Only if they are reinventing how the alphabet sounds.
"Zeus" definitely would've been better!
Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
Autonomous Vehicle Startup Zoox Snags 17 Former Apple Automotive Engineers
- Thread starter MacRumors
- Start date
- Sort by reaction score
Only if they are reinventing how the alphabet sounds.
"Zeus" definitely would've been better!
I agree. I'm sick of half measures (like notched screens without Touch ID! LOL). If you can't make the car, don't make it. Don't do the exact opposite of what made you a successful company. That seems like idiocy. Software for someone else hardware? Beyond stupid.
"Google". (Among many other examples.) No matter how dorky, once a name is around long enough it seems normal.
Do Tesla cars have a pile of cameras on the roof?
--Eric
Gadzooks punks dead, last good gut wrenched.
Antenna Sam Cook, black book Pushead.
Cat skulls stacked to the black hole sunset.
Olive on a toothpick..... Dagwood Bumstead!
;0)
Came here to write this. That funding is far from what’s necessary to complete the task. They will be in constant funding raising rounds or acquisition talks.
I do still think, the gathering of visual/positional data is NOT the problem.
It's just the technical aspect of the hardware to gather the data.
Identity the object it has seen.
This is the simple nuts and bolts stuff.
The problem is understanding this data and making decisions based upon it.
That's the hard bit. The VERY hard bit.
I'm still struggling to see how this is going to work yet, as it's up against the human brain, which I know makes mistakes and bad judgements, but is great for deciding what to do in an instant in a real life scenario.
Driving up on the pavement, going thru a red light, stopping with someone waves you down, seeing a driver make a hand gesture that means they are going to let you out of a junction.
Following hand signals when driving off a road.
So far I've just seen them drive along nice easy roads.
It's just the technical aspect of the hardware to gather the data.
Identity the object it has seen.
This is the simple nuts and bolts stuff.
The problem is understanding this data and making decisions based upon it.
That's the hard bit. The VERY hard bit.
I'm still struggling to see how this is going to work yet, as it's up against the human brain, which I know makes mistakes and bad judgements, but is great for deciding what to do in an instant in a real life scenario.
Driving up on the pavement, going thru a red light, stopping with someone waves you down, seeing a driver make a hand gesture that means they are going to let you out of a junction.
Following hand signals when driving off a road.
So far I've just seen them drive along nice easy roads.
Except that the name zoox, but the trend with names with z and x must cease.
You're misinterpreting what many people are now predicting to be Apple's path. Apple isn't simply going to develop yet another software platform for automakers to use. There will soon be many software platforms, e.g., Google, Uber, Apple and many of the car manufacturers themselves. Like most everything Apple does, Apple will develop the hardware AND software. Apple's expertise is developing and designing great products, and then contracting out the manufacture of them. It won't be "someone else's car." It will be "Designed by Apple in California." There will be plenty of manufacturers lining up to build cars for Apple.
And no Apple won't be selling cars; Apple knows that model is dying. Apple knows that the future is what Uber is building, and Apple will offer a great user experience when you use the Apple Car Service where your privacy is protected. Folks looking for bare bones transport will use Google's and be subject to having Google record everything that is being said, record where people live and go, who they go with, etc., and be inundated with ads.
[doublepost=1504192075][/doublepost]
You can relax. Again, the "idiocy" is suggesting that Apple's plan would be to make software for car manufacturers. The car manufacturers are already developing their own software and systems, and there are multiple companies already developing just software, e.g., Google.
Last edited:
Wtf are you talking about????
Apple decided NOT to build a physical car, just the software. So the experts on stuff like “braking systems” became completely superfluous & a company that IS attempting a physical car hired them.
You’re suggesting that Apple will do a 2nd feasibility study... conclude the 1st one was wrong, pivot back to plans of building a car, then hire back everyone that just left, through acquisitions??
Omg, trying to understand what you could possibly be thinking is making my brain hurt.
About a minute after I saved my post I thought a "wink" emoticon better conveyed my kidding around nature than a "smiley" so I updated the emoticon, but in that time between my original post and updating the emoticon folks were already starting to post comments so I think a subsequent post grabbed my original emoticon off the server.
Last edited:
When Nvidia sell something better than what you built in a box, you know it is time to move on.
http://www.nvidia.com/object/drive-px.html
[doublepost=1504197584][/doublepost]
It is all based on the data. The systems learn from data, and are not explicitly programming with things like parameter tables, if statements, etc. They "watch" human drivers and the environment, and see how the human driver reacts to the environment. And the more drivers the system has to learn from, the better it learns what to do, and more importantly want not to do in different situations.
This part about more data is key, and is one reason Tesla has their vehicle upload their data every night. Their self drive system as the experience of 1.5 Billion miles of driving, in all sorts of conditions and weather. And as the system gains more knowledge, Tesla updates a car's systems with the latest versions of the self-drive software.
Check out some papers on Deep Learning or Deep Neural Networks if you want to know more about a key underlying technology.
http://www.nvidia.com/object/drive-px.html
[doublepost=1504197584][/doublepost]
It is all based on the data. The systems learn from data, and are not explicitly programming with things like parameter tables, if statements, etc. They "watch" human drivers and the environment, and see how the human driver reacts to the environment. And the more drivers the system has to learn from, the better it learns what to do, and more importantly want not to do in different situations.
This part about more data is key, and is one reason Tesla has their vehicle upload their data every night. Their self drive system as the experience of 1.5 Billion miles of driving, in all sorts of conditions and weather. And as the system gains more knowledge, Tesla updates a car's systems with the latest versions of the self-drive software.
Check out some papers on Deep Learning or Deep Neural Networks if you want to know more about a key underlying technology.
Last edited:
Oh yes, I do understand that.
And I'm sure that works well for normal driving.
I just struggle with all the non normal driving aspects.
The road is flooded, do I drive thru it and take a gamble as to how deep it will be.
Backing up a narrow UK country lane when you are face to face with a van coming the other way.
Do you stop when a police, fire, medical officer waves their hands at you, well what about a hitchhiker?
Do you break the law when a road situation arises?
Driving up onto the path for example
If someone flashes their headlights (which means nothing legally) does that mean you should pull out in front of them even though they have the right of way?
As people have said, heavy snow/rain etc.
Will people in "power" be able to press a button and make your car stop any time they like?
How to deal with human drivers who know a driveless car will always give way and stop.
How to deal with kids? who can just cross the road anywhere as they know the cars will automatically always stop for them.
The system learns how to react in these situations based on how humans react in these situations. So over time these will be encountered and incorporated into the rules, or edited out of the data set.
Really, driving through a flooded road is probably are really dumb idea as shown by the events in Texas where doing this results in deaths.
They already stop for police, not sure about hitchikers.
Do you trust when someone signals you to cut in front of them by flashing your lights. I don't
Snow and rain. The Tesla model S is the #1 selling car in Norway, so I imagine the data set has a lot of data dealing with nasty weather.
Their have been incidents where people have tried to "mess" with self driving cars. I did it myself with the Google Lexus's they drive around where I live. I go in front and purposely slowed down. It went around me in a shot.
People fool them by getting rocking their bicycles back and forth at stop lights so they did not have to put their foot down. That no longer has any effect on the Google system.
Give that 45,000 people in the US and 1.2 M worldwide die in traffic accidents, 95% of which are caused by human error, we should not hold human behavior as the "gold standard". Let's face it, we suck at driving.
And the systems already very adapt at handing emergencies and human error. Check out this set of action by Tesla's old self drive and decide whether you would have done better. Note signal tones indicate the system is taking action because the driver is not doing anything
Last edited:
Sorry, I do have trouble seeing true level 5 in the real hard world becoming a reality for far longer than current media outlets are saying.
On pre-setup roads sure.
On certain routes only sure.
Whilst I don't doubt for 1 moment that a driverless car can drive along a normal road in normal conditions even now, I just can imagine so many scenarios where it's just going to get stuck.
Re the flood, I'm not speaking of cataclysmic 4ft water, I'm talking river has gone over the road, perhaps 6" even 8" deep.
It's down a secondary road, on your way to work, people are behind you, you see people on the opposite lane, one risks it, moves to the centre of the road, straddling your lane, you see them get thru, so you decide to take a risk.
After all you want to get to work and there are cars behind you.
So you move across lanes, keep the revs up and hopefully make it thru.
All judgement based on this new event.
Trust me, you won't find anyone more PRO driverless cars as a concept, I just worry we are not being sold the truth right now and I hope I'm wrong.
Drivers do thing that are not in the law books, like the flashing headlights to let someone in, a small gesture with the hands, or a nod of the head.
One aspect I genuinely don't think we are ready for is the "Car kills human" headline.
Hell we fall to pieces with lawsuits all the time with tiny issues.
Let alone reports every day of someone being killed in the driverless car.
We are no where near ready yet.
I did wonder if there will be some cop-out new laws introduced which stated you had to give way to driverless cars.
Another aspect I cannot yet see working in the human to car interface.
How do you get this fast and accurate enough to work?
Forget travelling from home to work car park
Or from home to the car park in town
I mean the tiny aspects of driving around town, do I go right or left, oh I see a van in the distance, I'll go left.
Going into this car park, but dam, looks like a queue building up, so snap decision I'll turn right, then left, go down the back street and I get get into that other car park.
There s no way for the human to tell a car that fast of a change of decision.
Human drivers could and would run rings around the queue of driverless cars.
So so much to deal with, in actual hard reality.
On pre-setup roads sure.
On certain routes only sure.
Whilst I don't doubt for 1 moment that a driverless car can drive along a normal road in normal conditions even now, I just can imagine so many scenarios where it's just going to get stuck.
Re the flood, I'm not speaking of cataclysmic 4ft water, I'm talking river has gone over the road, perhaps 6" even 8" deep.
It's down a secondary road, on your way to work, people are behind you, you see people on the opposite lane, one risks it, moves to the centre of the road, straddling your lane, you see them get thru, so you decide to take a risk.
After all you want to get to work and there are cars behind you.
So you move across lanes, keep the revs up and hopefully make it thru.
All judgement based on this new event.
Trust me, you won't find anyone more PRO driverless cars as a concept, I just worry we are not being sold the truth right now and I hope I'm wrong.
Drivers do thing that are not in the law books, like the flashing headlights to let someone in, a small gesture with the hands, or a nod of the head.
One aspect I genuinely don't think we are ready for is the "Car kills human" headline.
Hell we fall to pieces with lawsuits all the time with tiny issues.
Let alone reports every day of someone being killed in the driverless car.
We are no where near ready yet.
I did wonder if there will be some cop-out new laws introduced which stated you had to give way to driverless cars.
Another aspect I cannot yet see working in the human to car interface.
How do you get this fast and accurate enough to work?
Forget travelling from home to work car park
Or from home to the car park in town
I mean the tiny aspects of driving around town, do I go right or left, oh I see a van in the distance, I'll go left.
Going into this car park, but dam, looks like a queue building up, so snap decision I'll turn right, then left, go down the back street and I get get into that other car park.
There s no way for the human to tell a car that fast of a change of decision.
Human drivers could and would run rings around the queue of driverless cars.
So so much to deal with, in actual hard reality.
Some around here would argue that if you can't swap sensors on your car, you shouldn't market it as "Pro"...
Agree about Level 5. Sorry, I missed your mention of Level 5. Most of the current systems, and the ones that being targeted for the near term are targeting Level 4 where not all scenarios are not covered. But even with current development they are getting closer to Level 5. And data, as in actual usage is what is needed to get to Level 5. And this data will be acquired as more and more miles are self-driven and scenarios encountered.
And yes, humans can outperform autonomous systems today, but things like your car park scenario may not even exists in the future. Soon a Tesla can drop you off and go find an empty spot on it's own. What do I care if it takes the car 1 or 15 minutes to find a spot in Christmas time parking or at crowded office park? I am out and shopping or working, and the car will come get me when I press a button.
And the water on the road scenario is one where a set of sensor can estimate depth and speed of the water better than the human, and make a decision based on water depth, speed, tread on the tires, etc as to whether it is safe to cross.
One thing about AI technology is that it is like electricity or the internet. In the sense that it is a transformative technology. That is, how we did things a certain way before, will seem quaint and obsolete after.
Last edited: