DD Pro on 3.6.4 Not Passenger Friendly

[Smorgasbord]

OP here, let me address some of the responses:
1) These construction "zones" were often just a few cars lengths long. Maybe they were doing median or wall repair. Most of the time the lines were still here.
2) Sometimes it wasn't a construction zone, just faint or missing lines.
3) I ended up keeping my hands on the wheel the whole time so I could take-over instantly as required by the car. In this mode, it was my passengers that were annoyed by all the chiming, thinking that I, the driver, was doing something wrong.
4) Comparing the system to a toddler misses that Lucid isn't charging for a toddler-capable version ($6,750).
5) The pre-mapping issue is "inside baseball." What Lucid is apparently doing is pre-figuring out how the car should drive each section of roadway, and then revert to driver if any kind of deviation from "normal" is detected (perhaps cones, perhaps anything orange on the shoulders, etc.). But, what it should do is look to see if it can still identify a clear, clean lane in which to drive, and then do so. Pre-mapping also should compensate for lane lines being faint or even missing for a short spell. It know the road from the pre-map, and the faint/missing lines were, in the vast majority of cases, there during the pre-mapping, so the map should have the path for the car to follow. But it doesn't. It's not smart enough to look ahead 75 feet to see the lane lines continuing and decide to drive between that span. The bottom line is that if Lucid will always require pre-mapping, but yet provide no advantage to that pre-mapping, what's the point? Tesla FSD drives these roads better without pre-mapping and without LiDAR.
6) Saying we're a decade away from autonomous cars misses what Waymo is doing today, and what Tesla is starting to do today. And even if so, how does Lucid get away with charging almost $7k for this?
7) I'm not asking for complicated urban environment driving, that has pedestrians, bicycles, stop lights, round-abouts, etc. This is supposed to be the easy drive me on the highway between entrance and exit (non-inclusive). If it can't handle these normal highways (280, 680, 80) without annoying passengers and requiring drivers to be hands-on at all times, then it's not that useful, not worth the money, and other car companies will be more successful.

 

[Aka-none]

So you’re driving along with well painted lines and then a Jersey Barrier encroaches on your space. And you want it to recognize there’s probably still room to continue on at 70ish? Rather than having it require you to participate?
And apparently tesla fsd will do this? And you would like the Lucid to behave the same way?
I personally don’t care to accept the risk that tesla should be building into their system, or not continuing to over promise. All the risk of failure is on you because you thought it really was fsd l4/5 and not the level 2+ it is.

 

[Smorgasbord]

So you’re driving along with well painted lines and then a Jersey Barrier encroaches on your space. And you want it to recognize there’s probably still room to continue on at 70ish? Rather than having it require you to participate?

Not what I said happened, not what I proposed.

Nothing on that trip encroached into my lane space. I want DD Pro to recognize that and continue on. That means not disengaging because there are some orange cones sitting on the side of the road. And certainly not disengaging without adequate warning and time for me to get my hands back on the wheel. In J3016, SAE normatively talks about "several seconds" of time for such degradation of driving level.

Another example: on a straight, or almost straight roadway with good daytime visibility, some of the upcoming painted lines are faded. There is clear visibility to clear lines just the faded few of them. I want DD Pro to recognize this and continue on.

 

[Saksak]

[…] There is clear visibility to clear lines just the faded few of them. I want DD Pro to recognize this and continue on.

I’m sure the autonomous driving division is well aware of the current limitations and is working hard on the next generation of DD Pro software. Today, DD Pro needs clear lane lines, and whether we like it or not, we have to accept it and drive accordingly.

I’m hopeful that at some point in the near future, we’ll get a significant upgrade in Dream Drive capabilities. We know that work hasn’t stopped with the new executive management.

 

[joec]

With any of these systems, there's going to be variance depending on where you are driving. Where I drive in Northern CA, so far Dream Drive 2 has been quite good. Even the auto lane-change, which I thought I would hate based on past experience with my Tesla, is very sensible. I only use it on major highways, not smaller two-lane roads. And I always take over when I'm about to get into something even remotely dicey. And I'm ready to take over at all times, regardless.

It feels like what some folks really want is a full-time driver. Sit in the back seat and let someone else do all the work. That's cool, if that's your thing. But it doesn't exist yet. Period. Some companies are closer than others to achieving that. But none of them are there yet. Pretending they are or expecting them to be is dangerous in the meantime.

For me, Dream Drive 2 is exactly what I want. Something to help on long road trips and especially in traffic jams with the monotony of the road.

I have a very simple ritual on work days. In the morning, I drive myself to work, old school. I leave early enough that there's little traffic, and I just eat up the pure joy of driving a beautifully engineered machine.

On the way home, in the late afternoon, where the traffic is going to be clogged up several times over the course of the 50-mile trip, I set Dream Drive, turn on my massage seats, pick out a good album or podcast to listen to, and let the worries of the work day fade away. Once in a long while, when someone does something totally unpredictable in front of me, I end up having to take over. More often than not, Dream Drive takes me all the way to about mile before my exit, where I take over and finish the drive home. Simple.

 

[Bdr100]

Road tripped from Houston to Austin this weekend (Eric Clapton Crossroads guitar festival - great show!) and had very mixed DDP2 results. While on I-10, handsfree was very on/off. I understand the construction zone thing but a lot of the disengagement was random. Did not stay engaged for more than 3-4 miles at a time. Like poster above, I got in habit of just keeping hand on wheel to be ready every time hands-free disengaged. Then when I turned off onto State Hwy 71, which is a divided double lane each way hwy, I somewhat expected hands-free to return to hands-on, but instead the grayed out steering wheel kept flashing on/off for 30+ miles before it finally went solid allowing me to engage. Before leaving for home yesterday I did a system reset and return trip was better. I came home different way (Hwy 290), so not sure if that was due to reset or different conditions. Looking forward to the NVIDIA system coming to Gravity in a few months (hopefully) that should be a significant upgrade on hands-free capabilities.

My 10 year old Mercedes GLE that I sold for this Gravity had very early generation ADAS but at least it was very consistent. It did good job lane centering and keeping distance from cars ahead. It could not handle sharp turns and tended to want to veer into left turn around lanes on divided highway. The key was that it was very, very consistent and you did not have to worry about it suddenly disengaging.

 

[Macgreiner]

Getting at least Autopilot equivalency will be important if Lucid wants to pull Tesla owners over.