How fast is Tesla's engineering staff growing? Hardware & software

The rush to finish Model X is behind them but the rush to finish Model 3 is ahead:frown:
Hopefully though much of the software code is common so it should actually be fixed in any case.

These days, well architected software is going to be broken into platform-independent and platform-deponent modules. So, if you look at something like the AP suite, the UI and control algorithms will be the same across cars, but there will be code specific to the hardware on a Model S and a Model X. The upside of this is you are not reinventing the wheel with each car.

BTW, the career section of the website shows about 250 openings in Engineering.

Since the Model S was introduced, externally, there doesn't appear to be any major improvement in their software strategy. Obvious bugs still appear in released software. Software updates are released in a start-stop cycle, with pauses likely caused by discovery of new bugs. Apps are released with basic (or beta) functionality, with few improvements after the initial release (still missing playlists, waypoints, ...). Major new features are announced officially or unofficially by Tesla (App Store, 3rd party SDK, improved browser, ...) and not delivered without any subsequent information provided by Tesla on their plans. And we continue to live with the very awkward combination of Navigon's software on the dashboard coupled with different software on the Google maps console display. And Tesla still seems to focus new releases primarily on the "next big thing" with only minor changes made to previously released apps. And, we're still waiting to hear Tesla's response to most of the other cars gaining Apple CarPlay and Android Auto.

Tesla's had the advantage the last 3.5 years in having the only long range EV on the market - without any real competition. The new Chevy Bolt seems more like a "mid range" than a "long range" EV, because of the lack of something like the supercharger network to enable road trips. But the Bolt may have better software for the core apps - navigation, media playback and smartphone integration - and should be on the market later this year.

My concern is that unless Tesla can do better with their software, should they ever have direct competitors for their vehicles, the software could become a drag on new vehicle sales. And, while Tesla may be the first to the market with AutoPilot, other vehicles will be getting comparable software soon - and Tesla's only long term advantage may be their willingness to more quickly deploy new software updates than the other manufacturers, giving them a temporary time-to-market advantage.

omarsultan said:

These days, well architected software is going to be broken into platform-independent and platform-deponent modules. So, if you look at something like the AP suite, the UI and control algorithms will be the same across cars, but there will be code specific to the hardware on a Model S and a Model X. The upside of this is you are not reinventing the wheel with each car.

BTW, the career section of the website shows about 250 openings in Engineering.

Click to expand...

I expect the software is the same across the fleet and features only get enabled if the hardware it works with is found. I've written a lot of firmware in my career and that's the easiest way to do it. It also makes it much easier to maintain down the road.

Maintaining similar but different code trees is a royal pain in the backside.

If I was designing the firmware for Tesla's cars, one code base for everything is the way I'd go. The only reason to do separate code would be some sort of safety regulation. Unless some bureaucrat is insisting it be done a certain way, there is always ways to make it safe with one code base.

Looking at the way software is done at Tesla, based on what gets released (or escapes :wink, my guess is that the software management and culture at Tesla is far from world class. The actual software required to do everything in the UI seems pretty unchallenging, and yet it's a mess, and it stays a mess release after release. Any organization that would even demo the current navigation system to an in-house audience, much less release it, alpha, beta or zeta, is clearly badly screwed up.

Since they are alone in the market, they can get away with it for now. However, it's pretty hard to change a company's culture when it eventually is faced by real external challenges.

doctorwho said:

The rush to finish Model X is behind them but the rush to finish Model 3 is ahead:frown:
Hopefully though much of the software code is common so it should actually be fixed in any case.

Click to expand...

Model X is far from finished considering that Tesla can barely push any out the door, prompting JPMorgan to downgrade TSLA's value based on poor Model X production. That's not what I would call a finished product.

Now hang on. Yes, nav could use improvement, but that's primarily on the vendor/ choice of vendor.

I've seen significant improvement from 7.0 to 7.1, with only one major problem being the front-end collision problem. I say that's due to some component being too busy to respond in time, which is most likely an unexpected consequence. Of course that's being fixed. Oh they've removed the Source button; that will be back.

Let's just all take a breath.

AmpedRealtor said:

Model X is far from finished considering that Tesla can barely push any out the door, prompting JPMorgan to downgrade TSLA's value based on poor Model X production. That's not what I would call a finished product.

Click to expand...

From an engineering perspective, the Model X is a finished product. If they are having problems getting production sorted out, that requires very little engineering support. And the support is mostly industrial engineers who are probably part of the production department rather than the R&D department. Some R&D engineers are probably still there fixing bugs and available for redesign of anything that is too much of a problem, but the bulk of the Model X R&D team is probably working on the Model 3 now.

The navigation problems will likely never get fixed using the current combination of Navigon and Google maps. By using different apps for the dashboard/navigation and the console - this has created a number of problems - maps are inconsistent - even the labelling of the exits/turns are different. Navigon is better at identifying the upcoming exits - while the console usually provides a confusing list of street names.

The Navigon maps are not being updated annually - as promised, and even after the only update that was distributed in 2014, the maps were already out of date compared to Google.

And the functionality restrictions may be due more to using the combination of Navigon and Google - if Navigon was used for everything, surely the software is capable of basic features like waypoints and route customization. But because the user interface is being done with Google maps on the display - that could be making it more difficult for Tesla to provide a fully featured navigation package.

Except for the AutoPilot (which doesn't work with the classic cars) and some relatively small changes to the user interface, what significant improvements have been made in 7.0 and 7.1?

The auto-open/close for Homelink is nice - but that was required to support AutoPilot.

We are still lacking basic functionality in navigation and media playback. The owner frustration on simple things like placement of time and temperature - plus removal of the media Source button seems to be simple things that should have been caught during the testing of the releases, before they were released. And Tesla is falling further behind in smartphone integration - the only major improvement made since the Model S was introduced was adding calendar events - which doesn't seem near as important as other features like voice text messaging or being able to use the console to navigate through media stored on the smartphones.

Almost everyone on the Tesla forums wants Tesla to be successful, and while we are really all "early adopters" and have been pretty patient with some of the inconveniences, Tesla's inability to make improvements in the functionality and quality of their software could become an increasingly larger problem - especially if other manufacturers start producing more competitive EVs - with better software (like the Bolt with Apple CarPlay and Android Auto - and playlists!).