4 ms·
It is a big story, and something about it isn’t adding up. Autopilot wouldn’t engage on the road as outlined, and even if it did wouldn’t exceed 35... that pl
by reasonabl_human 5y ago
It is a big story, and something about it isn’t adding up.
Autopilot wouldn’t engage on the road as outlined, and even if it did wouldn’t exceed 35... that plus the driver would’ve had to activate it then leave to the back seat, and ask the passenger to tug the wheel every 15-30 seconds to keep it active.
I’ve never tried to leave the drivers seat with autopilot engaged but I doubt it would work. Would be willing to try just to put some of this misinformation surrounding the news to rest.
Even if it was engaged, this is a massively human error, not autonomy error. It’s a beta service- you have to hold the wheel and pay attention. It isn’t L5 FSD.
- camjohnson26 5y agoHere’s a video of someone not only exiting the driver seat, but exiting the CAR while autopilot is engaged, and almost getting locked out: https://youtu.be/Z_N5zrFJgII https://youtu.be/Z_N5zrFJgII Here’s another video showing autopilot engaging with no lane markings in a neighborhood and setting the target speed too high: https://twitter.com/LyftGyft/status/1383917552762384386?s=20 https://twitter.com/LyftGyft/status/1383917552762384386?s=20
- reasonabl_human 5y agoThanks for the videos! For the first one: The conclusion as outlined by the producer was that Tesla has implemented restrictions around how autopilot can be used and that it is safe when used correctly. (Someone actually in the seat, holding the wheel, paying attention). He had to jump through a significant number of hoops in a controlled environment to get exiting the car to work. The autopilot disengages when you unbuckle or open the front door, which sounds very reasonable to me. As he stated, idiots are going to be idiots, you can also get out of a moving plane if you try hard enough... Also, he didn’t ‘almost get locked out’, that is an exaggeration. He just observed that the car won’t open the doors while driving. Pretty standard. For the second one: That’s concerning that autopilot would allow you to engage without lane lines at such a high speed. What’s more concerning is that any human would engage it in that situation. You can technically land a small plane with Garmin autopilot systems but it’s really only used for emergencies when a pilot is incapacitated. Should such features be disabled? Or should it be up to the end user to not be an idiot when situationally applying the technology? As all Tesla owners know, this isn’t FSD level 5 and it’s up to the driver to situationally deem the environment safe to use autopilot. Pretty much limited to highways and large multi lane roads at this point. From my experience of ~5k autopilot miles it works very well in those environments. I’d never engage it on a backroad or parking lot etc. even if it ‘let me’. What happened to the men in this recent crash was tragic but easily avoidable.
- camjohnson26 5y agoI see where you’re coming from but have a few major disagreements. First you say that all Tesla owners know this isn’t FSD level 5. The occupants were 70 and 60 years old, I’ve seen plenty of situations of technology blind spots with older acquaintances. That’s why vehicle safety often includes the concept of predictable abuse, features that will predictably be abused. Pair this crash with all the TikTok videos with no one driving and even Elon driving hands free and claiming self driving is imminent. And I think it’s pedantic to say he didn’t almost get locked out, not sure what else you would call not being able to open the door. If that other car wasn’t there he would have had to climb through the window.
- reasonabl_human 5y agoI think we are ultimately on the same page.. as far predictable abuse goes, the seatbelt and door detection strikes me as reasonable. Adding detection for weight on the front seat would be another good call but could further be defeated with a dumbbell. My point was that it isn’t up to Tesla to ‘baby-proof’ a potentially dangerous feature, but to make reasonable safety precautions, which it seems they’ve done. Totally agree on the technology blind spot point and think the autopilot capability is over-hyped and the marketing angles could be dangerous when considering non-tech or older folks think it’s autonomous. However, that technology blind spot is addressed with the above precautions; if an elderly woman kicks it into autopilot thinking she can then leave the car or leave the drivers seat, the safety features will prevent her from doing so. These guys had to ‘trick’ the vehicle in order to function as it did, and evidently evidently from the video the tricks were fairly involved and required a lot of understanding how the system worked, so I don’t think the blind spot argument can be used. It isn’t like an elderly woman started napping behind the wheel because she had a false sense of safety (also wouldn’t work bc you need to actively touch the wheel), they had to have intentionally circumvented safety checks and left the seat. The ‘locked out’ bit is less relevant, agreed probably pedantic both ways, but to me locked out implies being unable to access your car without a locksmith or the like. In this case it seemed to further imply that the car was in a bad state, and mistakenly wasn’t letting its owner in... but that’s not the case- it was in drive and thus locked, it would be dangerous to have the car unlocked from the outside while driving, or from the inside too, preventing bumping the open button while groping for the window switches etc. No he wouldn’t have to had climbed through the window... autopilot disengages when you take the wheel, all he had to do was grab the wheel a bit. He mentioned this actually. Or reach inside and pull the stalk up... or adjust the speed to zero on the wheel control knob.. regardless it isn’t relevant to the discussion at hand IMO. The first video proves that these guys in the crash intentionally defeated the autopilot safety checks, nothing more. The second video, on the other hand, shows that Tesla has a long, long way to go to bring autopilot to a generally viable control system. So thank you for sharing these videos. I’ll pay even more attention on the highway now...
- Decker87 5y agoAs long as it's called "autopilot" it's going to be a problem.
- reasonabl_human 5y agoI disagree. It is an autopilot that works well when applied correctly, it is not generally viable full self driving which is what I think people tend to hear when they read ‘autopilot’. Wikipedia: “An autopilot is a system used to control the trajectory of an aircraft, marine craft or spacecraft without requiring constant manual control by a human operator. Autopilots do not replace human operators. Instead, the autopilot assists the operator's control of the vehicle, allowing the operator to focus on broader aspects of operations.” The problem is when people don’t understand the difference between an autopilot driver assist system and a fully autonomous vehicle.
- alphabetting 5y agowould you agree though that "full self driving" is not such a great name for a feature which isn't full self driving? Telsa's are rated as very safe cars, but this marketing tactic is definitely not.
- reasonabl_human 5y agoYes, I very much agree with that. ‘FSD Capability’ in its functional form doesn’t exist yet, people are buying it in advance to access beta autopilot enhancements which are then problematically referred to as ‘FSD’ colloquially. It would be more informative for Tesla to clarify these or change the naming convention... but it’s a convenient hype angle, probably why they keep this naming setup. The issue seems to be when people think that these limited beta enhancements of autopilot are FSD, when in fact, nobody has FSD yet. Tesla doesn’t do marketing in a traditional sense so I don’t think it’s a marketing failure but considering the dangers of confusing this FSD sentiment it should be clarified abundantly when signing up, signing waivers, something like that...