Is Autopilot that much better than Competitors?

[ToddRLockwood]

How good is that last feature you mentioned? I don't have my Model S yet, but during my test drive I found that after setting the cruise control, a steep downhill caused the car to accelerate - it wasn't able to regen enough to keep the speed steady. Have you tested it on steep downhills?

That might be the case on hilly San Francisco streets, but in my experience it's no problem on secondary roads or mountain roads. Was the regen on your demo car set to Standard or Low. I'm willing to bet it was set too low.

 

[breser]

Honestly, I know a lot of people are really excited about autopilot, but this doesn't strike me as something that Tesla can disrupt, the way that it is disrupting so many other things. The hardware, software, and overall functionality of autopilot is going to be common across the industry within a few years. There's no reason to think Tesla will be better at doing this than Mercedes or Ford.

I think how they can disrupt this is they have a capability that other car manufacturers don't. Software updates. This is not a trivial capability. It requires a great deal of effort to plan to support software updates and then actually implement them in a way that doesn't cause problems. Because Tesla has this capability they have a lot of room to add features and improve them. The other guys might be able to do updates when a car comes in for service. So they're a lot harder to get into the hands of the users. Which means the rest of the manufacturers have to be much more conservative about what features they put out. Mercedes and Audi have been working on this much longer to get what they have done. Tesla is just starting and seems to be moving quickly. Once they've caught up they can begin to move beyond. Not just for future model years but for many vehicles already on the road.

So in my opinion Autopilot brings an opportunity to Tesla to have their software updates be far more important to the driving experience, not just extras like adding a Calendar or improving Navigation.

I view trying to compete in this space as a bit of a distraction. Tesla's core mission (and competency, and competitive differentiator, is its EV knowledge (and supercharger infrastructure). They have to have driver assist to keep up with the other companies, but any time they spend on that that takes away from BEV and SpC development is, in my mind, a waste.

This sort of argument is not really that true in my experience. People tend to presume that effort towards one thing is fungible towards another. Once the hardware sensors are in place for Autopilot most of the work is on software. Whereas the BEV/SpC development is mostly a matter of Electrical Engineering. Now you can argue that Tesla could hire more EE's and fewer Software Developers. But again this isn't really that simple. Putting more people on a project doesn't speed things up, in many cases it actually slows things down.

 

[andrewket]

I sure hope Tesla is able to do all the things demonstrated in the Mercedes demo. I worry that tesla doesn't yet have all the sensors required, most notably rear radar, and that we're going to be disappointed. Has tesla oversold autopilot?

 

[JST]

To refine my point a bit, I'm not saying Tesla shouldn't invest in driver assistance tech. They have to--it's basically table stakes at this point.

But it's also an area where every other automaker and supplier a) has the know how, and b) has the *desire* to invest and innovate. Moreover, you've got other players (like Google) that have a lot invested here already.

BEV tech and (most importantly to me) SpCs are areas where other companies are *not* investing. They are actively ignoring those opportunities. Why? I have no idea. But that's an investment multiplier--every dollar you spend there is vastly more effective than the dollars you spend where everyone else is investing, too.

On software downloading, yes, that's a great feature. But as andrewket observes, do the cars being built now have the hardware they need? Downloading won't help my 2013 do any of this stuff, of course. And clearly some level of autopilot will be available on the cars being built now. But is there going to be another wave of assistance tech that becomes available only with additional hardware?

 

[gregincal]

Sorry, but if you really need or want your car to park and unpark itself, you are missing the point of owning an automobile. I love driving and want nothing to do with it handling those functions myself.

I love the driving experience, everything about it. If you automate any of that it would be like playing a game of baseball and having the bat swing itself for you. No thanks.

I'm sorry, but automatic parking does not equate to swinging the bat in baseball, more like insisting that you're not really playing the game if somebody hands you your bat instead of picking it out of the rack.

 

[mdevp]

I think breser has a point with the updates. Tesla is the only manufacturer that will be able to update real time making its Autopilot superior over time. However at the event, Musk didn't say our autopilot will be far ahead of everyone else in the future, he meant ours is better now. IMO he knows something about Tesla's autopilot and how its superior to others. We just don't know what it is! (hopefully)

 

[yobigd20]

There was a detailed discussion about this on the Tesla Motors forum and the consensus among some was that with the lack of rear and side radar whatever Tesla can activate with the current sensor array via software is limited.

The Mercedes system uses two different types of rear and side radar. The sensor array Tesla has implemented is only aware of 16 feet around the car but based on a graphic on the thread below the Mercedes system can sense traffic behind the car at up to 260 feet away. IMHO the autopilot abilities they can activate via software will only be as good as the sensor array and without rear and side radar the Tesla sensor array is not as advanced as what Mercedes offers.

http://www.teslamotors.com/forum/forums/theoretical-capability-auto-pilot

If that's true about the side and rear radars that they only see 16 feet, what happens when you go to change lanes at 50mph and some jackass (like me) is coming up at like 100mph? You'd pull right in front and cause an accident. That worries me that there isn't long range radar on the rear.

 

[yobigd20]

Really? I suggest you put the Tesla on a narrow and twisting alpine back road and I'll be happy to see the results.

The systems available e.g. for the Mercedes S-Class are more likely restricted in their abilities for legal rather than for technological reasons. As to the feature of the car going off to park itself and to return when called - I'm sure that it's just a huge lawsuit waiting to happen. How many people will really put it to use at home rather than to call the car when it's pouring down and they don't want to cross a public parking lot in such conditions?

BTW, is Tesla's technology its own or do they get it from some supplier?

I can't see these types of auto driving systems being complete without things like IR sensors to pick up, oh like a baby crawling in the driveway or something that might be below the radar's range of sight.

 

[Canuck]

In my opinion, driver assist and autopilot are being hyped far beyond the actual capability of the technology. I think people will find that it's very limited with promises coming for years about what it is capable of doing, without that happening for at least another 5 years, and probably more likely in the next decade, and new hardware will be required to meet that potential. I really hope I'm proven wrong, but reading the experts on this issue, it seems we are much further away from solving many key problems with this technology than many people think:

Google’s Self-Driving Cars Still Face Many Obstacles | MIT Technology Review

 

[ItsNotAboutTheMoney]

On software downloading, yes, that's a great feature. But as andrewket observes, do the cars being built now have the hardware they need? Downloading won't help my 2013 do any of this stuff, of course. And clearly some level of autopilot will be available on the cars being built now. But is there going to be another wave of assistance tech that becomes available only with additional hardware?

There shouldn't need to be a change until they want to do autonomy. Front camera and radar for advanced adaptive cruise and crash detection, all-round sonar sensors for blind spot detection and low-speed autonomy.

Tesla's approach was to figure out where they could and would need to be with driver assist, then work on it until they could be sure of the hardware requirements and could introduce some basic features, and then add it (with demos of interesting stuff to come, of course).

But, to answer the thread's question: no, it's worse. Right now Tesla only has basic lane keep assist and software to identify speed limit signs. Who knows if or when everything else will be ready?

 

[donv]

It's nice to know someone enjoys creeping along in traffic jams! Takes all kinds, I guess...

I love the driving experience, everything about it. If you automate any of that it would be like playing a game of baseball and having the bat swing itself for you. No thanks.

 

[Canuck]

It's nice to know someone enjoys creeping along in traffic jams! Takes all kinds, I guess...

There's no need to attack Perfect_Flaw as if he is some odd kind of character and this new technology somehow mysteriously has us no longer creeping along in traffic jams.

In fact, you will still be creeping along in traffic jams. The only difference is that the car will do the creeping along by itself with only limited involvement of the driver. I can see someone saying: "As long as I am creeping along in this traffic jam, I'd rather drive the car than have it drive me." I don't know how I feel about this issue because I have never experienced a self-driving car. However, my gut reaction is much the same as Perfect_Flaw, in that after playing with the technology, if I am stuck in traffic, I might as well be driving. If that surprises you, then perhaps it's the difference of people who like to be in control, rather than have external forces control them.

 

[breser]

If that's true about the side and rear radars that they only see 16 feet, what happens when you go to change lanes at 50mph and some jackass (like me) is coming up at like 100mph? You'd pull right in front and cause an accident. That worries me that there isn't long range radar on the rear.

I really don't understand why this is such a big deal. These are driver assist features not driver replacement features. It's called autopilot because just like an airplane you still need the pilot, the autopilot just reduces work load.

Driver-less cars are much further off than people assume. If you look into it Google's cars need very detailed maps and can't function in poor weather. These are not easy problems to solve.

Would a rear facing radar be good to have? Sure. But all it does is improve the safety of the system. Whether you have it or not doesn't change the fact that the driver is still responsible for the operation of the vehicle. Much the same as the pilot is responsible for operation of their plane flying under autopilot.

You don't have to have absolutely perfect sensors for every scenario to be able to implement a lot of useful features. Especially if the driver is made aware of the limitations of the system.

 

[voidptr]

Sorry, but if you really need or want your car to park and unpark itself, you are missing the point of owning an automobile. I love driving and want nothing to do with it handling those functions myself.

I love the driving experience, everything about it. If you automate any of that it would be like playing a game of baseball and having the bat swing itself for you. No thanks.

Ever come out of work, a store, or a restaurant, and while you were inside, the weather turned into an absolute downpour? And the car's on the other side of the parking lot, with your umbrella inside?

I have about a half-dozen times just since they announced autopilot. I would have absolutely loved to have the car come pick me up at the door instead of running across the lot and getting soaked.

 

[Canuck]

It will be interesting to see the accident data statistics once these features become more common. When red light cameras came out, they were sold as a tool to decrease accidents and that certainly made sense to me. I wanted them to prevent people from running read lights. Now we come to find out that the opposite is true:

Red Light Camera Studies

Maybe causing drivers to be less attentive will lead to more accidents? It certainly makes sense that a car that can stop itself, and steer around obstacles, will have less accidents, but I wonder if the corresponding effect on the driver being less attentive will be worse than the benefits, especially in the early stages of this new technology? I guess only time will tell.

- - - Updated - - -

Ever come out of work, a store, or a restaurant, and while you were inside, the weather turned into an absolute downpour? And the car's on the other side of the parking lot, with your umbrella inside?

I have about a half-dozen times just since they announced autopilot. I would have absolutely loved to have the car come pick me up at the door instead of running across the lot and getting soaked.

The autopilot announcement wouldn't have prevented you from getting wet. One of the people at the forefront of this research has a 11 year old kid and he "hopes" self-driving will be available by the time his kid is driving in 5 years. 99% of roads are not mapped. Parking lots are probably over a decade away from being mapped. Read the link I posted above on this issue which quotes from the leading experts. Despite the recent announcement, you might want to consider buying an umbrella to prevent you from getting soaked in the future -- but I do get your point and it would be a great feature to have, especially for where I live with all our rain.

 

[ItsNotAboutTheMoney]

The autopilot announcement wouldn't have prevented you from getting wet. One of the people at the forefront of this research has a 11 year old kid and he "hopes" self-driving will be available by the time his kid is driving in 5 years. 99% of roads are not mapped. Parking lots are probably over a decade away from being mapped. Read the link I posted above on this issue which quotes from the leading experts. Despite the recent announcement, you might want to consider buying an umbrella to prevent you from getting soaked in the future -- but I do get your point and it would be a great feature to have, especially for where I live with all our rain.

Parking lots don't need to be mapped. You can drive any way you like , because they're private land with no rules. That's what the sonar sensors are for. As long as it's on private land, the car can roll slowly to towards you with two simple rules: don't hit anything; keep right in gaps that are two cars wide.

 

[sitter_k]

I bet there will be people lined up to have an accident with a self parking tesla, whiplash claim already in hand.

 

[ItsNotAboutTheMoney]

It will be interesting to see the accident data statistics once these features become more common. When red light cameras came out, they were sold as a tool to decrease accidents and that certainly made sense to me. I wanted them to prevent people from running read lights. Now we come to find out that the opposite is true:

Red Light Camera Studies

Maybe causing drivers to be less attentive will lead to more accidents? It certainly makes sense that a car that can stop itself, and steer around obstacles, will have less accidents, but I wonder if the corresponding effect on the driver being less attentive will be worse than the benefits, especially in the early stages of this new technology? I guess only time will tell.

The big problem with red light cameras is that they cost money and are used as a revenue stream. The result is that in some locations, instead of thinking "Oh crap, I've misjudged the light, it'll be red, but it's too late now", people think "Oh crap, I've misjudged the light, it'll be red. Brake hard!", then they brake hard and then they get rear-ended.

 

[Canuck]

Parking lots don't need to be mapped. You can drive any way you like , because they're private land with no rules. That's what the sonar sensors are for. As long as it's on private land, the car can roll slowly to towards you with two simple rules: don't hit anything; keep right in gaps that are two cars wide.

That's wrong. Private roads, roads within shopping centers, parking lots, airports, sports arenas, business and recreation facilities, etc. that are open to public travel are subject to the same traffic control standards, rules and legislation as public streets and highways.

 

[breser]

That's wrong. Private roads, roads within shopping centers, parking lots, airports, sports arenas, business and recreation facilities, etc. that are open to public travel are subject to the same traffic control standards, rules and legislation as public streets and highways.

That's a regulation about signage, line markings etc... The very next question makes it clear that while they have to comply with the regulations about signage, nobody actually has authority to enforce it.

My experience with traffic regulations on private property is that there isn't any. E.G. if someone hits your car in a parking lot, the cops either won't come or if they do come won't issue a ticket. But that may vary by location.