would never be able to completely wipe out the human driven standard cars.
From India's point of view autonomous cars aren't seeing the day of light in our country for a decade or even more. Traffic habits and road discipline being the major hurdle. Not to forget the general paranoia most if not all would have letting a computer take complete control of a moving vehicle!
Another major hurdle would be "dodging" the two-three wheelers plying on our roads. No matter what kind of technology arrives the bikes,scootys and the rickshaws will always be human driven.

The article in a way gives a very positive vibe to the purist "Live to Drive" folks in here including me!

If anyone could answer this- In case of an accident between two autonomous driving vehicles would the car manufacturers of the respective car models involved be held guilty for the property loss or loss of life since no human(owner) was responsible for the mishap?

[groom]

100% Safe, 100% Automated within a specific geographical area (Geofenced) vehicles will likely be there in half a decade. There could be some surprise package during the 2020 Olympics. These could be vehicles like automated pods / taxis.

Among all the manufacturers, Daimler seems far ahead in this race but tight lipped about this unlike the outspoken American OEMs. Google is actually catching up, Tesla is learning the hardway.

"Self driven cars behave as programmers meant to do" --> This is changing, Deep learning and Neural Networks will change the way / already changed the way automotive computing takes place in real time. This will also change the "common sensical" and "humaness" talk about automated driving.This is usual talk that is happening because of logical computing as a basis of algorithm. Neural Network based image and perception management / algorithms have now reached 2% point better than human understanding. I believe this is what will drive automated driving success story and also change the conventional myth around the logic and comparator based thinking.

These algorithms think like humans do due to the way they learn over a period of time.There are many demonstrations of automated bots that help people, or reach a destination is different moving obstacle scenarios or changing contexts successfully just like humans. Same / similar algorithms will enable automated driving.

All such things will pull the commercially available, for end user sale 100% automated car a decade away.

Its coming, watch out. Its really coming.

Originally Posted by Skyline_GT

In case of an accident between two autonomous driving vehicles would the car manufacturers of the respective car models involved be held guilty for the property loss or loss of life since no human(owner) was responsible for the mishap?

For benefit of all, i am attaching the link of Federal Vehicle Automated Driving Policy released just a few weeks ago.

https://www.transportation.gov/AV/fe...september-2016

Policies across is shaping up, there are lot of questions from several entities. But good part is, these are getting answered.

[hybridpetrol]

Originally Posted by GTO

One thing we've missed out on is the cost. If & when self-driving cars do come, how much more will they cost over a regular car? 30 - 40%? How many will be willing to pay that kind of premium?

We are a country where the cars with no airbag / ABS is preferred over the version with these features due to the cost factor. Manuals are preferred over automatics to save the premium & the mileage considerations. Electric cars are completely ignored while the SHVS took off.
While there are other factors such as non-availability of models with airbag / ABS, fun of driving a manual, lack of charging infrastructure for EVs & their horrible designs, cost factor is one major point.
There must be a complete ban on the current type of cars in order to bring in Autonomous vehicles, which is highly unlikely with our infrastructure. While the infrastructure can be improved, there is no point in introducing a technology which does not solve our current traffic issues faced by 1.25 billion people.
Autonomous cars will remain a fancy technology, like android robots washing dishes at kitchen.

[Nissan1180]

Singapore had an ongoing trial of an autonomous taxi service provided by a company called Nutonomy.
People were excited. The test roads were extended to 12 km from the initial 7 km.
And today the car met with the first accident: It collided with a truck while switching lanes. This is Singapore where people follow rules and lanes.
The problem with Autonomous cars is two fold: First, there's always an element of uncertainty. What if due to some weird reason one of the sensors malfunction? If the occupant cannot drive, then he/she won't be able to take control. This would mean you need to have extra sensors to use in case of a failure-and this means they would be very expensive.
Second, if you drive them on public roads, aren't you endangering the safety of everyone else? Aircrafts are tested in a fixed flying region that won't have air traffic. But you need to test cars on public roads because only then the machine learning algorithms get realistic data to optimize the movement response of the vehicle.
I don't see them becoming common anytime soon. The only way they can be included in the transport ecosystem is by creating a separate lane for them, like the BRTS lanes for a bus. But then, what's the point? If you're making a new road to include the autonomous car, it won't get the real world data and hence it won't be able to predict responses fast enough when it is driven on a public road.
I'm not sure how people are addressing these two points.
One of my seniors has a startup in Silicon valley-they make automated shuttles for intra-campus transport in the US universities. That's a good application because the traffic density is low and people generally behave predictably inside college campuses.

[vibbs]

Originally Posted by smartcat

Some 80 years ago, people thought we'd be "driving" cars that can also fly. 40 years ago, people thought all commercial flights will be supersonic. 20 years ago, it was about hydrogen fuel cells. These future predictions about fully autonomous cars falls in pretty much the same category.

Spot on!!

The Hollywood movie triology of Back to the future made in the 80s I believe, shows people in 2015 driving cars that can fly. We are no where near that now.

Although there is a lot of difference between Spielberg's wishful thinking and prediction by Auto industry people, I still believe 10 years is too short a time period for self driven cars being a norm in US.