Engineering Explained. This is bad.

[Smorgasbord]

I call this the rental car test. A car should be designed (as much as possible) to be an acceptable rental car. You roll in to Avis and rent a car, you jump into it, look around and evaluate the various controls and within 5 minutes you have a good handle on how everything works, start the car and drive off. No need to pull the manual from glove box (if there even is one).

I somewhat disagree. I'm OK with having to learn the way a new car in this price range works. What I'm not OK with is having to learn how to get around software bugs.

For instance, learning from the manual that the top left of the Instrument Cluster is actually a touchscreen is OK. Learning that to switch driver profiles means working around software bugs is not.

 

[Smorgasbord]

If they are really using QNX, that is a major mistake, IMHO. There are many ways to achieve soft real-time with normal OSes like Linux that are much easier to program.

QNX has a number of advantages over Linux. Not just inherent real-time support, but isolation of processes, fault-tolerance, etc from its micro-kernel architecture. And, you can still program in common programming languages. Shoe-horning real-time performance into Linux has never truly worked, and those that use it simply live with less than a real-time guarantee.

Heck, Nvidia uses QNX for its self-driving platform, Drive, where real-time performance is safety critical. Nothing wrong with QNX, and it doesn't by itself explain Lucid's buggy software.

--

If it's true that the majority of problems are key card or key fob related, then that's maybe on the third party doing that software/hardware and/or on Lucid's integration of that into the vehicle. That should be something addressable in 3 months, as I would insist if I were in charge.

 

[LVAGT]

I forget the name of the Tier 2, but Lucid's key card and fob software is done by a third-party company. The question is whether the bugs are that company's fault, or in Lucid's integration of that company's product (which is probably delivered as an ECU).

They use the same supplier as the Tesla Model S/X for the key fobs. That was told to me by the Lucid Tech about a year ago. Not sure if anything has changed. I had the Model X from 2020 to 2023, very rarely ever had any issues with my fob. In the AGT, I've noticed, it works great when there are no other cars around, but once I get into a school with lots of vehicles parked or a parking garage, it takes a good 3 long awkward seconds for the door handles to present themselves.

 

[tarator]

QNX has a number of advantages over Linux. Not just inherent real-time support, but isolation of processes, fault-tolerance, etc from its micro-kernel architecture. And, you can still program in common programming languages. Shoe-horning real-time performance into Linux has never truly worked, and those that use it simply live with less than a real-time guarantee.

Heck, Nvidia uses QNX for its self-driving platform, Drive, where real-time performance is safety critical. Nothing wrong with QNX, and it doesn't by itself explain Lucid's buggy software.

--

If it's true that the majority of problems are key card or key fob related, then that's maybe on the third party doing that software/hardware and/or on Lucid's integration of that into the vehicle. That should be something addressable in 3 months, as I would insist if I were in charge.

I don't know what isolation of processes does QNX have, but I know that Lucid's infotainment software crashes in ways that reboot screens. That doesn't sound very isolated for me.

For infotainment system real-time OS doesn't make any sense. Compute is embarrassingly cheap anyway and all (screen) UI seems real-time to humans.

Of course, for the real driving, the requirements are different and there might be need for something like QNX, although I personally don't buy it. That's why Nvidia's DriveOS supports both QNX and Linux.

As for the key fobs, my suspicion is that instead of designating a microcontroller that is always on scanning for them, they added it to another system that wakes up only few times a second and does a poor job scanning for signals.

 

[DreamTour]

For everyone spinning away and hand wringing at this video, as an owner, how does one influencers bad experience being documented affect your satisfaction with the vehicle?

I.e. if you were dissatisfied before, does a YouTuber being dissatisfied make you even less satisfied?

Re those who said lucid should front run the narrative he’s putting out. Maybe. But public grovelling no. He’s already committed a cardinal sin automotive journalism (I know he’s not an automotive journalist) but launching a complaint fest / airing out dirty laundry is not a way to get constructive engagement. Perhaps it was because he felt frustrated he was not getting attention from service (which is totally valid).

Lucids execution problems are real. But as owners you already know that. What you will get an auto oem to acknowledge publically is different from privately. If you have gone to any of the lucid hosted events and spoken with their senior management team, not only do they know where execution is lacking but acknowledge the need to improve. Knowledge of poor execution alone does not right the ship, nor does “throwing more people at it”

The leadership reshuffle should give you an indication of the board actually doing corporate governance and trying to get more effective management in place. In the auto oem world, this amount of leadership change in a short period of time is unheard of.

 

[SaaSManKS]

Tell me about it. We've had two Gravities, both on identical software. The array of software glitches they have -- and how different those glitches are between the two cars -- is absolutely baffling.

One of the Gravities got totaled in a wreck last weekend, and we have a third one on order that has been built almost exactly six months after the first one. All Dream editions, identical software. It will be interesting to see how it behaves compared to the other two.

this right here is the most interesting thing IMO, the vast differences in issues between the cars and software issues is so concerning

 

[hmp10]

They use the same supplier as the Tesla Model S/X for the key fobs.

That's interesting, because one of the most consistent software/hardware failures in our 2021 Model S Plaid was that the key fob periodically refused to work, sometimes for several days at a stretch. During those episodes we had to use the key card to enter the car and rub it over the phone charger pad to start the car.

Lucid switched to Marquardt (MB's supplier) for the Gravity key fobs . . . but we all know how that went.

 

[Smorgasbord]

They use the same supplier as the Tesla Model S/X for the key fobs. That was told to me by the Lucid Tech about a year ago. Not sure if anything has changed.

Lucid did use Pektron for the Air, but switched suppliers for the Gravity. Someone here posted who the new company was, but I don't have the link anymore.

as an owner, how does one influencers bad experience being documented affect your satisfaction with the vehicle?

Well, as a potential owner, my wife and I are rethinking whether to follow-through with our order placed back in Sept.

 

[Zinee]

Like most things, reactions to the video depend on your perspective. As a Lucid owner and regular forum lurker here, it had a "water is wet" quality to it. Most people I run into have never even heard of Lucid (when folks ask me about my Air, I tell them its the best car I've ever driven and I love it, even though the software and UI are buggy and can be temperamental). Anyone researching a possible Lucid purchase will quickly discover the software issues, and have to weigh those against the car's positives.

 

[socktucker]

I call this the rental car test. A car should be designed (as much as possible) to be an acceptable rental car. You roll in to Avis and rent a car, you jump into it, look around and evaluate the various controls and within 5 minutes you have a good handle on how everything works, start the car and drive off. No need to pull the manual from glove box (if there even is one).

 

[hmp10]

For everyone spinning away and hand wringing at this video, as an owner, how does one influencers bad experience being documented affect your satisfaction with the vehicle?

For me, it is less about my satisfaction as an owner and more about my satisfaction as a stockholder with a fair-sized investment in the company. My satisfaction as an owner was already in a tailspin over the continuing problems with the Gravity software.

I.e. if you were dissatisfied before, does a YouTuber being dissatisfied make you even less satisfied?

While I have been brutally candid about Lucid's perpetual software woes on this forum, everywhere else I have tried to be a brand ambassador. I post about it on YouTube comments only when I have something to say about Lucid's achievements with the design and engineering of the car. I have put every friend and neighbor that I could behind the wheels of our four Lucids (two Airs and two Gravities). I start by telling them that they're going to see some software glitches, but assure them that they will be taken care of via OTA updates. Then I turn to pointing out the amazing driving and riding experience they're about to have.

However, as the software problems continued, I lost my enthusiasm and quit talking the Gravity up to others. I actually became worried that one of them might buy one on my recommendation and then blame me for misleading them.

But, most of all, I have constantly been pushing my brother to move from his 2018 Tesla Model S to a Lucid Earth when it comes out. His main point of resistance had been the lack of FSD, but I recently saw a comprehensive test of the Air's most recent DreamDrive updates in which it performed very impressively. I promptly sent it to him, and he responded that it really had pretty much brought him over the hump. However, he had been taking note of the software bugs in the Gravity -- not to mention that it was on a drive up to Atlanta to visit him that our Gravity rear drive unit failed, sending us home in a Mazda rental car.

Then yesterday, shortly after I came across this thread and began watching the video, I got a text from him. He's a big "Engineering Explained" fan and had just seen the video and wanted to tell me that an Earth was now off the table for him. He's going to wait instead for the BMW iX3.

So, yes. This YouTuber being dissatisfied makes me even less satisfied with Lucid.

 

[electric]

If it's true that the majority of problems are key card or key fob related, then that's maybe on the third party doing that software/hardware and/or on Lucid's integration of that into the vehicle. That should be something addressable in 3 months, as I would insist if I were in charge.

Majority of software problems are not key card or fob related, even if many are manifesting as such to an average user. I'm 100% convinced that wide range of manifestations actually points to a single or a small number of related deeper bugs in vehicle communication protocol and/or real time events processing. In other words, multiple subsystems and low level controllers are failing to stay in sync or failing to communicate real time status changes to the central brain. Impossible to know if comms stack can be tuned up to perfection or needs to be thrown out and redone from scratch, but I feel confident it's all fixable in software once the right skills are acquired and mobilized by Lucid's management.

 

[DreamTour]

Lucid did use Pektron for the Air, but switched suppliers for the Gravity. Someone here posted who the new company was, but I don't have the link anymore.


Well, as a potential owner, my wife and I are rethinking whether to follow-through with our order placed back in Sept.

That’s fair, though personally I would weight experiences of 1) trusted parties 2) people without a commercial motivation I.e forum members over someone who has a commercial motivation to get views / stir the pot.

I don’t watch this guys channel, but you have to always wonder how much of it is embellished to tell a story. (Both negative and positive spun.)

 

[electric]

One thing I haven't seen mentioned in this thread is that you don't build a 4M follower YouTube channel on dry engineering alone, Jason is often quite dramatic and humorous in his videos, which makes them fun to watch. I noted plenty of dramatization in this video as well, having personally experienced most of his issues in my Gravity and loaner Air. These are real issues, but dramatized in a way that most people would not, although it really depends on one's mood and personality. My own Lucid experience ranges from ecstatic to "I'm going to drive it off the cliff and collect insurance" pretty much every day, depending on which bugs pop up that day and how my wife has treated me the night before.

 

[TahoeMichael]

Yeah, and I bet the first few iterations of the wheel weren’t that great either. But if a car company came out today with some weird shaped, rough riding wheels, would your reply be the same?

Tesla was in a completely different environment altogether than Lucid is now - zero competition then and it was exciting to see a company bring this fantasy to life, so they had plenty of time and people gave them the longest leash possible to get stuff to work right because it seemed absolutely impossible - and they were the only game in town. Even Tesla couldn’t do now what they did in the beginning because consumers aren't that patient and there are other options that didn’t exist back then.

If Lucid insists on following the same roadmap that Tesla created, they are going to fail, because that means that they didn’t learn anything about Tesla’s challenges and struggles. Excellent engineering won’t save them if they can’t get the other stuff right from the beginning because most people don’t care about the technical wizardry of engineers - they want a car that works all the time like the other cars they have owned for decades. Most potential customers definitely don’t want to hear “but Tesla had the same problems in the beginning”, because if Tesla doesn’t have those problems now, they can just go buy a Tesla (or other brands, or just a regular old ICE car). The path of least resistance is definitely the most travelled.

To be fair, Lucid didn't follow the roadmap of Tesla. Early Tesla hardware was a real shit-show. Individual vehicles could be fine, but the overall hardware engineering, manufacturing, and quality control was amateur hour. To a lesser degree, Tesla struggles with that even today. Lucid seems to be doing much much better in that regard. What they failed to recognize is that software is both important and hard, and requires a substantial investment.

Failing to recognize that software is incredibly difficult is a common problem with hardware people, and Peter Rawlinson is apparently not unusual in that regard. These days, hardware people usually know how to program. They whip out a couple thousand lines of code to solve some problem they have, and they think "gee, pretty easy compared to my day job!". They fail to recognize that a billion lines (or more!) of code is a completely different problem, particularly when that consists of multiple million+ line modules written by different organizations who never expected their modules to have to work together. They also don't understand that the discipline of software engineering just doesn't look like hardware engineering. The software industry has long prioritized feature bloat over reliability and quality, and that goes right down into the academic programs that produce new software engineers. Disciplines like design-for-testability effectively don't exist in the software world. That means that when you actually want a highly reliable software system, you're engaging in hire-wire work, while engineering for reliability is table-stakes in the hardware world.

 

[MGRMLN]

Lucid created something great. But my concern is the with perpetual software woes.
As vehicles become increasingly software defined, this hits especially hard.
No one argues how brilliant the cars are from a handling/dynamics standpoint.
But the software issue easily overcome any of that in daily use.
Silly things become a major pain. My persistent "last audio source" issue with Native Tidal damages my enjoyment of my car.
A car at this price point should just never have this issue. Full stop.
I sincerely hope Lucid is listening and agree that they should reach out to EE (Jason Fenske) and solve his car's issues.

 

[tarator]

QNX has a number of advantages over Linux. Not just inherent real-time support, but isolation of processes, fault-tolerance, etc from its micro-kernel architecture.

BTW, the tempo of the sound for the turn signals the car keeps tells me everything I need to know about the real-time-ness the car's OS has. Or rather the lack of it.

 

[TahoeMichael]

Actually, Yamaha outsells Harley handily when you consider all motorcycle offerings. More importantly, Harley's customers are literally dying off and new customers aren't coming into the fold.

In terms of Lucid's market, Tesla has essentially stopped trying to compete there. Sales of Model S and X are down from their heyday, and Tesla really isn't giving those models the attention they need. Back in 2020, Tesla upped Model S's range after Lucid Air came out with a 400 mile range. Back then, Tesla would make performance versions and run them on The Ring for bragging rights. They now instead put most of their efforts into the 3/Y, with new engineering going into the CyberTruck (steer by wire, 4 wheel steering, 48 volt architecture, etc.), with no indication of that tech moving into the S/X supposed flagships.

What I don't get is why Lucid feels the need to spend big bucks on advertising with famous movie stars when they not only don't make cars that many people in this price range will tolerate, they actually don't make enough cars for the demand they do have already. I've been waiting almost 3 months now for my build - with the only news being that if Lucid doesn't make my car in time I lose the deal they promised me when I ordered it.

This is no way to run a car company. Lucid's lucky my wife hates Elon as much as she does.

I'm all for bashing Lucid over their quality problems. But be real: the Lucid Air is outselling every other luxury sedan EV. Apparently the target market does fine value in the product. With the Gravity, it's too early to say. People have long bashed Lucid about their lack of marketing, and that too many potential buyers have never heard of the brand. It's not unreasonable for them to decide to increase market awareness.

 

[TahoeMichael]

BTW, the tempo of the sound for the turn signals the car keeps tells me everything I need to know about the real-time-ness the car's OS has. Or rather the lack of it.

I think you're just trying to be funny, in which case:

But turn signal sounds aren't a great way to assess the OS. Unless they made turn signal sounds a high priority process (which would be weird), you could expect to hear some jitter.

 

[tarator]

I think you're just trying to be funny, in which case:

But turn signal sounds aren't a great way to assess the OS. Unless they made turn signal sounds a high priority process (which would be weird), you could expect to hear some jitter.

Actually, the turn signal sounds are the perfect way to asses the OS. "Some jitter" is not something human should be able to hear. "Some jitter" means millisecond more or less. That's not what I hear.

I really wish I was able to see if the blinkers follow the sound, but I am too lazy to set that experiment up. My suspicion is that the tempo (or the lack of a tempo) is the same.