Intel, BMW and Mobileye To Have Autonomous Test Vehicles On Roads In 2017

[HardOCP News]

BMW Group, Intel and Mobileye today announced that a fleet of approximately 40 autonomous BMW vehicles will be on the roads by the second half of 2017, demonstrating the significant advancements made by the three companies towards fully autonomous driving. Revealing this at a podium discussion held during a joint press conference at CES, the companies further explained that the BMW 7 Series will employ cutting-edge Intel and Mobileye technologies during global trials starting in the U.S. and Europe. This news follows the partnership that was announced between the BMW Group, Intel and Mobileye in July of last year. The companies have since developed a scalable architecture that can be adopted by other automotive developers and carmakers to pursue state of the art designs and create differentiated brands. The offerings scale from individual key integrated modules to a complete end-to-end solution providing a wide range of differentiated consumer experiences.

 

[rudy]

I actually just bought a new car, it doesn't have anything amazing but it does have some safety features. Autonomous breaking, some warnings sensors for blind spots and so on, backup camera with guides, and adaptive cruise control. All I have to say is even now I can see the massive benefits in its infancy. This stuff can't come soon enough to make our roads safer. In my car their are lots of limitations but I have already felt how great it is to allow cruise control to just do what it was always meant to do. And know that if I have a lapse of concentration a system might wake me up or alert me and try to avoid the crash. Right now my system can only track other cars for cruise but as these systems get better they will be able to track everything. Deer come to mind as a major factor here.

Even if you don't like autonomous driving all the hardware and software that is used to get it working directly applies to safety redundancy systems. IE the car knows you are going to crash and warns you and applies breaks for you.

 

[TechLarry]

Right now we have Uninsured Motorist Coverage we are required to have.

Will we soon be forced to pay more for Autonomous Car Fuck-Up Insurance ?

 

[thejokker]

You can't scam investors for money without going through the motions of having a prototype vehicle...

Myself, a co-worker and our boss went on a business trip in the bosses new Honda CR-V with all the bells and whistles. On one level it was kind of impressive that it could follow curves on the interstate and follow a safe distance behind other cars but on the other hand a human driver is more precise. There will be people who will allow this to be a distraction and accidents will occur that would not happen with a manual system with the driver paying full attention.

All software has bugs. Computers crash. Sensors fail. Human beings are a far more robust platform for an automotive application.

 

[rudy]

You can't scam investors for money without going through the motions of having a prototype vehicle...

Myself, a co-worker and our boss went on a business trip in the bosses new Honda CR-V with all the bells and whistles. On one level it was kind of impressive that it could follow curves on the interstate and follow a safe distance behind other cars but on the other hand a human driver is more precise. There will be people who will allow this to be a distraction and accidents will occur that would not happen with a manual system with the driver paying full attention.

All software has bugs. Computers crash. Sensors fail. Human beings are a far more robust platform for an automotive application.

https://crashstats.nhtsa.dot.gov/Api/Public/ViewPublication/812115

Check that document out, in there it says that about 2% of crashes are caused by vehicle errors, this would be your software bugs, computer crashes, sensor errors. Meanwhile humans the "more robust platform" accounts for 94% of crashes. Even if all the automation increased the vehicle-related crashes by 10 fold, raising it to 20% our roads would still be ~75% safer.

https://crashstats.nhtsa.dot.gov/Api/Public/ViewPublication/811059

Now read through that document to get lots of other interesting details. Like did you know the most common maneuver prior to a crash is when they are going straight? And the third most common maneuver is they are stopped in traffic? And did you know the most common time to crash a car is during clear days and daytime lighting? Sounds like human drivers may not be the most robust platform. Give them the best conditions and they still, manage to screw it up.

 

[M76]

You can't scam investors for money without going through the motions of having a prototype vehicle...

Myself, a co-worker and our boss went on a business trip in the bosses new Honda CR-V with all the bells and whistles. On one level it was kind of impressive that it could follow curves on the interstate and follow a safe distance behind other cars but on the other hand a human driver is more precise. There will be people who will allow this to be a distraction and accidents will occur that would not happen with a manual system with the driver paying full attention.

All software has bugs. Computers crash. Sensors fail. Human beings are a far more robust platform for an automotive application.

Wow, talk about confirmation bias. Humans as the most robust platform that's heavy. Guess the mantra "I'm only human, I make mistakes" was wrong all along.

 

[thejokker]

https://crashstats.nhtsa.dot.gov/Api/Public/ViewPublication/812115

Check that document out, in there it says that about 2% of crashes are caused by vehicle errors, this would be your software bugs, computer crashes, sensor errors. Meanwhile humans the "more robust platform" accounts for 94% of crashes. Even if all the automation increased the vehicle-related crashes by 10 fold, raising it to 20% our roads would still be ~75% safer.

https://crashstats.nhtsa.dot.gov/Api/Public/ViewPublication/811059

Now read through that document to get lots of other interesting details. Like did you know the most common maneuver prior to a crash is when they are going straight? And the third most common maneuver is they are stopped in traffic? And did you know the most common time to crash a car is during clear days and daytime lighting? Sounds like human drivers may not be the most robust platform. Give them the best conditions and they still, manage to screw it up.

Invalid conclusions...
A self-driving car will utilize an entirely new system which will dwarf the level of technology currently used in today's cars. An IC that controls whether your outside lights come when a person approaches your front door is much different that your personal computer that runs complex tasks. A self-driving car will require sensors that will need to be very clean at all times in order to function. Driving a car in complex and changing conditions will require software exponentially more complex than anything currently found in a car today. The software required to drive a car may be the most advanced software ever developed and it will never be completely bug-free. Most people are totally under-estimating just how difficult (perhaps impossible) developing the operating system and software required for a self-driving car.

People "are" more robust compared to a self-driving car. A self-driving car will essentially be a computer controlled car. How many times has your computer crashed or locked up due to hardware or software issues? How many times have you fainted or passed out (excluding alcohol or drug related circumstances) or died? What happens when a computer encounters a unique, unanticipated situation? People are far more capable of reacting in an innovated fashion than a programmed device.

I suspect landing a robot on Mars is easier than developing a self-driving car program. We can mathematically anticipate exactly where Mars will be in a future point and time and program an intercepting course with certainty. Driving a car at high speeds in congested conditions requires programming reactions to dynamic and unpredictable situations. How can you program for unanticipated situations?

Do I believe self-driving cars will ever be developed? Yes but it will happen in several stages. The initial self-driving car will be rejected by the public just like google glasses because self driving cars will not be able to integrate transparently with our current infrastructure and driving environment. The initial failure set back development for years. Military research will continue behind the scenes and eventually hardware and software and the traffic infrastructure will mature until it becomes feasible. We are decades away...

 

[thejokker]

Wow, talk about confirmation bias. Humans as the most robust platform that's heavy. Guess the mantra "I'm only human, I make mistakes" was wrong all along.

You are over-estimating computers and under-estimating humans. Computers crash and freeze up. There are computer bugs and computer errors. Malware and hackers. There are reasons why humans control computers and not vice versa.

 

[rudy]

Invalid conclusions...
A self-driving car will utilize an entirely new system which will dwarf the level of technology currently used in today's cars. An IC that controls whether your outside lights come when a person approaches your front door is much different that your personal computer that runs complex tasks. A self-driving car will require sensors that will need to be very clean at all times in order to function. Driving a car in complex and changing conditions will require software exponentially more complex than anything currently found in a car today. The software required to drive a car may be the most advanced software ever developed and it will never be completely bug-free. Most people are totally under-estimating just how difficult (perhaps impossible) developing the operating system and software required for a self-driving car.

People "are" more robust compared to a self-driving car. A self-driving car will essentially be a computer controlled car. How many times has your computer crashed or locked up due to hardware or software issues? How many times have you fainted or passed out (excluding alcohol or drug related circumstances) or died? What happens when a computer encounters a unique, unanticipated situation? People are far more capable of reacting in an innovated fashion than a programmed device.

I suspect landing a robot on Mars is easier than developing a self-driving car program. We can mathematically anticipate exactly where Mars will be in a future point and time and program an intercepting course with certainty. Driving a car at high speeds in congested conditions requires programming reactions to dynamic and unpredictable situations. How can you program for unanticipated situations?

Do I believe self-driving cars will ever be developed? Yes but it will happen in several stages. The initial self-driving car will be rejected by the public just like google glasses because self driving cars will not be able to integrate transparently with our current infrastructure and driving environment. The initial failure set back development for years. Military research will continue behind the scenes and eventually hardware and software and the traffic infrastructure will mature until it becomes feasible. We are decades away...

The sensors that are on cars now are already pretty good and they don't need to be perfectly clean. Even if they do it won't be much work to implement some sort of self cleaning mechanism. This will only improve with time.

Being against autonomous cars causes people to make extreme conclusions. I showed definite data showing that humans are causing the vast majority of accidents. And tons of those accidents are simply a lapse in concentration. Meanwhile you talk about computers crashing and compare it to humans feinting. I think that is an extreme example not based in reality. The software that will run cars will be built to be redundant and stable its a car driving people and protecting their lives not a crappy console port. There are many computers that need to have high uptime that run stable of years. A car wouldn't even need that as it will reboot everytime it is shut down likely. Also none of these systems need to be perfect, another extreme conclusion people draw. As I already pointed out these systems could multiply failures by 10 fold and still beat humans by a large margin. In the end human selfishness will win out and drive the adoption of autonomous cars and subsystems faster than many people think. Selfish motivations include, laziness, the desire to stay safe, cut costs etc.... Even if mainstream consumers reject autonomous cars industry players like the trucking industry will just push it forward without them.

Even in the short term many of the downsides of autonomous systems will be offset by their upsides. A computer system doesn't sleep or get drowsy it doesn't lose concentration. And in the interim autonomous systems will simply detect if they are unable to do a good job and alert the user to maintain control.

 

[thejokker]

The sensors that are on cars now are already pretty good and they don't need to be perfectly clean. Even if they do it won't be much work to implement some sort of self cleaning mechanism. This will only improve with time.

Being against autonomous cars causes people to make extreme conclusions. I showed definite data showing that humans are causing the vast majority of accidents. And tons of those accidents are simply a lapse in concentration. Meanwhile you talk about computers crashing and compare it to humans feinting. I think that is an extreme example not based in reality. The software that will run cars will be built to be redundant and stable its a car driving people and protecting their lives not a crappy console port. There are many computers that need to have high uptime that run stable of years. A car wouldn't even need that as it will reboot everytime it is shut down likely. Also none of these systems need to be perfect, another extreme conclusion people draw. As I already pointed out these systems could multiply failures by 10 fold and still beat humans by a large margin. In the end human selfishness will win out and drive the adoption of autonomous cars and subsystems faster than many people think. Selfish motivations include, laziness, the desire to stay safe, cut costs etc.... Even if mainstream consumers reject autonomous cars industry players like the trucking industry will just push it forward without them.

Even in the short term many of the downsides of autonomous systems will be offset by their upsides. A computer system doesn't sleep or get drowsy it doesn't lose concentration. And in the interim autonomous systems will simply detect if they are unable to do a good job and alert the user to maintain control.

The vast majority of crashes are caused by humans because the vast majority of cars are driven by humans (nearly 100%). If self-driving cars are ever available in sufficient numbers than the data you cite may become more meaningful.

Yes, there are many computers that have run for years without crashing however those computers are not subjected to vibrations and environmental conditions that self-driven cars will be subjected to. Likewise your conclusions about the superiority of self-driving cars to humans is completely unfounded as there are no self-driving cars that can drive under the wide range of conditions humans are capable of today. All of your "assumptions" are unfounded presumptions as there is a total absence of data to support your conclusions. Similarly you cynical appraisal of humans suggests a bias that distorts your conclusions.

A downside of autonomous cars are that they are programmed. Novel situations unanticipated by the programmers will be problematic while a human driver will be able to make a judgement based upon past experiences. While undoubtedly there are many incompetent drivers you seem to be disregarding all the highly skilled human drivers. In those situations where an autonomous system alerts a user to maintain control what happens during the time that the user has to shift his focus from a task at hand to driving? Those few seconds could be fatal.

IMO you are over-estimating computerized cars and under-estimating the majority of human drivers. Self-driving cars will be consumer items and you don't seem to realize that humans may reject buying self-driving cars. The first self-driving cars will "not" be perfect and those cars and their early-adapter owners may find themselves rejected by the majority similar to the rejection of Google-glasses and Glass-holes.

 

[rudy]

The vast majority of crashes are caused by humans because the vast majority of cars are driven by humans (nearly 100%). If self-driving cars are ever available in sufficient numbers than the data you cite may become more meaningful.

Yes, there are many computers that have run for years without crashing however those computers are not subjected to vibrations and environmental conditions that self-driven cars will be subjected to. Likewise your conclusions about the superiority of self-driving cars to humans is completely unfounded as there are no self-driving cars that can drive under the wide range of conditions humans are capable of today. All of your "assumptions" are unfounded presumptions as there is a total absence of data to support your conclusions. Similarly you cynical appraisal of humans suggests a bias that distorts your conclusions.

A downside of autonomous cars are that they are programmed. Novel situations unanticipated by the programmers will be problematic while a human driver will be able to make a judgement based upon past experiences. While undoubtedly there are many incompetent drivers you seem to be disregarding all the highly skilled human drivers. In those situations where an autonomous system alerts a user to maintain control what happens during the time that the user has to shift his focus from a task at hand to driving? Those few seconds could be fatal.

IMO you are over-estimating computerized cars and under-estimating the majority of human drivers. Self-driving cars will be consumer items and you don't seem to realize that humans may reject buying self-driving cars. The first self-driving cars will "not" be perfect and those cars and their early-adapter owners may find themselves rejected by the majority similar to the rejection of Google-glasses and Glass-holes.

If an autonomous car can't drive through a situations it can be programmed to ask for the human to take over and at the very same time the car will be slowing down or stopping if the human does not take over in time, in fact that is exactly where autonomous systems are making their first major inroads, helping drivers who are not capable of breaking in time slow down or stop cars to avoid crashes. Many of the novel situations we encounter on the road are created by human drivers that do not know or do not follow appropriate rules of the road.

You are misinterpreting the data, yes nearly all cars on the roads are driven by humans but the major cause of an accident could easily be a car failure or weather condition, you know those things that humans make excuses about everytime they crash. But the major causes of crashes are not those items. Logic would dictate that if cars can be made as they are today and as reliable as they are that similar reliability may be possible with the electronics and sensors. If a sensor or computer part goes down the driver is asked to take over or the car slows down to a stop and a tow truck may be called.

 

[thejokker]

If an autonomous car can't drive through a situations it can be programmed to ask for the human to take over and at the very same time the car will be slowing down or stopping if the human does not take over in time, in fact that is exactly where autonomous systems are making their first major inroads, helping drivers who are not capable of breaking in time slow down or stop cars to avoid crashes. Many of the novel situations we encounter on the road are created by human drivers that do not know or do not follow appropriate rules of the road.

You are misinterpreting the data, yes nearly all cars on the roads are driven by humans but the major cause of an accident could easily be a car failure or weather condition, you know those things that humans make excuses about everytime they crash. But the major causes of crashes are not those items. Logic would dictate that if cars can be made as they are today and as reliable as they are that similar reliability may be possible with the electronics and sensors. If a sensor or computer part goes down the driver is asked to take over or the car slows down to a stop and a tow truck may be called.

How much computing power will be required to drive a car at a level where it will be considered self-driving? As much power as a smartphone? An iPad? A laptop? More? Will it require storage for upgrades and updates?

As the computing technology required increases than the validity of the analogy to today's car's reliability decreases. It is one thing to create a working prototype but it is another to create a dependable commercial product. Initially that product will be very expensive and not affordable by the common man.

How well with that product integrate into our existing automotive environment? Of course it will be easier to develop a successful product if we simply eliminate all manually controlled vehicles but that would only be possible under totalitarian rule that has no concern about creating a huge underclass of those without cars and an elite class who can afford cars.

There are technical hurdles to over-come which (IMO) are under-estimated by most and political considerations that are not being discussed at all and are far more insidious...