wagonsrock

A friend of mine recently got a Tesla model 3 with enhanced autopilot. This weekend I drove the model 3, and he drove my 6 month old E400S back to back over the same stretch of highway, with similar traffic conditions. Obviously the build quality and appointments of the E400S were far nicer than the model 3, but apart from lack of a HUD and air suspension, the model 3 was otherwise far superior. My main focus was comparing the two autonomous driving systems, but the model 3 (which had dual motors for AWD, but not the performance package) had much better acceleration.

I've been using drive pilot for 6 months and am quite used to its quirks:

Poor lane keeping ability, the car wanders like a slamom skier within its lane.

Limited ability to corner in tight turns, car will run off road without driver intervention.

Terrifying emergency breaking is applied when approaching cars at a stop light because drive pilot does not sense stationary cars ahead unless its previously been following them.

When following cars in stop and go traffic, braking is applied too abruptly compared to normal slowing when driven manually.

No feedback about surrounding traffic and lane detection, other than an indication of when a car ahead is detected, and its distance ahead.

Automatic speed sign recognition is unreliable, e.g. in school zones car slows to 15 mph even though the limit only applies during school hours.

Inability to detect merging traffic pulling into a lane ahead of you: the E400 only detects other vehicles when they are immediately in front, with no forward lateral detection.

Failure to give clear warnings when the system disengages (which it does frequently) other than the change is color from green to yellow to grey of the steering wheel color in the HUD.

No over the air updates, so that when better algorithms are developed these are uploaded to cars.

As far as I understand it, no software updates at all. Even though the 2019 drive pilot system is supposed to be improved, 2018 models will not get this.

The drive in the Model 3 was amazing.

The car stayed firmly centered in the lane, with no wandering.

Excellent behavior when merging with other traffic, sensing cars in front as well as on the side, and adapting appropriately.

Very smooth approach to traffic at stop lights, similar to the performance of a human driver anticipating slower and stopped traffic ahead.

Very good driver feedback on what the car sensed around it, with the location of multiple vehicles behind, at the side and in front all individually identified in the driver information display.

Automatic speed recognition seemed to work well.

Its astonishing to me how primitive the Mercedes system is compared to Tesla, in a car that in principle can cost 50% of an E class. Most impressively, because the ever growing fleet of Tesla cars send data back to software development teams, they are constantly learning and improving drive pilot, and sending software updates back to the fleet multiple times a year. Mercedes seems so backwards by comparison. That said, the E400S is a nicer environment in which to drive, but it could be so much better. I'd encourage anyone who has not experienced a Tesla to take a test drive, just to experience the technology.

JDPEClassUK

Cambridgehank

I was at a "Cars and Coffee" event yesterday. There were four Tesla Model 3 and One S. The guy with the performance 3 showed off his iPhone App. His 12 year old son was able to start the car, drive it forward and back up. He claimed the next generation will be able to be parked without a valet. He called this the movie star treatment. I had to admit that MBUSA deleted the E Class's ability to be driven by the app. The guy with the S claimed he had one of the fastest sedans, and does not have to stop for gas. He has solar panels on his house and does not pay to charge his car.

nycebo

But he pays the upfront cost for the car. It's double the price of an E nicely appointed.

I agree, electric should be the future, especially since we wouldn't send our petrodollars out of the country. But right now, they aren't really good for anything more than commuting. Try to make it to Vermont or Maine, park in a condo without supercharging, and make it back.

Further, I think that the Tesla interiors are cheap as hell. They ought to be able to do better than that for the price they charge.

And, back on topic, all of these auto-pilots ought to be better regulated. None are ready for true autonomous driving and we've already read about the mortal accidents owing to malfunctions. Instead, I like MB's approach: it's auto for a short period of time but then demands that the driver place their hands back on the wheel to confirm that they are still awake, or alive.

JTK44

Just my $.02:

I test drove a Tesla s several years ago. About 45 minutes. Clearly the best overall car that I had ever driven. Was ready to sign there on the spot. Then reality stepped in:

1) While the range is sufficient, 200 plus miles, which means a stop every 2 to 3 hours, which I normally do, the charging time was horrific: I drive to Vermont and once I am in Vermont, there are no charging stations within 50 miles of where I stay in Ludlow. That meant that going from Long Island to Vermont, instead of two 10 minute "pit stops" I would have to make two 1 hour 20 minute stops to fully charge the Tesla. A 4 hour 15 minute trip now became over 6 hours! Ditto for the return trip. If there was snow on the ground even more stops to make sure the battery was fully charged at all times.
2) Cost of operation: Besides the Model S costing as much as an S Class and nearly $40,000 more than E class the actual cost of operations was not that much of a savings: The cost of electricity is calculated at between $.07 and $.11 per KWH. On Long Island we pay $.23 per KWH. Through in the cost of home charger, $2,000 and there is no cost savings until the third year of ownership or 30,000 miles.
3) Simple trip to for example Washington DC. 244 miles or Boston, MA 210 miles that with my E class make one or two pit stops of 10 minutes each, require 1 hour 20 minutes for a full charge. A trip to the Hamptons, 90 miles each way, on the return trip you get range "anxiety"
5) Often the charging stations are in "less desirable places", places you would not want to be in at night, alone.
6) Fit, finish and quality of interior materials not up to luxury car standards.
7) As Tesla bonds are near junk, their cost of borrowing is quite high. This is reflected in the extremely high lease rates for the Model S, nearly as high as the S Class and no leases at all for the Model 3.


On the plus side:

1) Acceleration that this breath taking.
2) Eerily quiet
3) With the ultra low center of gravity, superb handling.
4) Very, very low maintenance costs. However, as I lease the costs of maintenance of the E class is also minimal.
5) The most important: it is far, far better for the environment than an ICE motor vehicle.
6) As 90% of our driving which in in reality is less than 100 miles, an EV makes sense. It is that other 10%, the road trips that require a lot of planning if you drive an EV. With the E 450 and its 21 gallon fuel tank, the range is between 500 and 600 miles. That means re-fueling is when I want to as opposed to the Tesla, when it needs to!

IMO, and again this opinion not fact, the time to charge will come down, I believe Porsche has a system in place in Europe where the charging time is under 30 minutes. Once the charging time comes down to under 20 minutes, the EV will replace the ICE.

I have just read where VW is planning on building 2 million EV a year: They are converting a factory and will be laying off 13,000 workers because the EV requires fewer workers to build than a comparable EV. Economically the EV makes sense.

As to the autonomous driving capabilities of the Tesla and the Mercedes: Think about this: when was the last time anyone bought a computer from Siemens or any European computer manufacturer? Then why would you expect the Mercedes computer aided functions to be any better?

I do not want to get into a discussion of whether Tesla will survive or not, but wanted to give only my $.02 as the major pluses and minus of the EV vs. ICE.

jsclarke

Thanks Wagonsrock- I think you're right. A friend with a Tesla S smokes my E300 on auto-pilot. And zero updates- a very old fashioned approach to new technology (that NEEDS updating). The car wanders in the lane- like auto-pilot by Mr. Magoo!

2012 merc amg

I can say first hand that the Distronic on the 2019's is greatly improved over the 2018's. I had a 2018 AMG E63 Wagon and now a 2019 E450 Wagon and the Distronic in the 2019 is much better and does not wander or try to cut off the highway when it sees an exit.

wagonsrock

Same as the MB system, Tesla now forces drivers to move the steering wheel after just a couple of mins to keep autopilot engaged. More aggressive. Different from MD the Tesla system will disengage autopilot for the rest of the trip if the driver repeatedly ignores these warnings (punishment mode!). To reset the system you have to pullover, shut the car down, and restart. Pretty effective.

wagonsrock

It would be really interesting to know what's changed between 2018 and 2019. If its just software there is simply no excuse not to offer a (free) update. If MB added additional sensors, then its understandable that the features cannot be retrospectively implemented to 2018 and earlier models with drive pilot.

Tesla's approach was to build in all the hardware into production models, including sensors that at 1st were not used, and then charge owners for the software ($5k at purchase, or $7k if enabled after purchase), and implement use of additional sensors as the software was upgraded. Apart from the radar, the cost of cameras and ultrasonic sensors are really inexpensive, so its easy to adopt this approach. For the model 3 they currently give new owners a 30 day free trial, and then deactivate the autopilot software if its not purchased.

JTK44

What truly amazes me is the over the air connectivity of the Tesla. I was reading a review of the Model 3 in Road and Track: with a lap top, the engineer was able to modify the braking and shocks. In the past these were mechanical things. My present Ford Edge you can modify the steering wheel effort. Again in the past these were mechanical.

This is where Tesla is light years, probably closer to 1 to 2 years, ahead of everyone else.

ua549

Unfortunately OTA update mechanisms provide an easy opening for hackers. It is possible (and demonstrated) for a hacker to take effective control of a moving vehicle from outside that vehicle.

CaptainE

wagonsrock

I never had the chance to drive through school zones in my friends Tesla M3, so don't know how it handles speed sign recognition. My MB is a 2018 model year and perhaps they have improved this aspect of drive drive pilot in 2019, and as you say these systems will continue to evolve. In my 2018 E400S the system also sometimes fails to update the speed correctly; e.g. it stays at 50 mph in a 30 zone until it sees the next sign, and also it frequently fails to read a speed sign if I am in the outside lane of a multi lane highway and the speed sign is on the left.

Bottom line is its a pain to use, so I turn off automatic speed adaptation, but leave on the speed sign recognition, so (for most of the time) I have feedback on what the posted speed limit is in the HUD.

pcascio

I've never driven a Tesla, and obviously there is going to be a certain amount of bias on this forum, which I'll admit to. That notwithstanding, I wanted a Tesla S, but even a used one with Warp Drive, or whatever it's called, was too expensive.I found a 2017 E300 with P3# and only 11k for a great price earlier this year. I love it, and drive from CT to Daytona with minimal intervention. Tesla crippled its AP, requiring your hands to be on, or hovering over, the wheel a lot more -- a position that is very uncomfortable.

I am disappointed that MB, knowing the newness of its Auto Pilot, has has not passed along updates as it's collected more data and implemented software changes on newer models but withheld them from existing models.

IMO, this is foolish. We're not talking about a sound system, we're talking about new technology that includes inherent safety considerations. If they are making improvements in a newer model that can be provided to me with only a software update, they owe it to me and all their customers.

Why do car manufacturers in general think their software shouldn't be upgraded without buying a new car? Imagine the lawsuits that may eventually ensue when accidents that were preventable, occurred simply because updates that were available were withheld.

nycebo

Read this. Apparently it's the dealers.

Boiled down to its essence, OEMs can't offer existing customers new features for their vehicles without the car dealerships getting their cut. This is in contrast to Tesla, which has done much to highlight the utility of OTA updates.

Love it or hate it, you can't deny that Tesla has been breathing fresh air into the US car market. By eschewing the traditional way of selling cars in the US—a network of dealerships which the OEM supplies with new cars for a price—it has also been unencumbered by the restrictive regulations that those dealers (and their lobby group NADA), have had written into law.

pcascio

Sounds like a lame excuse, but the Dealers Associations are going to expose themselves, and their members, to liability. However, the manufacturer should still circumvent this as it's in the public interest and safety. If the automaker pushes it will get its way.

wagonsrock

Likewise, if the customer complains to MBUSA, the dealers will eventually get the message

ua549

IMO a potential conflict arises because MB has no control over the software state once a vehicle is sold. There are literally dozens of software hacks available to consumers. Those software hacks may conflict with software updates from the manufacturer which may cause unintended consequences. At least in an MB service bay the software is re-imaged, not patched. OTA updates are notoriously unreliable because wifi and cellular protocols have very limited error detection and/or error correction.

JTK44

Sorry I cannot agree:

You wrote:

"At least in an MB service bay the software is re-imaged, not patched. OTA updates are notoriously unreliable because wifi and cellular protocols have very limited error detection and/or error correction."

If Tesla can do it then every other manufacturer can as well.

Again, I actually have driven the Tesla S and would have gotten it - as it was far and away the best vehicle I had every driven - but for the time it takes to recharge.

What you write might make sense if it were not for Tesla.

ua549

Tesla is at great risk for hackers. One reason it is not a big problem currently is due to the scarcity of Tesla's to target. I speak from experience as a retired international networking and security consultant to governments, large banks, ... Patching requires complete knowledge of the target code being patched. The OBD2 devices sold by third party vendors have the potential to modify the code (a work around) in addition to simply turning an option on or off. Three years ago Tesla was relatively easy to hack. Researchers were able to download an update from the Tesla server from hacking the Model S for network info. Then they were able to crack the weak passwords contained in the update file. Hopefully Tesla has greatly improved the security of their update server as well as the security in the vehicle itself. I would never purchase or even ride in a car with OTA update capability until security is vastly improved.

JTK44

Dear Ua549:

You wrote:

"Tesla is at great risk for hackers. One reason it is not a big problem currently is due to the scarcity of Tesla's to target".

IMO, this is causation without correlation.

In fact there are over 200,000 Tesla on the road today. At an average price of $75,000 that is over $15 billion worth of cars that could be hacked. Surely that is a big enough target, hardly "scarce" as you put it, for hackers. The fact that Tesla's are not regularly being hacked, is proof, at least to me that the Tesla OTA are secure.


"Hopefully Tesla has greatly improved the security of their update server as well as the security in the vehicle itself" and " I would never purchase or even ride in a car with OTA update capability until security is vastly improved."

See above. The OTA is already secure and like everything else in technology, improvements are being made daily, if not hourly.



Just think about this: When Consumer Reports found problems with the braking in the Model 3, within days Tesla corrected this problem. Isn't this far better than having to make an appointment with your dealer, waiting several weeks for the appointment, and until the appointment, you continue to drive your car with the braking problem?

I assume you have a smart phone and/or tablet. I have both an iPhone and an iPad. They are regularly updated. Do you allow your smartphone and/or tablet to be updated as well? I do!


OTA is the future. Embrace it.

nycebo

The researchers detailed the approach:

The amount of work that goes into the hack is surprisingly minimal, and not much equipment is needed. All of the necessary hardware costs less than $600, and the Model S can be hacked within seconds without the owner realizing it. The researchers built their device with a Raspberry Pi computer, a Yard Stick One radio, a Proxima radio, an external hard drive, and batteries.

JTK44

Thanks for the link.

The most important thing is that Tesla took immediate steps and corrected the flaw in their security. This confirms what I posted: "...like everything else in technology, improvements are being made daily, if not hourly"

ua549

Researchers have been able to hack into moving vehicles from outside the vehicle (FCA Jeep). Would you feel safe in such a car? All manufacturers need to harden their vehicle operating systems to reduce the risk profile. Although industry standards say that onboard systems are supposed to be protected against unauthorized firmware updates, the researchers found that they could change the firmware on some systems without any sort of authentication. In one attack that the researchers call "Self-destruct" they launch a 60 second countdown on the driver's dashboard that's accompanied by a clicking noise, and then finally warning honks in the final seconds. As the time hits zero, the car's engine is killed and the doors are locked. This attack takes less than 200 lines of code -- most of it devoted to keeping time during the countdown.

Ten years ago researchers needed physical access to the CAN via the OBD2 connector. With OTA physical access to the CAN is not necessary.

JTK44

Here is a link to an article that is quite informative, at least imo:

see: https://www.industryweek.com/technol...oems-need-know


Here is a quote from the article:


"+By 2020, Statista predicts that worldwide, connected cars will make up 98% of the new car market. And with this innovation comes increased vulnerabilities."