Park Assist just ruined my front bumper - cut too short!

[pchilds]

I could tell that he was going to hit the pickup, before it hit, I guess I am the only one here that is not blind or blinded by wanting to make Tesla responsible for this driver's actions.

 

[u00mem9]

Troll? Can't be serious.

 

[Tam]

I could tell that he was going to hit the pickup, before it hit, I guess I am the only one here that is not blind or blinded by wanting to make Tesla responsible for this driver's actions.

It is nonproductive to assign blame to owners or Tesla when there is already a waiver saying that an owner is responsible when Autopilot (self parking is a subset of Autopilot) is in use.

The right thing to do is to disclose problems so others can become knowledgeable of the current system's limitations which is what exactly Philippe did here.

The next thing to do is for engineers to plan how to improve on the system to solve as much as possible all those limitations.

 

[trils0n]

How will this crash scenario be prevented in future upgrade/improvement?

Off topic (kinda): NHTSA should require all trucks to have rear bumpers. Many semi trucks and trailers (legally) don't have bumpers/rear impact guards (like in this unfortunate event). Really dangerous for cars to be able to get under the trailer in the event of a collision. Impact goes about face level and avoids crush zones/air bag deployment, etc. Not sure what Tesla could do, but it would be safer for everyone if the trailers and trucks had better safety measures (and bumpers would have the side effect of being easier to see by current parking sensors)

 

[RogerHScott]

I could tell that he was going to hit the pickup, before it hit, I guess I am the only one here that is not blind or blinded by wanting to make Tesla responsible for this driver's actions.

We've got the video right there. Please cite the time mark at which you make this observation and what, specifically, you observe at that point. Thanks.

 

[Canuck]

I could tell that he was going to hit the pickup, before it hit, I guess I am the only one here that is not blind or blinded by wanting to make Tesla responsible for this driver's actions.

Right. I just wish you were driving. I'm certain you would have intervened in time. You're so good, I wonder why you even need to use this feature? (And I'm the one here who is often called a "fanboy" of Tesla?)

It is nonproductive to assign blame to owners or Tesla when there is already a waiver saying that an owner is responsible when Autopilot (self parking is a subset of Autopilot) is in use.

Waivers are not always upheld in court and there's a good reason for that. Do you want to live in a society where if you sign a waiver before you do an activity, regardless of the negligence, or even gross negligence, of the waiver provider they are not liable? I sure don't. And we have decided, collectively as a society, by way of our common-law precedents, that we don't want to live in that kind of a society. For example, you sign a waiver to do a bungy jump. In your society, there's no real penalty for the owner to provide proper equipment, or do safety checks or maintenance. Yes, if he kills someone, he probably won't get any more business but he can take his money and run, no harm done, no damages to be paid, he has a signed waiver. Sound good to you? Yes, according to what you posted above. So then you jump, end up paralyzed because of an undersized cord, and find out that the owner used a cord not suited for bungy jumping in order to save money? So the judge says to you: "tough luck, you signed a waiver". Or do you want to live in a society where the courts will say that kind of waiver goes against the collective values of our society? Now, if you throw your back out, or even become paralyzed because of the jump where everything went as planned, then the waiver has a good and valid purpose. The shop did nothing wrong and you understood the risks involved. I don't accept the risks of an owner failing to provide proper equipment regardless of any waiver I sign and I don't accept this accident as being a risk I accepted by using Tesla's parallel parking feature. You may, but I do not. And if it happened to me, and Tesla didn't pay, I would commence a Small Claims action and let a judge decide who is at fault and how much weight any waiver holds. I can almost guarantee you, though, that it would never reach a courtroom.

 

[jeffro01]

I'm not trying to be antagonistic, but you didn't answer my question. The reason I asked is that if you are encountering these same circumstances and not having these problems then you might be able to share with us your "secret" to making it work, which is what we
all ultimately want (not to tear down Tesla). Btw, the ("perceived") problems with parallel auto-park go way beyond the topic of this particular thread.

Ah... Okay, that's a my bad on me...

I'm not doing anything special or using any tricks. The car, to date, has fit into every spot it said it could without hitting the curb or the surrounding vehicles.

Jeff

 

[RogerHScott]

I'm not doing anything special or using any tricks. The car, to date, has fit into every spot it said it could without hitting the curb or the surrounding vehicles.

Jeff

The hitting things issue is, to my knowledge, an extremely rare occurrence. It is not the primary basis for my "bridge too far" claim. That
claim is based on the large, growing, and ill-defined set of circumstances in which it simply doesn't work at all (as I said above,
certainly the preferred failure mode, but...). So do you find you can routinely use it on streets with traffic behind you? For you does
it routinely recognize non-square curbs? Does it never claim to see a parking spot in the middle of traffic, such as at a stop light?

 

[TaoJones]

Autopark has badly missed the last couple of parallel parking attempts at a curb area at which it did fine a couple of weeks ago. Without abrupt intervention on my part, it would have driven the right rear wheel/rim into the concrete curb and it wouldn't have been a near miss, either. That's $200 of damage avoided (standard quote from Tesla for rim repair, per rim). $400 if it would have failed with the right front as well - although the impact of the rear would have stopped the car one presumes - it's a high enough curb.

Disappointing, as Autopark had worked more than 50% of the time in the past. Not comfortable demo-ing it now for prospective owner/passengers.

Will give the ownership line a call to have the logs reviewed for sensor faults.

Time to park at an SvC for 20 minutes to facilitate the latest patch. Perhaps avoiding hundreds of dollars of wheel damage qualifies as a "minor bug fix or enhancement".

Bleh.

 

[Zybd1201]

I disagree completely, I've used it well over 100 times without fail, without issue, without even a close call in tight spaces. I'm not saying it's perfect, but to say its "a bridge too far for Tesla" is absurd.

Jeff

This is exactly the problem, Jeff. When your autopilot fails and actually causes damage I'm confident you will feel differently. However, it is easy to say "this hasn't happened to me yet therefore...."

The autopark and at times autopilot act in a way which doesn't allow human intervention to occur soon enough to prevent damage. In that scenario, where no reasonable driver could be expected to intervene soon enough to prevent that damage then no amount of disclaimers will protect Tesla and the long list of disclaimers they already have essentially say, "this doesn't work as intended." The only possible way to not be liable for Tesla's software and hardware faults is to simply not use their systems. Otherwise, they will abruptly place the blame on you saying "we told them it doesn't work!"

Doesn't that seem kind of strange to you?

 

[Zybd1201]

Will give the ownership line a call to have the logs reviewed for sensor faults.

Bleh.

Let me know how that goes. It took me a month and an uncountable number of calls, reports, emails, and various other methods to get Tesla to actually inspect the logs. In that process they SAID they inspected them but did not actually expect them. Make sure you get a legitimate answer and keep pressing until they actually inspect your logs and give you specifics.

 

[K5ING]

RHD cars need to park left-side in UK, Australia and Japan, etc. Needs to operate as software that can "flip the circumstances" and work the same whether left or right. In computer programming, it can come down to a negative sign on a formula rather than positive when the car location is mapped out on the X and Y axis. If anyone has ever "programmed Pong" in any computer language and worked on X/Y coordinates, you know what I mean.

I'm a retired CNC programmer, which is why I asked about this. I was asking whether the hardware and software was the same for LHD and RHD cars, or if it's optimized for parking on the right for LHD cars and on the left for RHD cars. Since he was parking on the left in a LHD car, I thought it may have contributed to the bump. If there is no optimization for LHD/RHD, then it isn't an issue.

 

[RogerHScott]

Let me know how that goes. It took me a month and an uncountable number of calls, reports, emails, and various other methods to get Tesla to actually inspect the logs. In that process they SAID they inspected them but did not actually expect them. Make sure you get a legitimate answer and keep pressing until they actually inspect your logs and give you specifics.

If you really wanted to be a snot about it you could ask them to tell you something from the logs that only you would know (without actually looking at the logs), such as what date some unusual event occurred on.

 

[TaoJones]

This is exactly the problem, Jeff. When your autopilot fails and actually causes damage I'm confident you will feel differently. However, it is easy to say "this hasn't happened to me yet therefore...."

The autopark and at times autopilot act in a way which doesn't allow human intervention to occur soon enough to prevent damage. In that scenario, where no reasonable driver could be expected to intervene soon enough to prevent that damage then no amount of disclaimers will protect Tesla and the long list of disclaimers they already have essentially say, "this doesn't work as intended." The only possible way to not be liable for Tesla's software and hardware faults is to simply not use their systems. Otherwise, they will abruptly place the blame on you saying "we told them it doesn't work!"

Doesn't that seem kind of strange to you?

Hear, hear.

Let me know how that goes. It took me a month and an uncountable number of calls, reports, emails, and various other methods to get Tesla to actually inspect the logs. In that process they SAID they inspected them but did not actually expect them. Make sure you get a legitimate answer and keep pressing until they actually inspect your logs and give you specifics.

Will do. Congrats for enduring the obligatory brain damage. I've had to escalate past concerns around and through multiple SvCs and layers of management to eventually resolve concerns. They may punt this particular one until AP2.0 (more and better sensors), but it won't hurt to ask in the meantime with regard to my specific chariot.

 

[dusterbuster]

Probably goes without saying, but the parking assist limitations are not unique to Tesla. I stopped using the parking assist on my wife's BMW i3 because of its tendency to want to cause curb rash on the rims.

 

[jeffro01]

This is exactly the problem, Jeff. When your autopilot fails and actually causes damage I'm confident you will feel differently. However, it is easy to say "this hasn't happened to me yet therefore...."

The autopark and at times autopilot act in a way which doesn't allow human intervention to occur soon enough to prevent damage."

Doesn't that seem kind of strange to you?

Um... I disagree completely. In fact, that's complete BS. A tap of the break disengages both systems, it's not as if there is some major process you have to go through...

In this situation specifically, my depth perception sucks so I would have been paranoid of the truck and would have hit the break well before the car hit the truck. There was ample time to react. I'm not accusing the OP of anything, some people have better depth perception which could lead them to believe that it would be close but not actually hit anything. I'm just saying for me, I'd have hit the break before the bumpers collided. To date, I've never had to interrupt auto park out of fear of striking the object. To date that is.

Your making an assumption of how I would feel if X, Y, or Z happens... You are completely unable to make that assumption. I am fully aware of the limitations of the system, how it works, where it works well, where it doesn't, and the fact that at any moment I might have to intervene.

Jeff

 

[jeffro01]

The hitting things issue is, to my knowledge, an extremely rare occurrence. It is not the primary basis for my "bridge too far" claim. That
claim is based on the large, growing, and ill-defined set of circumstances in which it simply doesn't work at all (as I said above,
certainly the preferred failure mode, but...). So do you find you can routinely use it on streets with traffic behind you? For you does
it routinely recognize non-square curbs? Does it never claim to see a parking spot in the middle of traffic, such as at a stop light?

To date, which I feel like I must preface in this thread, here are my experiences to your questions.

1. Yes, without issue.
2. Don't know, the curbs I've used the feature on to date have all been square.
3. Yes. Granted, I know there is no parking spot there but yes the P has come up in certain traffic situations, such as at a stop light, which I have chuckled at and will continue to do so. I certainly haven't tried to use it in traffic... LOL

The most common problem I have with auto park isn't in the parking itself, it's my preference as to where the car sits in the opening. I usually have to adjust it from a front\back clearance standpoint. On the flip side, I've been impressed on how well it handles double opening parallel parking spaces and placing the car in one of the open spots, not directly in the middle of the two...

Jeff

 

[Gizmotoy]

How will this crash scenario be prevented in future upgrade/improvement?

After an owner who didn't know that he activated the summoning and left the car unsupervised and it crashed into the top front a tall semi-trailer bed,

Tesla revised the procedure to include another acknowledgement step to make sure driver is aware of summoning and is to hang around to babysit the system to make sure an owner would intervene to stop the car from crashing in the same scenario.

Could it be that Tesla has problems of crashing onto taller obstacles such as the mark of bumper damage is just above the sensor in this parallel parking case?

Should more sensors be added, and may be at a higher locations to detect taller obstacles such as pickup and semi trucks?

This is pretty easy to test yourself. I played with mine by holding a box in front of the car (with a driver you trust to not run you over). I don't have the Autopilot sensors, but my ultrasonics have a shockingly small vertical field of view. They miss anything besides giant curbs, and split-rail fences can be an issue if the horizontal crossbars are higher than about where the hood starts.

That said, their horizontal view was pretty good. They have trouble with thin objects, but there don't seem to be any gaps in coverage once the object is about a foot wide or so.

 

[GlmnAlyAirCar]

I'm thinking if that plastic step thingy wasn't there it probably wouldn't have happened. My take away is avoid PA when pickups are there but out of curiosity has anyone else had an issue using this feature when pickups or other higher vehicles are in the space ahead or behind?

FYI just took delivery of 60D and except for the test drive haven't used parking assist yet.

Trailer hitches concern me.

 

[jeffro01]

Trailer hitches concern me.

Me too... I've used the auto park in one instance where the front car\truck had a ball hitch sticking out the back so I was extra vigilant. Car never came close to it and appeared to "see" it given the ultra sonic visual representation.

Jeff