Tesla uberbulls often like to say that Tesla is the leader in self-driving because while it doesn’t have a commercially available autonomous ride-hailing service like Waymo, it doesn’t rely on geo-fencing and mapping like Waymo.
They argue that if Tesla wanted to do that it could, but it prefers to focus on an autonomous system that could drive anywhere, anytime, without mapping.
However, it is questionable that they could do it if they wanted to because they still haven’t done it on a project much simpler than Waymo’s operations in Pheonix and other cities: the tunnels under Las Vegas.
The Las Vegas Convention Center Loop is The Boring Company’s first full-scale loop project currently in commercial use.
Elon Musk’s tunneling start-up completed the $50 million project in just over a year.
A Boring Company Loop system consists of tunnels in which Tesla electric vehicles travel at high speeds between stations to transport people within a city. The Boring Company said that it was working with Tesla to use its self-driving system inside those tunnels, which would enables to get rid of the current drivers and lower the cost of operation.
However, 2 years and several more tunnels connected to the Loop later, The Boring Company is still using drivers in the tunnels.
The last paragraph sums up my own thoughts pretty well
Why hasn’t this problem been solved yet?
This is a very good question. Elon why are you so pathetic at self driving?
Anyone is free to go to YouTube and see how well the current version of FSD does in traffic. To imply its performance is anything short of amazing in the year 2024 means they’re either uninformed or lying. It’s not flawless and it’s never going to be but it’s quite safe to say it drives much better than the average human driver and when it fails it’s virtually always that it got stuck - not that it caused an accident.
As a heavy FSD user in a car I hate that I own at times I…mostly agree with you. It can take me flawlessly to and from the store or work on a regular basis with a mix of freeway, city streets, unprotected turns, pedestrian activity, etc.
But dear god sometimes it is just dumb. Like getting into the opposite lane of the turn lane it needs to be in now and then bailing on its current route and taking some longer one (I’ve now anticipated when it will happen and will just make it stop, but c’mon).
It’s also safer than a human at times because it drives like a slug going uphill on a hot day. It’ll sometimes stop too long, inch forward too slow, wait too long to go, or whatever. And I get, don’t get me wrong, better safe than sorry, but 1/3rd of the time I just take over because I need to goooo.
edit: What’s confusing to me is if no cars are around, it will not always choose the same behavior in the same exact situation. Most notably the turn into my neighborhood, sometimes it just decides to miss it. Sometimes it decides to route elsewhere first. Then I’m like…yoooo, you were so close and there are no cars, no people, it’s light out, wtf are you doing.
This sounds like a problem setting for physics 101. Is that why they expected the hyperloop to work in a vacuum?
Might have something to do with Musk’s insistence on not using LIDAR. The cameras probably struggle with the lack of distinct / unique features in the tunnels.
That’s what I’m thinking. It’s probably some obstacle created by management, not a technical issue.
Didn’t they go with cameras only for their automation and like zero radar/lidar that could probably sort this out?
I believe they did. So why not paint a pattern on the walls for the cameras to follow? Elon should be ashamed that this problem hasn’t been solved in two years, but he’s a shameless jackass, so…
I hope, at least, that it’s his company that’s responsible for paying the drivers, and not the city.
Because it’s not an easy problem to solve.
The issue is that morons like Elon want people to believe that it is.
I’ll be the first to acknowledge that I’m a moron on this and many other topics but…
This is an environment that doesnt include weather, pedestrians, many other cars, or other obstacles. I feel like my 4 year old Honda could almost manage that. It can follow a lane at a set speed without me actually driving. It can’t manage obstacles or signs or anything like that so it clearly isn’t self driving and I’m not claiming it is. The only hard part would be the intersection between two tunnels but I feel like that part has already pretty much been addressed by Tesla FSD tech.
Part of the problem is to meet their quoted throughput of passengers they would need to fully load/unload each vehicle in ~30 seconds. 4 adults in, 4 out, with luggage, with no delays or struggling. That’s… not very feasible for a commercial passenger car. They’re not designed for quick loading and unloading.
The tunnels are a single lane without a service tunnel, which the Victorians used in the 1800s for their subways. Because if a single car has mechanical issues the entire service has to stop and empty to clear it. They’re electric, so there are less mechanical systems, but they are still putting a significant amount of wear and tear on tires/axels/steering systems, all the mechanical systems they still have. Even without meeting the their goal throughput, they’re putting orders of magnitude more use on each vehicle, which are consumer cars. They’re meant to spend most of their lives parked.
If they made a “Tesla train/trolly” where the engine car was pulling a simple enclosed cart with seats it would significantly improve their throughput and loading times, and require less maintenance per passenger. But at that point you’ve just invented a train that uses significantly less efficient rubber tires on asphalt.