Samsung said it would invest heavily in four key areas through 2020. Auto tech (1), artificial intelligence(2) and new fifth-generation, or 5G, cellular technology (3) [...] will draw funding, as will its nascent drug companies specializing in contract manufacturing and biosimilar medications (4)
The summary is correct, even if the phrasing is tortuous
I expect the kind of people that replace their $1000 phone every two years to earn enough to afford the phone, as well as some new, more expensive cars. They will never need to switch to a 2008 flip phone to save money.
You're conflating here several systems, all coming under the label of "self-drive".
Tesla's Autopilot is level 2 because it requires constant supervision and Tesla never claimed it's perfect - they expect some corner cases where, for instance, it will drift and hit concrete barriers. It might not even be the best level 2 system out there - Cadillac, Audi, Mercedes, Nissan, Volvo have similar systems, it's only that their marketing doesn't call them "self-drive"
Uber's system is in early development, eventually it should be level 4 - no driver required in a limited, mapped area - but it's not there yet, that's why they have safety drivers. The accident happened because both the automatic system and the safety driver backup did not work.
It's wrong to judge the performance and safety of autonomous cars by relating to accidents caused by those projects. You will find out that Teslas are not truly self-driving, that Uber is far behind its rivals, and that lone safety drivers get bored and aren't really good at it. But it doesn't tell you how good Waymo's cars are, for example.
True, but what's stopping another company to rush another poorly developed prototype onto public roads? Even more, how do you differentiate between a poorly developed prototype and a working production-grade system? Where do you draw the line?
The Arizona crash should be a call for authorities to create metrics of maturity for a self driving car program. They can be millions of miles driven on urban roads, number of incidents, or time between safety-related disconnects of the system. Build upon what Waymo and Cruise has discovered. Maybe Waymo doesn't need them anymore, its systems are mature enough, but other wannabe companies should pass the same gates.
If you cared to RTFA, those "idiots" begin with recapping how their last year predictions turned out.
And yes, it's a lightweight article in a tech magazine to be read as entertainment on a lazy holiday evening. Not a serious analysis with billion dollar decisions at stake. Chill.
There is something artificial and Orwellian about metric, Its a synthetic system, a poor fit for everyday use, while customary measures feel more organic for everyday use. George Orwell mentioned this in relation to metric beer in 1984.
No, it's just about the system you learned when you were young. It's not more "organic" or anything, it's just custom and tradition.
It's as easy an natural for me, living in continental Europe, to say I got a 120 cm TV as it is for you to say you have a 48' one. Celsius is pretty good for the weather too - I know that if it's sub-zero, then roads are slippery and pipes are freezing.
It's simply much more difficult for you to make unit conversions than it is for me.
Germans are cash loving people, but also they have their own debit card network - EC - which is completely separate from MasterCard and Visa. Most stores don't take credit cards, not even Visa/MC credit cards, but on the other hand almoste everyone takes EC.
It is actually rather straightforward to write some code that transfers your playlist and preferred seat position from one car to another. I worked for a such a project a few years ago - the end customer being a car sharing service.
You probably need more than a single decent GPU just to run the self driving algorithms. Also take into account that these are safety critical systems - so you don't provision to run at 100% capacity. The code will probably spend 25% of CPU time doing.the math and 75% doing checks that everything's ok.
Moreover, they don't put a single unit in a car for safety reasons, we're talking here either 2-out-of-3 voting systems, or a primary system to do computation, a mirror system to verify it, and a 3rd hot standby for fallback if something goes wrong. In any case, you need 3x the power just to ensure very high reliability.
This way code than needs 200W on a desktop GPU during development will require 1-2 kW in a safety critical production system.
Then don't step into that elevator! Take the stairs!
Joking aside, nobody steps into an elevator thinking "I'm locked in a box dangling on top of a deep shaft - I put my physical safety into the strength of a cable". They trust the machine.
Same when flying across the ocean, or driving at speed down the motorway. Planes and cars are orders of magnitude more complex and they can easily kill their passengers if anything goes wrong. Yet people trust them because they work just fine the vast majority of the time.
Safe driving cars will be the same. Yet another layer of complexity that will eventually work fine and people will get used to. IMO it will happen rather quickly.
I sent your questions to Google and the answers were spot-on...
"How many cylinders in a V6 engine?"
An engine of a car with six cylinders is called a V6 engine
"How many doughnughts are in a dozen doougnughts?"
At many bakeries, though, the fun doesn't end with 12 donuts. Instead, a lot of bakeries will give you an extra donut to create what's known as a "baker's dozen." That's 13 amazing donuts for the price of 12.
"What is the nominal size of a 2X4 board?"
2x4s are not actually 2 inches by 4 inches. When the board is first rough sawn from the log, it is a true 2x4, but the drying process and planing of the board reduce it to the finished 1.5x3.5 size
"What time is it on Mars?"
Listen to Nagin Cox's TED Talk: "What time is it on Mars?" : As a spacecraft engineer at NASA's Jet Propulsion Laboratory, Cox works on the team that manages the United States' rovers on Mars. But working a 9-to-5 on another planet -- whose day is 40 minutes longer than Earth's -- has particular, often comical challenges.
Most people I know don't drive for 16h without taking a break. Come to think of it, it's highly dangerous, and actually illegal if you're a professional driver.
A more realistic scenario is doing, say, 1000km in a day. With an ICE car, that will take 10 hrs driving; take a few breaks and we are talking at least 11 hrs in total. With an EV , you stop 3 times every 250km for a recharge. With a fast charger, that may take about 30 minutes each stop , 1 hour and a half in total. So maybe a 12 hr trip? It's not much of a difference.
What's missing is the infrastructure. When there will be a network of fast chargers everywhere, we would have similar travel times for ICEs and EVs doing long trips.
I bought a $30K car while my previous one was $15K. But the new car is bigger, accelerates faster, is quieter, has a prettier color, has a sun roof, navigates using Google Maps, the brand is better perceived by people I meet... there are a hundred benefits that overall made me feel it's worth the extra dollars and cents.
However when I decided between gas, diesel, or plug-in hybrid, I did the math and counted the dollars. I didn't feel going hybrid or EV is worth anything but saving money.
People who buy EVs have to accept the inconvenience of long charging and short range, and for most of us, there are not enough benefits of going electric to compensate for the cost and troubles. EVs cost now a lot more than ICEs, and what do you get for the extra money? Silent drive, extra torque, a warm fuzzy feeling of running a cleaner car? how many people care about that?
True, they got the Pilot Assist, which is pretty good at keeping the lane at any speed. But if you check the manual it says it's semi-autonomous:
Warning Pilot Assist is an aid which cannot handle all traffic, weather and road conditions. The driver must always be observant with regard to the prevailing traffic conditions and intervene when Pilot Assist is not maintaining a suitable speed or suitable distance. [...] The driver always bears responsibility for how the car is controlled as well as for maintaining the correct distance and speed, even when Pilot Assist is being used.
So you can't really take your eyes off the road; while in the Audi you can - as long as the car is slow enough.
Watch the movie in TFA and at 5:35 you get a slide with the processors involved. There's MobilEye (now Intel) that does image recognition, Altera (now Intel too) that does sensor fusion, Nvidia that does camera processing, and Infineon Aurix that does the Traffic Jam Pilot. The Aurix is the weakest of all four, but it's the only one considered safe enough to drive a vehicle (ISO26262 ASIL-D and similar standards)
Then you understand why the system is so limited. It can't change lanes. It doesn't overtake other cars. It does not plan ahead your way home and takes the second exit if needed. All it does is drive more or less in a straight line following the car in front, and stop if the radar detects an obstacle. There's no AI or image recognition there, just classical control algorithms that can run on a microcontroller such as the Aurix.
All I can think of is that they haven't figured out yet how to make the AI required for real self driving safe, so they can assume responsibility if anything goes wrong.
Audi is the first manufacturer to assume responsibility for any accidents that happen while the car is in piloted mode. That's what Level 3 means. No one else, not even Tesla, is here yet.
The first time you read the title, it seems like Volvo will only sell electric cars in 2 years time and they will become another Tesla.
Then, if you pay attention, it says that starting 2019, all new models will have an electric engine in them. Yes, this includes mild hybrids, basically energy recovery systems where the electric engine only gives a boost, but it's too small to drive on electric power alone. And yes, they will keep producing the old models for a while.
This is good news, but by no means earth-shattering. I understand most of the European manufacturers will introduce mild hybrids across their range, due to very strict emission standards coming 2020. A PR coup for Volvo, for making public a decision that everyone in the industry will eventually take, and soon.
That's a good point. On the other hand I looked up who those "Romanian carriers" are.
You've got Orange, headquartered in France, 7 EU countries. Then Vodafone, HQ in UK, present in 25 EU countries, almost everywhere, Then Telekom from Germany, with 6 networks in the EU. Last one is a local carrier who has just gone over the border in Hungary.
That means a Romanian traveling to Germany, for example, has a great chance to use the very same company network but pay much more for it.
Actually it makes sense for carriers to share the same infrastructure for different countries - from core networks to call centers, it makes sense to have them big and build economies of scale. They would probably keep all but the radio access network in one place, if it weren't for regulations like having to provide legal intercept to local law enforcement.
I think this evolution will be much faster in Europe. Hybrid cars are fairly common here and if, as you mention, it were possible to charge a pluggable hybrid up in 10 minutes and get 100 km out of it at a speed of around 100 km/h I'd definitely buy one. In fact the last time I was in the market for a car the only reason I did not buy a hybrid was that no pluggable ones were available. I'd do about 80-90% of my driving, commuting to work and running errands in electric mode on a car like that. If electric cars could be had with a charge time of 10 minutes and a subsequent range of ~3-400 km and if the charge station infrastructure was in place I'd buy one in a heartbeat and never look back.
Let's do some math here. A car consumes ~20 kW at 100km/h . It takes 1h to go 100 km, that means the total energy consumed is 20 kWh. To get this kind of charge back in 10 minutes (1/6 of an hour), you need to charge it at 6 x 20 = 120 kW . That's possible now with a Tesla supercharger. On the other hand, if you want to go 3-400 km, then you need to charge it at 360 to 500 kW. I won't say it's impossible, but I'd say it's quite hard to build such a high power charging system for a car.
BTW, most plug-in hybrids today don't support fast charging; that would make them too expensive. The typical usage pattern is to leave them charged overnight and wake up with just enough juice for the daily commute. Not unlike how we use phones, to think of it.
Deregulation - or lack of regulation at the right time. In the late '90s and early 2000's the cable TV and Internet market was a wild wild West with thousands of small companies fighting for users. No rules, cables stretched everywhere across streets and buildings. They consolidated now in a few big players and the cables got buried underground, but there is still competition - the apartment I lived in Bucharest had three cables from three different providers passing in front of the door.
It also helps that most people leave in densely populated apartment buildings in city centers, instead of sprawling suburbia.
But they won't lose all their jobs at once. First commercially available self driving trucks will appear no earlier than 2020. They'll probably only be certified for highway driving when the weather is fine. They will be expensive and they'll be used by a few companies in pilot projects. They won't replace the driver but extend the hours while the vehicle is moving - the ex-driver can sleep while the truck is driving itself on the interstate.
Slowly they become popular across big rigs but for a long time they'll still need someone on board, even if not actively driving. That's for security or for negotiating town streets. We are talking 20-30 years until those millions of jobs will be lost. There's plenty of time for those people to retire or move on.
Samsung said it would invest heavily in four key areas through 2020. Auto tech (1), artificial intelligence(2) and new fifth-generation, or 5G, cellular technology (3) [...] will draw funding, as will its nascent drug companies specializing in contract manufacturing and biosimilar medications (4)
The summary is correct, even if the phrasing is tortuous
True. Also a tent is much cheaper than a house.
I expect the kind of people that replace their $1000 phone every two years to earn enough to afford the phone, as well as some new, more expensive cars. They will never need to switch to a 2008 flip phone to save money.
You're conflating here several systems, all coming under the label of "self-drive".
Tesla's Autopilot is level 2 because it requires constant supervision and Tesla never claimed it's perfect - they expect some corner cases where, for instance, it will drift and hit concrete barriers. It might not even be the best level 2 system out there - Cadillac, Audi, Mercedes, Nissan, Volvo have similar systems, it's only that their marketing doesn't call them "self-drive"
Uber's system is in early development, eventually it should be level 4 - no driver required in a limited, mapped area - but it's not there yet, that's why they have safety drivers. The accident happened because both the automatic system and the safety driver backup did not work.
It's wrong to judge the performance and safety of autonomous cars by relating to accidents caused by those projects. You will find out that Teslas are not truly self-driving, that Uber is far behind its rivals, and that lone safety drivers get bored and aren't really good at it. But it doesn't tell you how good Waymo's cars are, for example.
True, but what's stopping another company to rush another poorly developed prototype onto public roads? Even more, how do you differentiate between a poorly developed prototype and a working production-grade system? Where do you draw the line?
The Arizona crash should be a call for authorities to create metrics of maturity for a self driving car program. They can be millions of miles driven on urban roads, number of incidents, or time between safety-related disconnects of the system. Build upon what Waymo and Cruise has discovered. Maybe Waymo doesn't need them anymore, its systems are mature enough, but other wannabe companies should pass the same gates.
It looks like it was an one-off event where the clocks ran behind during the course of a day or so.
I've been away from home for a week, and when I returned the clocks from my ovens (Samsung and Electrolux) were both 5 minutes behind.
This explains it! I did scratch my head for a while trying to figure out how this could happen.
If you cared to RTFA, those "idiots" begin with recapping how their last year predictions turned out.
And yes, it's a lightweight article in a tech magazine to be read as entertainment on a lazy holiday evening. Not a serious analysis with billion dollar decisions at stake. Chill.
There is something artificial and Orwellian about metric, Its a synthetic system, a poor fit for everyday use, while customary measures feel more organic for everyday use. George Orwell mentioned this in relation to metric beer in 1984.
No, it's just about the system you learned when you were young. It's not more "organic" or anything, it's just custom and tradition.
It's as easy an natural for me, living in continental Europe, to say I got a 120 cm TV as it is for you to say you have a 48' one. Celsius is pretty good for the weather too - I know that if it's sub-zero, then roads are slippery and pipes are freezing.
It's simply much more difficult for you to make unit conversions than it is for me.
Germans are cash loving people, but also they have their own debit card network - EC - which is completely separate from MasterCard and Visa. Most stores don't take credit cards, not even Visa/MC credit cards, but on the other hand almoste everyone takes EC.
I'm all you with you, but Tesla Autopilot doesn't let you do any of those things, yet. You still have to pay attention to the traffic.
It is actually rather straightforward to write some code that transfers your playlist and preferred seat position from one car to another. I worked for a such a project a few years ago - the end customer being a car sharing service.
They don't take 2.5% from you, they take it from the merchants where you buy things.
You probably need more than a single decent GPU just to run the self driving algorithms. Also take into account that these are safety critical systems - so you don't provision to run at 100% capacity. The code will probably spend 25% of CPU time doing.the math and 75% doing checks that everything's ok.
Moreover, they don't put a single unit in a car for safety reasons, we're talking here either 2-out-of-3 voting systems, or a primary system to do computation, a mirror system to verify it, and a 3rd hot standby for fallback if something goes wrong. In any case, you need 3x the power just to ensure very high reliability.
This way code than needs 200W on a desktop GPU during development will require 1-2 kW in a safety critical production system.
Then don't step into that elevator! Take the stairs!
Joking aside, nobody steps into an elevator thinking "I'm locked in a box dangling on top of a deep shaft - I put my physical safety into the strength of a cable". They trust the machine.
Same when flying across the ocean, or driving at speed down the motorway. Planes and cars are orders of magnitude more complex and they can easily kill their passengers if anything goes wrong. Yet people trust them because they work just fine the vast majority of the time.
Safe driving cars will be the same. Yet another layer of complexity that will eventually work fine and people will get used to. IMO it will happen rather quickly.
I sent your questions to Google and the answers were spot-on...
"How many cylinders in a V6 engine?"
An engine of a car with six cylinders is called a V6 engine
"How many doughnughts are in a dozen doougnughts?"
At many bakeries, though, the fun doesn't end with 12 donuts. Instead, a lot of bakeries will give you an extra donut to create what's known as a "baker's dozen." That's 13 amazing donuts for the price of 12.
"What is the nominal size of a 2X4 board?"
2x4s are not actually 2 inches by 4 inches. When the board is first rough sawn from the log, it is a true 2x4, but the drying process and planing of the board reduce it to the finished 1.5x3.5 size
"What time is it on Mars?"
Listen to Nagin Cox's TED Talk: "What time is it on Mars?" : As a spacecraft engineer at NASA's Jet Propulsion Laboratory, Cox works on the team that manages the United States' rovers on Mars. But working a 9-to-5 on another planet -- whose day is 40 minutes longer than Earth's -- has particular, often comical challenges.
That feels pretty intelligent for me.
Most people I know don't drive for 16h without taking a break. Come to think of it, it's highly dangerous, and actually illegal if you're a professional driver.
A more realistic scenario is doing, say, 1000km in a day. With an ICE car, that will take 10 hrs driving; take a few breaks and we are talking at least 11 hrs in total.
With an EV , you stop 3 times every 250km for a recharge. With a fast charger, that may take about 30 minutes each stop , 1 hour and a half in total. So maybe a 12 hr trip? It's not much of a difference.
What's missing is the infrastructure. When there will be a network of fast chargers everywhere, we would have similar travel times for ICEs and EVs doing long trips.
I bought a $30K car while my previous one was $15K. But the new car is bigger, accelerates faster, is quieter, has a prettier color, has a sun roof, navigates using Google Maps, the brand is better perceived by people I meet... there are a hundred benefits that overall made me feel it's worth the extra dollars and cents.
However when I decided between gas, diesel, or plug-in hybrid, I did the math and counted the dollars. I didn't feel going hybrid or EV is worth anything but saving money.
People who buy EVs have to accept the inconvenience of long charging and short range, and for most of us, there are not enough benefits of going electric to compensate for the cost and troubles. EVs cost now a lot more than ICEs, and what do you get for the extra money? Silent drive, extra torque, a warm fuzzy feeling of running a cleaner car? how many people care about that?
True, they got the Pilot Assist, which is pretty good at keeping the lane at any speed. But if you check the manual it says it's semi-autonomous:
Warning
Pilot Assist is an aid which cannot handle all traffic, weather and road conditions.
The driver must always be observant with regard to the prevailing traffic conditions and intervene when Pilot Assist is not maintaining a suitable speed or suitable distance.
[...]
The driver always bears responsibility for how the car is controlled as well as for maintaining the correct distance and speed, even when Pilot Assist is being used.
So you can't really take your eyes off the road; while in the Audi you can - as long as the car is slow enough.
Watch the movie in TFA and at 5:35 you get a slide with the processors involved. There's MobilEye (now Intel) that does image recognition, Altera (now Intel too) that does sensor fusion, Nvidia that does camera processing, and Infineon Aurix that does the Traffic Jam Pilot. The Aurix is the weakest of all four, but it's the only one considered safe enough to drive a vehicle (ISO26262 ASIL-D and similar standards)
Then you understand why the system is so limited. It can't change lanes. It doesn't overtake other cars. It does not plan ahead your way home and takes the second exit if needed. All it does is drive more or less in a straight line following the car in front, and stop if the radar detects an obstacle. There's no AI or image recognition there, just classical control algorithms that can run on a microcontroller such as the Aurix.
All I can think of is that they haven't figured out yet how to make the AI required for real self driving safe, so they can assume responsibility if anything goes wrong.
Name one such production model please.
Audi is the first manufacturer to assume responsibility for any accidents that happen while the car is in piloted mode. That's what Level 3 means. No one else, not even Tesla, is here yet.
The first time you read the title, it seems like Volvo will only sell electric cars in 2 years time and they will become another Tesla.
Then, if you pay attention, it says that starting 2019, all new models will have an electric engine in them. Yes, this includes mild hybrids, basically energy recovery systems where the electric engine only gives a boost, but it's too small to drive on electric power alone. And yes, they will keep producing the old models for a while.
This is good news, but by no means earth-shattering. I understand most of the European manufacturers will introduce mild hybrids across their range, due to very strict emission standards coming 2020. A PR coup for Volvo, for making public a decision that everyone in the industry will eventually take, and soon.
That's a good point. On the other hand I looked up who those "Romanian carriers" are.
You've got Orange, headquartered in France, 7 EU countries. Then Vodafone, HQ in UK, present in 25 EU countries, almost everywhere, Then Telekom from Germany, with 6 networks in the EU. Last one is a local carrier who has just gone over the border in Hungary.
That means a Romanian traveling to Germany, for example, has a great chance to use the very same company network but pay much more for it.
Actually it makes sense for carriers to share the same infrastructure for different countries - from core networks to call centers, it makes sense to have them big and build economies of scale. They would probably keep all but the radio access network in one place, if it weren't for regulations like having to provide legal intercept to local law enforcement.
I think this evolution will be much faster in Europe. Hybrid cars are fairly common here and if, as you mention, it were possible to charge a pluggable hybrid up in 10 minutes and get 100 km out of it at a speed of around 100 km/h I'd definitely buy one. In fact the last time I was in the market for a car the only reason I did not buy a hybrid was that no pluggable ones were available. I'd do about 80-90% of my driving, commuting to work and running errands in electric mode on a car like that. If electric cars could be had with a charge time of 10 minutes and a subsequent range of ~3-400 km and if the charge station infrastructure was in place I'd buy one in a heartbeat and never look back.
Let's do some math here.
A car consumes ~20 kW at 100km/h . It takes 1h to go 100 km, that means the total energy consumed is 20 kWh. To get this kind of charge back in 10 minutes (1/6 of an hour), you need to charge it at 6 x 20 = 120 kW . That's possible now with a Tesla supercharger.
On the other hand, if you want to go 3-400 km, then you need to charge it at 360 to 500 kW. I won't say it's impossible, but I'd say it's quite hard to build such a high power charging system for a car.
BTW, most plug-in hybrids today don't support fast charging; that would make them too expensive. The typical usage pattern is to leave them charged overnight and wake up with just enough juice for the daily commute. Not unlike how we use phones, to think of it.
Deregulation - or lack of regulation at the right time. In the late '90s and early 2000's the cable TV and Internet market was a wild wild West with thousands of small companies fighting for users. No rules, cables stretched everywhere across streets and buildings. They consolidated now in a few big players and the cables got buried underground, but there is still competition - the apartment I lived in Bucharest had three cables from three different providers passing in front of the door.
It also helps that most people leave in densely populated apartment buildings in city centers, instead of sprawling suburbia.
But they won't lose all their jobs at once. First commercially available self driving trucks will appear no earlier than 2020. They'll probably only be certified for highway driving when the weather is fine. They will be expensive and they'll be used by a few companies in pilot projects. They won't replace the driver but extend the hours while the vehicle is moving - the ex-driver can sleep while the truck is driving itself on the interstate.
Slowly they become popular across big rigs but for a long time they'll still need someone on board, even if not actively driving. That's for security or for negotiating town streets. We are talking 20-30 years until those millions of jobs will be lost. There's plenty of time for those people to retire or move on.