[livetodrive]

Surprised to see so many votes for 'Yes'. The most electronic gadget which we are playing with everytime and everyday is our phone and we can see the number of issues with such a device, either software or hardware. Till it is stopping us to talk, spoiling our games or getting hot we do not bother. But will we be alive to press the reset once our car behaves abnormally?

There are several parameters which human brain control while driving through an 'exception'. Sometimes we are lucky to handle that fatal error depending on our state of mind and also road and car conditions. Will a computer software be able to handle that exception at all?

Just take the example of the court room scene from the movie "Sully". Such nicely it has been demonstrated that how simulation failed compared to human intelligence to take a quick decision which looked bizarre, but the pilot could take it because of his skill and confidence.

Why we are not scared when the plane is flying in auto? Because we know at 40,000 feet there is almost no chance of something going wrong. And in other times, i.e. take off and landing, we are aware that our pilot has taken control.

Can the same thing happen with cars? Will the software and hardware along with mechanical engineering deliver the same impulse response to a certain circumstance for two different makers of cars ? Or, even for the same maker, will the response vary across different segments of cars ?

For a Maruti I know I have to hit the brake hard, for a Hyundai it will be less. That "machine learning" happens to a driver in a few meters of driving. How about autonomous cars? Will we not doubt that whether the car is fed with enough data points to resolve a non-linear or multidimensional situation?

Atleast for now, till lot of test drives and results, my vote will remain 'NO'

[for_cars1]

At some point in the future, all vehicles will be driven autonomously. That is a certainty whether we like it or not.

Over time, the infrastructure to support autonomous vehicles will get standardised as well as the common protocol to be followed by all autonomous vehicles.

There is bound to be electronic communication between autonomous vehicles that are in proximity and the protocol (well defined set of rules and behavior to be adhered to) will bring in the consistent predictability in the actions of autonomous vehicles unlike human actions which is very individual specific (driver's skills, mood/emotions, attentiveness and alertness levels, drugs/alcohol influence, physical and mental stress levels).

Ofcourse, there will be accidents whenever the autonomous vehicles are faced with a totally new scenario which is not defined in the protocol. Such new events will be a feedback to strengthen the protocol making the entire system more and more resilient over time and making it almost foolproof to regular use cases.
There will always be security/hacking loopholes to deal with in technology. Its a matter of staying on top and keeping the autonomous protocol secure and foolproof and will be a continuous process.

In addition, with vehicles becoming autonomous and having the ability to go to a user, the very concept of a personal car will disappear and that itself will significantly reduce the number of vehicles on the road with reduced accident rates.

[speedmiester]

One way to force people to use autonomous features/cars is by way of offering discount on insurance premiums as this article indicates.
Directline offers 5% discount on insurance premiums if drivers use autopilot feature in their Tesla.

British insurer Direct Line offers Tesla Autopilot discount

Britain's largest motor insurer Direct Line is offering Tesla Inc drivers in Britain a 5-percent discount for switching on the car's Autopilot system, seeking to encourage use of a system it hopes will cut down on accidents.

[clevermax]

Surprised to see the higher percentage of "Yes" voters. I too voted "Yes".

The way present day Tesla is able to predict crashes brings in a lot of confidence on the R&D that is happening in the autonomous driving area. There's no reason not to believe that autonomous driving and AI will surpass human vision, response time and judgment to avoid accidents that happen due to human error.

On the lighter side, it will be funny to put Tesla's crash prediction to test on Indian roads. It will keep beeping all the time

[King_pin09]

I firmly believe that the major causes for road accidents are,

1. Arrogance,
2. Ego,
3. Unwillingness to co-operate with other road users,
4. Unwillingness to follow traffic rules,
5. Driving under influence (alcohol, tension, stress, day dreaming, hurry, et-al),

All the above factors are more of a form of psychological imbalance/ disorder and do not kick in when it comes to autonomous cars. Cars that are programmed to read road signage and follow specific measures are lesser prone to end up in an accident.
However there are other several factors that lead to accidents and are totally circumstantial in nature,

1. A tree branch falling on the vehicle,
2. A huge pothole, an unmarked speed breaker that has come up overnight, or a large boulder left over by trucker, at unwarranted locations such as expressway (esp during nights),
3. Land slides, cave-in, etc,
4. Tyre burst at speed,
5. Vehicle related issues,

The above are far from driver errors and cannot be factored by autonomous vehicles. At the same time an experienced and alert driver can foresee danger in advance and initiate measures to avert such hazard sufficiently in advance. Thus, it is not possible to avert accidents in entirety; however the extent which they occur can be reduced substantially by implementing autonomous vehicles.

The best way to reduce accidents to the maximum possible extent is to subject each individual seeking license through stringent tests (involving driving skill, eye sight, color blindness, reflexes, understanding signboards, psychological tests, et-al), maintaining qualifying criteria of the highest standards and issue permits only to those who satisfactorily pass through each of them based on their own abilities.

[aerohit]

Going by the trend of Indian government, one accident is enough to put a ban on all autonomous cars. They dont have concept of pragmatism.

(Student died, entire school closed. Baby died, entire hospital closed.

Actually, by above logic, entire govt should be banned for not fixing pot hole and causing death)

[YaeJay]

Voted No. Maybe in the future. Still think human brain is faster with all kinds of chaos that we see on the road these days. It would perhaps, work well in a controlled environment. And what would the automated cars do in case there is a situation where it has to take a chance to save 'itself'? What kind of rules would it follow?

[ButterRider 7]

Voted for yes. People make mistakes. They’re prone to poor judgment and are easily distracted. While a self-driving car may not be perfect, it promises to be less flawed than a human driver. This could lead to the biggest potential benefit of the widespread adoption of self-driving cars: a massive reduction in traffic accidents, injuries, and deaths.

Again, autonomous cars aren’t perfect, but the computers they use to calculate things like stopping distance and proximity to other vehicles are more reliable than a human. And, unlike people, computers can’t be distracted by other passengers, cell phones or the radio.

[SS-Traveller]

Originally Posted by Varun_HexaGuy

I was reading a newspaper article a week ago which caught my attention. The article was about the road accidents. As per this article, human error has got to play a major role in the road accidents and Human error also accounts for 90% of the road accidents.

I like your question - but I don't think there is any clarity on Yes/No as the right answer. Hence voted for the "Can't say" option.

Our first concern is about the road accident statistics in India. If we say 90% of crashes in India are due to human error, we would probably be wrong. Human error is not the same as human ignorance. What we see in India is unskilled, untrained drivers crashing cars at 3-digit speeds, which is far different from the NHTSA study, where distracted driving causes crashes.

The article also says that autonomous cars can help decrease the rate of road accidents world-wide by a huge margin. Now my question to the fellow BHPians is that do you think that autonomous cars can really help decrease the rate of accidents world-wide?

The thought behind autonomous cars is to eliminate crashes due to distracted drivers. The first step that was taken in developed countries was to build in a number of driver assistance technologies to
(1) prevent drivers being distracted at the wheel;
(2) detect and alert distracted drivers to return their focus to the task of driving; and
(3) control the vehicle and prevent it from crashing when the driver is not alert or focussed on driving (applicable to short distances on open roads only).

So we've all heard of ADAS (Advanced Driver Assistance Systems) - inventions like AEB (autonomous emergency braking), lane-keeping assistance, blind-spot detection, ACC (adaptive cruise control) and AII (alcohol ignition interlock). These are the systems that address the points (1) and (2) above. Yet, recent studies point out that ADAS do not perform consistently, especially in real-word scenarios.

Asking this because we've heard quite some accidents which have been caused because of autonomous cars (a couple of links attached for reference below) and that kind of accidents could've been avoided with human intervention.

Right - so we have not got to that point anywhere in the world, where zero human input is required to drive a car from Point A to Point B. Forget cars, we do not have trains and aircraft yet, which do not require a trained and alert operator at the controls when the train / aircraft is in motion (even on autopilot, an aircraft cockpit is never empty of pilots AFAIK).

Also given that autonomous cars aren't coming to India, what's the other way around to decrease the road accidents in India (if you think that autonomous cars can really decrease the rate of road-accidents) since India also accounts for a fair share of road accidents world-wide.

This last question is a really tough one to answer, but I am sure increasing penalties and lowering speed limits is not the way to go. Re-training and re-licensing our drivers is an unsurmountable task, and the government is not thinking about the long-term fallout of gifting driving licences to people after 15 minutes (or less) of testing at speeds below 20 kmph.

Autonomous cars need to be introduced worldwide first, they must run without glitches for a few years, have their cost reduced, and only then should India even think of introducing autonomous cars on the roads. That time period should easily see you to my age, and I probably would not live to see that day.

[shancz]

Sorry for waking up an old thread but just doing as mentioned.
Voted Yes.

If and when implemented it will do the following :

• reduce nuisance caused by jokers with a license(can't call them drivers) who don't even know driving manners let alone driving rules
• bring order to the chaos which we face everyday
• drivers following the rules will finally feel welcome rather than threatened

Will this be implemented here ? Sadly not in the next 20 years at least to have any credible effect(not talking about pilot projects).
So driving education and enforcement is the way to go.

[gkveda]

I voted NO

Simple reason is, the people who walk on roads, cattles who jump out of nowhere, roads which creates craters on their own are not autonomous

Also, the sensors installed in the cars for autonomous driving are NOT autonomously manufactured.

Thirdly, the softwares written for autonomous driving is not written by autonomous bodies.

When all these are manufactured or developed by humans, they are all error prone.

Just an analogy. The 1st gen windows OS (3.1) was written in 1980s but even today,(almost 40 years) although there are multiple versions are released and several features are added, it still crashes and needs reloading of OS once in a while.

So, how can we expect softwares to be written error free in next 3-4 years.

[Jeroen]

Originally Posted by gkveda

I.

When all these are manufactured or developed by humans, they are all error prone.

Just an analogy. The 1st gen windows OS (3.1) was written in 1980s but even today,(almost 40 years) although there are multiple versions are released and several features are added, it still crashes and needs reloading of OS once in a while.

So, how can we expect softwares to be written error free in next 3-4 years.

Why should these automated systems be 100% error free? They only need to outperform humans. They are far from error free.

If we can get the automation to make half the mistakes humans do, we reduce death, accidents and mayhem by 50% already.

There are no error free systems. Yet, we have many industries whom rely heavily on automation to improve their safety record tremendously. E.g. aviation, railroads, power generation etc. Not a single one of them runs on 100% error free software or 100% error free systems. But they are still doing a heck of a better job than with humans at the controls.

The only way not to have an accident caused by an auto-pilot is not to have an autopilot. But when that autopilot does a considerable better job then the human, it’s time to switch. The number of human caused accidents will drop dramaticallly and we will be left with a few, very unfortunate, auto pilot related accidents.

Jeroen

Jeroen