Tesla User Interface

28 February, 2022 06:00AM ยท 14 minute read

I’ve written about my new Model 3, but wanted to pull together some thoughts about the Tesla User Interface (if I can call knobs, buttons, switches and a screen in a car a User Interface…pardon my Engineering parlance…) now that I’ve driven my Tesla for nearly 8,000ks (5,000mi) over the past 4 months.

V10 to V11

The changes made between V10 and V11 I’ll cover off in future maybe in another article. Most of the changes haven’t been improvements overall and I’ll cover those changes that are specifically detrimental below. Suffice it to say, most users/drivers didn’t like the V11 upgrade very much. For me, I’d only spent two months using V10 so re-learning where most things were wasn’t a big change for me. Others that had been using their V10 interface for years…were much more annoyed and I can understand why.

Easy Access Is Subjective

Having designed HMI Screens for two decades in critical plant control systems, my view on what information is critical vs what information is nice-to-have is very different from most people it’s fair to say. I’ve read so many peoples opinions about whether tap-tap-swipe to get to a function or not is just too hard and if glanceable information is missing. Frankly it’s opinionated and most people are too influenced by their own biases to be objective and to be honest with themselves.

Before we tackle that, let’s think about Tesla’s perspective, because it makes a big difference in understanding some of their choices.

Tesla: We expect you to be using AutoPilot

Not all the time, but most of the time. Yes, it’s true that AutoPilot on city streets is a bad idea - too many variables, obstacles, disjointed lines and things to account for but it’s getting better every month. On the highway or freeway it’s pretty solid and reliable.

In a non-touchscreen vehicle people can use tactile feel and use spatial awareness to locate the buttons, knobs and switches without taking their eyes off the road. Of course in recent years many vehicles have introduced touchscreens with CarPlay/Android Auto or their own entertainment system, as well as voice controls on some mid/high-end models. Whilst Smartphones have taken most of the blame in recent years for driver inattention there’s no denying that touchscreen entertainment consoles in cars also play a role in driver distraction. One might wonder if it’s safe to have any screens in a vehicle without Auto-lane Keep Assist (or whatever name you like) in the vehicle as well, but I digress.

For this discussion we’ll exclude the entertainment system-like functionality as it’s clearly a heavier touch than other interactions. For that matter we should define the four categories for functionality:

Let’s look at examples of each below, in no order of priority of preference.

Glanceable Information

I’m excluding non-automatic, and non-electric vehicle items like Oil Temperature, Oil Pressure, Engine Warning, tachometer and so on. Also worth noting that some items like Windscreen wiper state is directly observable without any indication on a dashboard, hence that’s excluded as well.

About Binnacles

In non-Tesla vehicles this information is almost always located in a Binnacle or Instrument Cluster if you prefer. It’s directly behind the steering wheel only requiring the driver to drop their gaze and look through a hole in the steering wheel. This isn’t perfect either since we can’t glance through the wheel at certain points of wheel rotation (when turning) and not all cars have good tilt/telescope and/or seat height/depth adjustment to suit every person. Due to my height in some cars I’ve driven, there’s simply no physical way to adjust the steering column and seat so I can see all of the Binnacle. The top is sometimes obscured and I’ve had to duck my head forward and down slightly to read the speedometer particularly at higher speeds.

An alternative to a Binnacle is a heads-up display, which super-imposes information that only the driver can see, on the windscreen directly in front of them. I drove a Prius-V for a few years that had this and I really loved it, however having something in your direct line of sight is not always ideal. Given the costs of implementing HUDs have dropped these remain an attractive alternative to a Binnacle that serve much the same function.

In a Model S and X there is a Binnacle, but on the 3 and Y there isn’t and no Teslas have HUDs. I have yet to find a seating position or steering wheel angle where the screen is obscured from view, but this visually clear sightline is offset by requiring not just a dip of vision, but also a glance to the side towards the center of the vehicle. Ultimately this takes slightly longer but in normal use isn’t much of an additional safely concern and with AutoPilot running, it’s no concern at all.

Glanceless Controls

Light Touch Controls

Heavy Touch Controls

To be fair there’s very little good reason for messing with heavy touch controls when you’re driving, even with AutoPilot. The risk is just too great. Some items require the car to be stopped to adjust and parked with the Park Brake applied for others - which feels like the right call. I don’t believe there are any truly heavy-touch control items of concern in the current Tesla UI.

Why AutoSteer is so Important

#RantBegins

Auto Lane-Keep Assist or AutoPilot/AutoSteer functionality is very important due to gaze-affected steering. There have been studies that look into the correlation of gaze and eye-position and induced movement of the steering wheel. In other words - we tend to steer in the direction we’re looking. The longer we look away from in front of us, the more we will drift in the direction we’re looking.

Before LKA/AP-AS was possible the only option was to ensure controls were tactile and binnacles were clear and concise to minimise this. With LKA/AP-AS technology, this is less of an issue provided it’s actually able to be used and is in use. The alternative position to think about though is, if AutoPilot isn’t working (poor conditions, dirty cameras/sensors) or disabled by the driver, then the only means to accessing information will then become standard tactile controls and a binnacle.

If we suppose AutoPilot isn’t always in use, the lack of a Binnacle and minimal tactile controls should result in a higher incidence of accidents with Tesla vehicles, particularly Model 3 and Y and yet there is not. Beyond the possibility that these are non-issues (could be…) a possible reason for this is that the higher cost of entry for these vehicles precludes ownership/access to younger drivers. It has been shown conclusively that younger drivers do not handle distraction as well as more experienced drivers. Therefore the marginally higher risk in non-AutoPilot situations is likely offset by the older driver demographic.

Projecting forward in time from here requires a lot of faith and assumption. If we assume that AutoPilot improves to the point where it can drive as well as or better than a human in all situations, then the interface deficiencies become a non-issue. Maybe even get rid of the Steering Wheel and the stalks? Hmm. The problem is that in that extreme scenario AutoPilot then becomes a requirement to use the vehicle in which case it is forced always on and the car can not move without it. Many people will never choose that option, therefore some balance needs to be found between full autonomy and catering for human attention to the road under non-AutoPilot operation.

#RantOver

Where Tesla’s Interface Falls Short

Much of the complaining for users is more centered around Heavy Touch controls or items that don’t directly affect safety or driveability or common usability of the vehicle. That said there are a few exceptions that fall under Glanceless Controls and Light Touch Controls that aren’t often covered that are currently not ideal:

Headlights: In every other vehicle I’ve driven, I can turn on and off headlights, set low and high beam and auto-highbeam (where fitted) using the stalk. In the Model 3 you can only turn lights off (if the Tesla decides they should be on in the first place) by using the touchscreen. Pulling the left stalk towards you flashes High Beam, whereas pushing it away from you and releasing toggles between Low and High Beam.

Headlight Suggestions: Pulling the left stalk towards you is common in many cars for flashing High Beams so should be retained, but add Pull and Hold for 2 seconds to toggle between High and Low Beam. Pushing forwards on the left stalk toward could toggle Auto-highbeam on and off, and Push forwards and Hold for 2 seconds could toggle the headlights on and off. Note: Normal cars don’t try to overload all of this functionality into two switch actions, they employ turn-dial positions and switches specifically for each independent control for a reason.

Wipers: Currently the end button on the left stalk has two depressed depths: Shallow preforms a single wipe, full depth performs a cleaning spray and wipe function. There is no way of changing the speed or cycle via the stalk, though pressing the button at all brings up a pop-up on the Touchscreen, this now becomes a Light Touch control for something you don’t want to be taking your eyes off the road in heavy rain conditions. It might be fine if the auto-wipers worked reliably but they don’t and given this impacts driver visibility it’s very important.

Wipers Suggestions: A single shallow press and release could cycle between OFF->Intermittent Slow->Int Fast->Continuous Slow->Cont Fast->OFF. A full depth press could single wipe with a full depth press and Hold for 2 seconds to trigger a cleaning wipe.

Demisting: In V10 demisting controls weren’t associated with a stalk, but rather on the bottom edge of the Touchscreen, easily available if needed. Considering the urgency or not employing these as they affect driver visibility it is very odd they were removed from the bottom bar in V11.

Demisting Suggestion: Put the controls permanently on the bottom bar for quick access.

Non-retentive Assumptions

If I could put my finger on one gripe about Tesla’s UI…it’s that many functions and settings are non-retentive. When you stop the car and get out, then get back in and drive off again later they reset to the factory defaults. There is no retention of my previous settings. Headlight for example…stop assuming that the light level outside require headlights. If I want to turn them off, when I do, they should stay in that state until I turn them on again.

Not-So-Auto

Nothing is perfect, but when things are designed to work in an Automatic mode, that mode needs to work in a pattern that is discernible, consistent and relatively well otherwise people can’t/won’t use it.

Conclusion

If we accept the addition of AutoPilot features easily offsets the marginal increase in delay for glanceable information then this is a tradeoff worth making for an otherwise unobstructed view of that information. The controls on the steering column (left and right hand stalks and thumb control wheels/joysticks) satisfy most of the key requirements for glanceless controls. They fall short however on: wiper speed control and headlight control and I suspect this is due to over-confidence by Tesla in their automatic headlight and automatic wiper controls.

Why Change Tesla, why? I’ve been thinking about what would/could motivate Tesla to modify their User Interface and thinking specifically about the Demisting features. One of the temptations would be to use end user real-world data to draw conclusions about the use regularity of certain features. If so I can absolutely understand how something like the demister button would be moved. In the real world, most people would rarely use this as the conditions need to be right for it to be required. The problem with that though is the thinking behind a big Red Button emergency stop. Sure we don’t push the big red button often, but it’s there in case you need it in a hurry - which is when you’ll REALLY, REALLY want to use it! Quick access therefore can not only be reserved for items of high regularity but to reduce the high risk of unwanted outcomes.

Of course I have no evidence that this is what Tesla have done or is doing, but if not I struggle with the thinking behind some of the V10 to V11 changes. Beyond this I think there’s a hard limit as to how many more functions could or should be pushed off physical controls and on to a touch screen provided the intention is to have a human driver in control of the vehicle AT ANY POINT. One user interface can serve both purposes but a little bit of tactile controls makes it safer in both use cases.

All of that said, the only good thing is that unlike most other car manufacturers, a fix to many of the above is only a software update away.