Tesla Priorities: Refine Autopilot or Fix Everything Else?

[Lump]

It makes me sad to say this...

Unfortunately, at this point I would not be able to recommend that any of my friends or family purchase a Tesla. We are three years into production, but Tesla still appears to be for beta testers only and not for those who want a finished, polished product. My business partner would never accept a navigation system that did not perform on par with her existing Lexus navigation system. My brother-in-law would not accept a Trip Planner that did not work well enough to rely on for determining when and where to charge. We live in the real world where we look for real solutions, not in a bubble where "just okay" is good enough.

Tesla lost me in a big way after Elon's idiotic and embarrassing "end to range anxiety" press conference. The last 12 months of software updates, with essentially no focus on polishing existing features, has left me disillusioned and questioning Tesla's focus and how it is being managed. The reality of Autopilot is that Tesla is catching up to existing auto makers, not blazing a new trail of its own. Self-parking, summoning, and auto-steering are all features that have been offered by other makers. The only feature Tesla has that others don't is automated lane changing. Other than that, BMW has already beaten Tesla to market with these features. Tesla has no first-mover advantage in anything anymore. I will note that BMW's software is NOT in beta and is release-level.

After two years I still love my car and how it drives, and service has been excellent, but in everything else my overall experience has been disappointing. Once you get beyond Tesla's electric drive train, you quickly realize that Tesla is far from a technological leader. In fact, it's quite a bit behind everyone else. BMW has already implemented gesture controls into its flagship vehicles, which is superior to a touch interface in terms of distractions. Tesla can't even give us a decent navigation system or provide a function as robust as EVTripplanner - a tool that was created by a student and is barely being maintained anymore. But even that is better to what Tesla came up with.

I have nobody to point to at Tesla other than Elon Musk for being ultimately responsible for the state of stagnation with the car's core software. Either Elon is spending too much time at SpaceX and not enough time at Tesla, or he is a terrible, awful manager who doesn't know how to properly delegate tasks and hold members of his team accountable for what they create. Given the horrible state of communications at Tesla for years despite multiple hires and executive turnover, the one common thread is Musk himself. There's nobody else visible enough at Tesla to hang this on but Musk.

The fact that Model X uses the same gimped navigation, trip planner, and media system as Model S should be alarming to the Tesla faithful. Even with its newest crown jewel that Tesla is taking extra time to perfect, nothing is happening on the software front to fix the core software and bring it up to par with the rest of the industry. Based upon this track record, I have dim hopes for Model 3. Tesla will most likely continue down this path and expect its Model 3 customers to cut it some slack, forgive the shortcomings of a new product, etc. I hope that's not the case, but if it is, Tesla and its shareholders are in for a major reality check.

It's my opinion that Tesla's Board of Directors needs to put Musk into more of an advisory/visionary capacity and should hand over the day-to-day functions of CEO to a more skilled and competent manager who can actually finish what he or she started.

Do I also need to bring up the Roadster 2.0 update, as in where is it? Maybe best not to go there.

Look at the garage auto open/close function as a prime and recent example of Tesla's failures. This feature was released but it doesn't work at all. Apparently nobody at Tesla actually tested the feature in any real world conditions before it was released. Who approved this feature being ready? Who tested it? NAMES! But we will never know. All I do know is that Tesla's software team is showing a propensity for incompetence. It's likely that the incompetence is coming from upper management, but ultimately it reflects back on the software team because they are going to be the ones blamed for the terrible state of software. This may offer one glimpse into why Tesla is losing talent to the likes of Apple and others.

About a year ago we were laughing at the horror that was Fisker's software interface, comparing it to the beautiful and functional Model S interface. Then we got version 7 which essentially "Fiskerized" a lot of what we thought was superior in version 6, and even lost some important informational feedback. Look no further than the speedo/instrument cluster to validate Tesla's software issues. Autopilot drivers, even though they may not be using Autopilot, are forced to watch a video game on their speedo instead of getting useful and valuable feedback as provided by the version 6 circular dial. Anyone with average intelligence would understand that the Autopilot interface is only useful when Autopilot is enabled, so use one mode without Autopilot and use another mode with Autopilot. Even those who are not paying for Autopilot are being forced to use the Autopilot interface - which is essentially useless for them.

I have tweeted my concerns to Elon Musk and encourage others to do the same. Unfortunately, short of a negative high profile media piece, I don't believe Tesla is going to change its focus. Has anyone noticed that Elon Musk no longer attends enthusiast gatherings? In 2013 we often saw him in front of enthusiastic throngs of owners and supporters, fielding their questions and offering his insights. The last time I remember seeing anything like this was at the last annual shareholders meeting. Tesla cannot even keep its own web site updated with regard to the executive management team, take a look:

Management | Tesla Motors

Only two people listed - JB and Elon. If you were an investor, wouldn't looking at this page scare the hell out of you? One does have to ask, on many levels, what on earth are Musk and Tesla thinking. Or are they?

100% agreement, thanks AR!

 

[msnow]

I have edited my comment to "Navigon and Google Maps". Better now?

If you don't know why it is important that Musk remains the CEO at least up until auto industry disruption is past the tipping point with the Model 3 in mass production then the best thing to do is to support him getting on with it.

This company is not getting built the way you imagine it is being built. If you think that Tesla doing nothing but running around responding to pressure from current consumers will make the oil industry and auto industry lobbyists lie down and go away then I think you have missed something critical. Tesla poses a direct threat to interests that real wars with guns have been fought over, repeatedly.

This is a war on at least thirty different battlefronts simultaneously and the impatience of a small number of current owners that don't appreciate what a big deal it is that a highway capable car with an electric drive train even exists is one battle out of thirty and the least of it because it will prove to be a self-solving problem. Your frustration will be replaced by delight and surprise in equal measure.

There is no better governance model than a Musk or a Jobs type character leading industrial revolutions with concentrated authority (and earned respect) i.e. genuine leadership. No committee or professional manager could run Tesla and your contention proves it. Managers and committees weigh probabilities and risks relying on experts analyzing available data. A true leader is absolutely necessary where the correct path is into uncharted territory. That is how Ford was built, it was not built by a professional manager and Apple provides a concrete example of what happens when you remove the leader and install a professional manager and I'm not talking about Cook I am talking about Mike Scully followed by the resumption of genuine leadership with the return of Steve Jobs - and yes I was one of Apple's consumers that was royally hacked off when Steve Jobs dropped support for OS 9 shortly after introducing the first OSX Macs to the point of actually buying a Sony Vaio in protest when my Mac's screen was broken in airline luggage (and now I am writing to you on an OSX Mac and I have probably bought gifted and recommended well over $200K worth of Apple OSX and iOS based stuff since then for both office use and and houses full of networked Apple AV equipment and media servers that I can control from my iPhone).

So to answer your question. Source of the Tesla Maps project. Sure: Elon Musk, presentation on Autopilot 7.0.

Someone later in this thread asked "how the heck are they going to collect the map data". Answer: 70+ thousand sensor-equipped scout vehicles feeding back to a central database and counting. And the beauty of it is that this is an ongoing and expanding effort, updated in real time, and it does not need to be complete at launch for every single minor road for it to be useful. The Auto Pilot can drop back to scouting mode or even manual mode and do a fine job whenever it encounters a road that no Tesla has ever driven before. Nobody has ever launched a navigable mapping effort on this scale or with anything like the data precision that is being collected. For example Tesla is out accelerating Google's mapping efforts by a factor of 1000 already without any dedicated mapping drivers - and of course Tesla can add non-customer mapping drivers at will for example in advance of launching a new sales territory.

Ok we are going to have to agree to disagree on Tesla management topic but regarding the maps project you may be confusing the "detailed maps" that Elon was talking about for Autopilot and the kind of maps needed for navigation. Also 70,000 or 100,000 cars collecting data in a few countries is not enough. It would make more sense to utilize an existing and refined system so they the company can focus on building cars and AP. I did see your edits to include Navigon.

 

[Julian Cox]

Ok we are going to have to agree to disagree on Tesla management topic but regarding the maps project you may be confusing the "detailed maps" that Elon was talking about for Autopilot and the kind of maps needed for navigation. Also 70,000 or 100,000 cars collecting data in a few countries is not enough. It would make more sense to utilize an existing and refined system so they the company can focus on building cars and AP. I did see your edits to include Navigon.

No I am not confused about either the management topic or the map topic. Yes absolutely it is frustrating that there would appear to be readily easily addressed priorities affecting current owners that are not yet addressed. I can think of lots of things the business could do that it hasn't got round to or optimized yet. Just never forget that every other auto maker is actively fighting addressing any of these issues starting with the most important (the main bit of the car). Check out VW with its diesel nonsense or Toyota with its Hydrogen nonsense. If you are happy to settle for a car with a V8 then BMW will be most happy to oblige. Ultimately I think a lot of these frustrations come from a good place - wanting Tesla to do better for its own good.

I would suggest an effort to collate your feature improvement requests in a constructive and actionable format rather than dreaming that you have a better grasp on Tesla's business priorities than Musk does or taking personal offense when those priorities don't coincide or roll out in the same order as you would have done it. Musk is the guy doing it and it isn't as easy as focusing on customer satisfaction (or immediate customer gratification) and the rest will take care of itself, if only it was.

 

[ohmman]

No I am not confused about either the management topic or the map topic.

Can you provide another source for the Tesla Maps initiative, then? Because it's been made very clear that the high precision maps he discussed were for AP ML models specifically. If this is something Tesla's putting resources into, I'm going to be pretty surprised. Not sure if it'll be in a good or bad way, though..

 

[AmpedRealtor]

I'm 100% with AmpedRealtor here. What he wrote pretty much is true. Tesla has created this image of being the future of technology by being the first to create a really good EV. We all give them credit for it! But looking at the details there are so many things where Tesla really is lacking compared to other cars.

I have been frustrated with Tesla's focus on few features all released as 'beta' while basic functionality is still not up to the standard of current upper class vehicles.

The reason is obvious. Summons or AP are features that everyone gets excited about and makes videos and talks about. It's creating a buzz and attention for Tesla. Fixing a broken navigation system won't get any attention. Tesla likes to do the cool features, the ones that get a 'WOW'. Everything else is on the back burner.

+ 1000

At some point the chickens will come home to roost.

 

[calisnow]

Also 70,000 or 100,000 cars collecting data in a few countries is not enough.

I don't claim to know how to go about building a mapping system but if the cars really are building some kind of high precision layer on top of the regular commercially available navigation maps then you're probably right - 70,000 or 100,000 cars isn't enough. But Tesla's sales - and thus its fleet - are currently on an exponential growth curve.

So if having a huge on-road fleet gathering data is in fact an important part of building a competitive moat, then as far as I know Tesla is the only auto manufacturer in existence which has this piece of the puzzle in place.

 

[Spidy]

Someone later in this thread asked "how the heck are they going to collect the map data". Answer: 70+ thousand sensor-equipped scout vehicles feeding back to a central database and counting. And the beauty of it is that this is an ongoing and expanding effort, updated in real time, and it does not need to be complete at launch for every single minor road for it to be useful. The Auto Pilot can drop back to scouting mode or even manual mode and do a fine job whenever it encounters a road that no Tesla has ever driven before. Nobody has ever launched a navigable mapping effort on this scale or with anything like the data precision that is being collected. For example Tesla is out accelerating Google's mapping efforts by a factor of 1000 already without any dedicated mapping drivers - and of course Tesla can add non-customer mapping drivers at will for example in advance of launching a new sales territory.

I disagree that it does not have to be complete to a huge extent for basic navigation. If I drive my kid to a birthday party in the next village here in Germany then I would want that road in the navigations system just like every other car had it that I have driven in the last 10 years. It's simply unacceptable to basically downgrade on a premium car.

And as someone else already pointed out those cars are not equipped to collect all the information like street signs, house numbers etc. (probably not even equipped for storing and transmitting the giant amount of image data) There is a reason google has a 360° camera. And 1000 dedicated cars are not comparable to Tesla's 100k cars. Most of those people will drive their daily commute and other repeating routes every day, which gives you little new information. If they drive to visit someone somewhere they drive the shortest way there and back. A tesla driver drives to a village then takes the side road to his friends house and that's it. A dedicated google mapper will spend 15minutes in the village driving down every single road.

Even with OpenStreetMap and people on the ground we still use government or if we are lucky company street name lists (OpenStreetMap Straßenlistenauswertung)) to match them with what is mapped and what we have missed. For a whole year I didn't realize a streetname in my village had been missing until I got lucky and a local political party gave me a street name list.

 

[calisnow]

I would suggest an effort to collate your feature improvement requests in a constructive and actionable format rather than dreaming that you have a better grasp on Tesla's business priorities than Musk does or taking personal offense when those priorities don't coincide or roll out in the same order as you would have done it. Musk is the guy doing it and it isn't as easy as focusing on customer satisfaction (or immediate customer gratification) and the rest will take care of itself, if only it was.

BINGO. Well put.

In addition all these complainers seem to be operating on the assumption that Tesla's dev teams are static - that Tesla isn't growing the teams to the point where at some point they *will* have the resources to address the UI issues. This is a silly assumption. Most likely these UI issues will in fact be ironed out as time goes by and Tesla's engineering resources grow. Right now they seem to focusing on fighting the most important battle there is in the auto industry after the powertrain issue - autonomy. Autonomy is the battle which matters more than anything else.

One thing I would love to see is a chart showing whether or not there is a correlation between the age of posters on this thread and how important they think Autopilot development is. I would hypothesize that the older the Tesla owner, the less important they think Autopilot is. But I could be wrong of course.

 

[morbot]

Everyone has a different view into what the priorities should be... and none of us are certain how much Tesla hears us or what they're up to. It's bound to cause anxiety.

It'd be nice if they enhanced the MYTESLA section of teslamotors.com for verified owners and included some sort of Cumulative Voting Prioritization mechanism. A user generated feature/bug backlog, prioritized by votes. Each owner could have 3 votes to identify their "top 3" wishlist. Items with highest votes float to the top of the priority list. If your item is resolved, you get that vote back to apply to another item (or you can remove a vote to place somewhere else as needed).

There'd be no guarantee from Tesla that they would (or could) address the issues in that order. But they could see where interest from owners was at. There'd be overhead, you'd need people to administer the list, review new submissions and periodically update the list (remove items that were addressed, either by fix or by confirmation that it cannot or will not be done).

Maybe a bit of a pipe-dream... but to me, that'd be my ideal way to communicate my desires to Tesla

 

[Cyberax]

Where are you seeing that timestamp?

On the SD card with the mapping data on the central console (yes, I did dismantle it).

 

[msnow]

On the SD card with the mapping data on the central console (yes, I did dismantle it).

Ok, I'll trust you on that then.

 

[Cyberax]

I would suggest an effort to collate your feature improvement requests in a constructive and actionable format rather than dreaming that you have a better grasp on Tesla's business priorities than Musk does or taking personal offense when those priorities don't coincide or roll out in the same order as you would have done it. Musk is the guy doing it and it isn't as easy as focusing on customer satisfaction (or immediate customer gratification) and the rest will take care of itself, if only it was.

There's a site with the prioritized list of issues. Tesla ignores it pretty much completely.

And we're not talking about new features (like key-linked profiles), but freaking bugfixes to widely known bugs.

 

[Julian Cox]

I disagree that it does not have to be complete to a huge extent for basic navigation. If I drive my kid to a birthday party in the next village here in Germany then I would want that road in the navigations system just like every other car had it that I have driven in the last 10 years. It's simply unacceptable to basically downgrade on a premium car.

And as someone else already pointed out those cars are not equipped to collect all the information like street signs, house numbers etc. (probably not even equipped for storing and transmitting the giant amount of image data) There is a reason google has a 360° camera. And 1000 dedicated cars are not comparable to Tesla's 100k cars. Most of those people will drive their daily commute and other repeating routes every day, which gives you little new information. If they drive to visit someone somewhere they drive the shortest way there and back. A tesla driver drives to a village then takes the side road to his friends house and that's it. A dedicated google mapper will spend 15minutes in the village driving down every single road.

Even with OpenStreetMap and people on the ground we still use government or if we are lucky company street name lists (OpenStreetMap Straßenlistenauswertung)) to match them with what is mapped and what we have missed. For a whole year I didn't realize a streetname in my village had been missing until I got lucky and a local political party gave me a street name list.

Pretty sure that I didn't say anything about a downgrade - or any kind of inability to add third party or open-source meta data like street names, restaurants and house numbers. (It's not one or the other). What Tesla is doing (uniquely and at unprecedented speed) is laying down a set of virtual rails along millions of miles of roads every day - perhaps 10-20% of that figure incrementally day by day. Every time an existing or new customer or a dedicated scout driver goes somewhere new the virtual rail network gets extended. As you will know it is entirely possible for a Tesla 'scout' vehicle to navigate an unfamiliar road for the first time in autopilot and to continue to scout the road layout and the desired path and speeds through curves and junctions having fallen back to manual control - and after a few passes cars no longer drop out of autopilot because they know where they are going. No other car or Autonomous system knows where it is going or anything about what the roads will be like before it gets there. Tesla's does.

Another thing: Nobody else even has the capacity to map the roads like this. Ford is apparently boasting the largest Autonomous exploratory fleet with thirty (30) vehicles! Talk about the prize for the alternates! On top of that you can't even have a central OTA network in the USA if you are forced to use dealers because dealers wont sell cars that are connected to the OEM for fear of challenges to their warranty work income.

Dear Ford: we're replacing the transmission because customer reports a terrible noise in the cabin. Dear dealer, our OTA network clearly shows the noise emanates from a loose speaker connector on the stereo, we're not paying for a transmissions swap even if you fill it with sand-blasting grit and run it for 48 hours before returning it to us for inspection.

 

[Spidy]

As you will know it is entirely possible for a Tesla 'scout' vehicle to navigate an unfamiliar road for the first time in autopilot and to continue to scout the road layout and the desired path and speeds through curves and junctions having fallen back to manual control - and after a few passes cars no longer drop out of autopilot because they know where they are going. No other car or Autonomous system knows where it is going or anything about what the roads will be like before it gets there. Tesla's does.

I don't even know what to say...

We are talking about Tesla providing good navigation with up to date data where you enter an address and then the car gives you directions and maybe some traffic data.

Leave Autopilot out of this, unless you can show how your Tesla maps provide us with a global map, with street names, road restrictions, house numbers etc. in a very short timeframe.

Because that is what is has to offer when it replaces the current systems as you said in a comment before:

Tesla's future Nav system, the one that they are actually working on is based on Tesla Maps. There is no point in Tesla developing incrementally on Google Maps because Tesla's system will blow it away.

 

[S4WRXTTCS]

The map "data" that both systems use is Navigon which, according to my Service Center, was last updated October 2014. Tesla claims this is the main issue with the NAV system but since mine seems to get lost and circle around destinations that existed years before 10/2014 I'm skeptical that's the only issue. The other issue is that if you have the feature that auto routes based on traffic conditions turned on it will reroute you mid trip without you knowing it happened.

The rumor I heard was there was a dispute between Tesla and Navigon (owned by Garmin) regarding map updates. I find this really troubling because I seem to recall Tesla selling the Tech package (before the AP came out) on having perpetual upgrading maps. I didn't purchase the car at the time because I felt like the whole Tech package on a high tech car was really quite silly. I waited to till the Autopilot stuff came out where Tesla wanted money for SW that didn't exist yet. Somehow I felt like that made more sense.

Anyways as to the re-route for traffic my biggest issue with it is it didn't work in V7.0. No matter what you selected for the time saving limit it would ignore it. I don't want to be re-routed to save 5 min, but I do want to be re-routed if it's 30 min. Not only does that time-saving adjustment not work, but the re-routing is sometimes completely silent (as you rightfully pointed out). I turned it off to prevent the sneaky route changes.

Just fixing some glitches with the CID (so it always restores the cell connection properly without a reboot), and fixing the Nav would go a long ways in restoring confidence of existing customers who feel like they've been left out for far too long. I'd feel the same way as some of you if I had a pre-autopilot car. Instead I'm too busy playing with new features to eloquently write anything.

 

[msnow]

The rumor I heard was there was a dispute between Tesla and Navigon (owned by Garmin) regarding map updates. I find this really troubling because I seem to recall Tesla selling the Tech package (before the AP came out) on having perpetual upgrading maps. I didn't purchase the car at the time because I felt like the whole Tech package on a high tech car was really quite silly. I waited to till the Autopilot stuff came out where Tesla wanted money for SW that didn't exist yet. Somehow I felt like that made more sense.

Anyways as to the re-route for traffic my biggest issue with it is it didn't work in V7.0. No matter what you selected for the time saving limit it would ignore it. I don't want to be re-routed to save 5 min, but I do want to be re-routed if it's 30 min. It has an adjustment, but that adjustment doesn't work. Not only does that adjustment not work, but the re-routing is sometimes completely silent (as you rightfully pointed out).

Couldn't agree more and I was told the same thing about the Tesla/Navigon relationship. I think the web site said we get 7 years of free map updates. What they didn't say was that would only be 1 update every 7 years. Sure hope this is all corrected soon.

 

[dpodoll]

One thing I would love to see is a chart showing whether or not there is a correlation between the age of posters on this thread and how important they think Autopilot development is. I would hypothesize that the older the Tesla owner, the less important they think Autopilot is. But I could be wrong of course.

67 (marginally old) put me down for AP

 

[cpa]

67 (marginally old) put me down for AP

Sixty-two, just a hair less than marginally old, and no interest in AP.

 

[calisnow]

No other car or Autonomous system knows where it is going or anything about what the roads will be like before it gets there. Tesla's does.

Actually how do you know this? Are you sure Mercedes isn't doing fleet learning and/or mapping with their lane assist cameras? Just because they're not publicly saying they do doesn't mean they're not doing it.

- - - Updated - - -

I'm 38, and think autopilot should be the number one priority (assuming resources are limited and indeed it is a binary choice between bug fixes and moving to full autonomy - in reality I think we'll get both eventually).

I don't know about Ampedrealtor and the OP for this thread - Brooklynrab - but I have been ignoring vehicle OEM interfaces for years now and simply using my phone. Why? Phones improve at a more rapid rate than vehicle OEM systems do, and Android is so good now that I don't need my car's built-in system to do anything.

I can pick up my Droid Turbo (or speak into my Bluetooth earpiece), keep my eyes on the road and without hitting a single button say "Navigate to 123 Smith Street, Anytown, USA" and it just does it. Voice imprinted, voila. Auto-re-routes, always updating the OS - I just don't care about my car's built-in interface very much.

If the car has the ONE MOST IMPORTANT KILLER FEATURE - AUTONOMY - THAT IS THE GAME CHANGER. If the car can drive me down the road safely, allowing me to take my hands off the wheel and eyes off the road - I can fiddle around with an imperfect interface happily for hours (what else would I do? You can only listen to so much NPR and audiobooks before you need a break).

- - - Updated - - -

Also - no matter what the whiners think about their imperfect interfaces - Tesla's sales are exploding. Model S up 46% year over year for 2015, while the luxury ICE makers, including BMW, Benz and Audi (with their refined interiors and perhaps less buggy interfaces) all showed double digit sales declines in the S Class, 7 series and A8. So whatever it is that Tesla is doing with its limited manpower - it's doing something right.

And I'm another data point - I can buy any car I want, and have enjoyed BMW's and Benzes over the years - but I've got two Teslas on order and I'm selling my E55 AMG. The autopilot and electric drivetrain are so compelling they FAR outweigh any buggy interface.

IF today Mercedes was selling an autopilot with the Model S's capabilities, and Tesla was selling Benz's gimped, error-prone lane assist system I would immediately buy the Benz instead. That's how important self driving is, at least to me.

Full autonomy is so important that I'm leasing these Teslas (which I never do - I always buy) because I want very badly the next level of Autopilot which I imagine Teslas will have out shortly. So yes - if you want some real world data on what is driving sales - it's autopilot, at least for this pilot.

A faulty OEM nav system, a buggy internet radio interface and other interface glitches - those are so far down my priority list that they don't even rank in my decision making process on a car purchase.

 

[ecarfan]

@calisnow, excellent post. Nice to see someone taking the long view. I agree with everything you said about the importance of autonomous driving and am glad Elon is making it a top priority.