Robotaxi Hype

Tesla Robotaxi unveil on 8/8

— Elon Musk (@elonmusk) April 5, 2024

Musk made the above Tweet on April 5 in response to a Reuters article that claimed Tesla would abandon an entry-level EV priced at $25,000 or less and instead focus on a robotaxi.

Musk denied the rumor, then essentially confirmed it with the above Tweet.

The big problem for Musk is Tesla’s self-driving capability is nonexistent.

Full Self Driving (FSD) Test

.@ElonMusk calls @Tesla Full Self-Driving v12 "mind-blowing." But see for yourself how 12.3 fails 4 times on our first drive. Don't be fooled by the Cult, FSD will still try to kill you.

The Cult's praying for FSD licensing deals but not even Yugo would use this pile of trash. pic.twitter.com/K2z50Md1Kg

— Dan O'Dowd (@RealDanODowd) April 5, 2024

Yes, I would describe that as mind-blowing, but in a negative way.

It’s easy to find videos touting FSD but that refutes all the claims. You cannot find a similar video for Waymo.

The Grandma of Robotaxis Is Winning

Bloomberg reports The Grandma of Robotaxis Is Winning

The Alphabet-owned robotaxi company recently announced that it plans to unleash its cars onto the Phoenix freeway soon. The news is a game changer for Waymo employees, some of whom use the cars to get to work. Until now, riders could only take surface streets to get to their destination. Adding the freeway into the mix has the potential to cut their commutes by half.

As it stands, Waymo operates in only two cities — Phoenix and San Francisco. But it’s expanding into Los Angeles and Austin very soon.

Elon Musk first attempted to cut corners in 2014, when he rolled out “Autopilot” at Tesla, which Dave notes was neither auto nor a pilot. Last summer, an investigation into the program said it had played a role in 17 deaths and 736 crashes in the years after its rollout. By winter, Tesla had to recall more than two million of its vehicles to make sure drivers weren’t using Autopilot incorrectly.

Waymo Has 7.1 Million Driverless Miles

The Verge reports Waymo Has 7.1 Million Driverless Miles

Waymo analyzed 7.13 million fully driverless miles in three cities — Phoenix, Los Angeles, and San Francisco — and compared the data to human driving benchmarks to determine whether its cars were involved in fewer injuring-causing and police-reported crashes. And it was the first time the company studied miles from fully driverless operations only, rather than a mix of autonomous and human-monitored driving.

The conclusion? Waymo’s driverless cars were 6.7 times less likely than human drivers to be involved a crash resulting in an injury, or an 85 percent reduction over the human benchmark, and 2.3 times less likely to be in a police-reported crash, or a 57 percent reduction.

The company’s main competitor, Cruise, [Not Tesla] has paused operations nationwide after a crash in San Francisco resulted in a pedestrian being dragged 20 feet by one of the company’s driverless cars. Cruise allegedly withheld video footage of the incident from regulators and is now facing up to $1.5 million in fines from the state.

\Waymo’s millions of miles were not totally incident-free. The company said that in total, over the entire 7 million-plus miles in all three cities, its vehicles were only involved in three crashes that resulted in injuries: two in Phoenix and one in San Francisco. All three injuries were minor, according to Kristofer Kusano, safety researcher at Waymo and a co-author of the study.

The human benchmark is 2.78 incidents per million miles. Waymo’s benchmark for its driverless vehicles was only 0.41. The analysis comes on the heels of a study that Waymo published in conjunction with Swiss Re that found that the company’s driverless vehicles reduced the frequency of bodily injury claims by 100 percent, compared to Swiss Re’s human baseline of 1.11 claims per million miles. 

Technical Perspective

Here’s an interesting set of facts from Reddit.

The primary technical difference is that Waymo/Cruise use multiple sensor modalities: Lidar / radar / camera, while Tesla is camera only.

In a vision-only system it is much harder to (1) infer distance / depth of perception tracks, (2) have redundancy in the system, (3) validate using sim or otherwise that your system would do the right thing across a broad range of scenarios, simulating what cameras would see is more difficult.

Technicalities aside, Waymo/Cruise are L4 robotaxis, Tesla is an L2 driver assist for personally owned vehicles.

Second Reddit Comment

Waymo and Cruise use short and longe range LiDAR and short and long range Radar for spatial (not just cameras for heuristics) plus acoustics and then sensors around the vehicle to detect even the most minute collision. You can tap on the bumper (or anywhere around the surroundings of the car) of a Waymo or Cruise vehicle and it’ll check in with RA to make sure everything’s OK or safe-stop the vehicle on it’s own. Tesla just uses cameras and mapping.

The compute units in a Waymo or Cruise vehicle cost more than an entire Model 3, and they have redundancy.

Third Reddit Comment

Tesla only uses cameras. The idea is to train the car to drive based just on what it seems with cameras, similar to how humans drive based on what we see with our eyes.

Tesla is now trying to do “end to end” which means training a single neural network to drive directly from video. Waymo uses a series of neural networks that are interconnected and trained separately.

The Tesla approach is a lot harder. For one, only using cameras and no detailed maps, means the car has no redundancy if the camera vision makes a mistake. This can be a problem since cameras can be blinded by the sun, occluded by objects or get dirty. And vision can make a mistake. It might misidentify an object or get a false positive or false negative.

That is one benefit of Waymo also using radar and lidar. Radar and lidar are active sensors, so they tend not to fail in the same way cameras do. And radar works great in rain or fog where cameras are less reliable. So having radar and lidar provides that secondary source to make your perception more reliable. And with more reliable perception, your car has better info to make decisions.

Translation

Musk wants to do this cheaply as possible and lags in technology because of it. And those Waymos are not exactly pretty.

Does Musk value prettiness and cheapness more than safety?

Levels of Driving Automation

Levels of Automation from Synopsis.

Musk and the Tesla cult claim that a camera-only system is better but the results and the lead video show the claim is a lie.

Claiming superiority of L2 over L4 is a joke.

The huge inferiority of Cruise and Tesla, with numerous accidents coupled with ridiculous claims likely held back Waymo which should be on the verge of prime time nearly everywhere.

Crash Rates of Waymo and Those of Human Drivers

Please consider Comparing Crash Rates of Waymo and those of Human Drivers

 Waymo has presented further comparisons comparing the 7.14 million driverless miles (11.424 million kilometers) driven so far with human drivers, and even in these early stages of Robotaxis, it is amazing how safe Robotaxis are. By the end of October 2023, the numbers in the three locations of Phoenix, San Francisco, and Los Angeles – where Waymo Robotaxis are driverless 24 hours a day, seven days a week – were analyzed and compared to the corresponding crash numbers caused by human drivers during the same period in the same region, resulting in police reports, injuries, and/or property damage. The results speak for themselves:

• An 85% or 6.8 times lower injury accident rate, from minor to serious and fatal cases
  (0.41 accidents per million miles – 0.256 accidents per million kilometers – for the Waymo driver vs. 2.78 per million miles – 1.74 per million kilometers – for human drivers).
• A 57% reduction, or 2.3 times lower police-reported accident rate
  (2.1 incidents per million miles – 1.3 per million kilometers – for the Waymo driver vs. 4.85 incidents per million miles – 3 per million kilometers – for human drivers)

In other words, in the 7.1 million miles (11.424 million kilometers) that Waymo drove, there were an estimated 17 fewer injuries and 20 fewer police-reported accidents than if human drivers had driven the same distance at the same accident rate in the areas where Waymo operates.

One point to consider in this comparison is the fact that many accidents involving human drivers are not reported. While robotaxis in the US have to report every collision, no matter how minor, to the authorities, it is estimated that only a third of accidents are reported to the police. Two thirds are reported to insurance companies, while one third never appear in the statistics. If the figures are corrected for these factors, autonomous cars would be far more than 90 percent safer on the road compared to humans.

Tesla is Dead Last

This Business Insider article is about a year old, but it tells a story of a desperate Tesla trying to catch up.

Please consider Tesla’s Not Even in the Top 10 Self-Driving Firms.

Of the 16 companies recently ranked by research and consulting firm Guidehouse Insights (which ranks some of the biggest names working on automated-driving technology each year), Tesla came in last. Tesla ranked last in similar lists in 2021 and 2020.

This year, Guidehouse specifically focused on ranking companies developing this tech for light- to medium-duty vehicles, rather than the automakers that might eventually deploy it in their cars.

Tesla will remain dead last as long as shuns Radar and lidar and relies only on cameras.

“Full Self Driving” Beta

Musk claims to have a “Full Self Driving Beta”.

It may be a “beta” but it sure isn’t anywhere close to self-driving.

Reuters Reports Tesla to Scrap $25,000 Entry EV

On April 5, I noted Reuters Reports Tesla to Scrap $25,000 Entry EV, Musk Cries Liar

RoboTaxi Zero Chance

Tesla has the worst autonomous driving capabilities around. I will do a separate report on where the technology stands. [This is it now]

Meanwhile, I await Musk’s August 8 robotaxi announcement. Expect to be underwhelmed and overpromised. A Tesla robotaxi is not close to being ready. Waymo is here and now.

I consider Musk a genius. But at best he overpromises and underdelivers for years on end, especially with autonomous driving, but also semis, the M2, and the cybertruck .

August 8 Vaporware Announcement

News Online says Tesla CEO Elon Musk’s Robotaxi Promise Raises Doubts.

After quickly denying a report Tesla reportedly dropping plans for a $25,000 electric vehicleCEO Elon Musk made his latest promise to investors and fans: a self-driving Tesla would be unveiled on August 8.

It’s a tried-and-true playbook for Musk, who has regularly promised new technology and vehicles, only for products to actually hit the market years later.

“We see an Aug. 8 Robotaxi unveiling as more ambitious, similar to Tesla’s Semi and Roadster announcements — and likely not Tesla’s next model,” Bernstein analysts wrote Monday.

The Cybertruck The electric pickup was first unveiled in November 2019 and production was scheduled for 2021, although the first model wouldn’t be made until 2023. Roadster The sports car was unveiled in 2017 and is expected to go on sale in 2021. Now Musk wants to present a prototype of the model by the end of 2024, with production planned for 2025.

The CEO has been promising Tesla EVs will be able to function completely independently of the driver for about a decade via the company’s driver assistance software, Full Self-Driving (FSD). Tesla recently updated the software out of ‘beta’, although it still reminds drivers not to ‘get complacent’ and that their cars are not fully autonomous. A national nonprofit auto safety organization said last month that a previous version of the software failed the tests.

“The top two U.S. robotaxi companies that operate without drivers do so in limited geo-fenced areas, with relatively small fleets of particularly expensive vehicles carrying a significant amount of hardware,” Deutsche Bank analysts wrote on Monday. “The fact that Mr. Musk announced a robotaxi unveiling for August 8 does not in any way mean that the technology is ready.”

Demonstration Coming Up

Please note that Musk Plans Coast-to-Coast Self-Driving Demonstration by End of Next Year.

Also note the above story is from 2016.

Tesla plans to do a Los Angeles-to-New York drive “without the need for a single touch” by the end of 2017, Musk told reporters Wednesday on a conference call.

Tesla Recalls Nearly All vehicles

Please note that on December 13, 2023, Tesla Recalls Nearly All vehicles on US Roads Over Lack of Autopilot Safeguards

Tesla is recalling over 2 million vehicles in the U.S. to install new safeguards in its Autopilot advanced driver-assistance system, after a federal safety regulator cited safety concerns.

The largest-ever Tesla recall appears to cover nearly all vehicles on U.S. roads to better ensure drivers pay attention when using the system. Tesla’s recall filing said that Autopilot’s software system controls “may not be sufficient to prevent driver misuse” and could increase the risk of a crash.

The National Highway Traffic Safety Administration (NHTSA) has spent over two years investigating whether vehicles produced by the electric automaker led by billionaire Elon Musk adequately ensure drivers pay attention.

Separately, since 2016, NHTSA has opened more than three dozen Tesla special crash investigations in cases where driver systems such as Autopilot were suspected of being used, with 23 crash deaths reported to date.

The “fix” will limit the amount of time a driver can have his hands off the wheel.

Yet, Musk brags of Full Self Driving capability.

FSD costs a $15,000 or a monthly subscription fee of either $99 per month or $199 per month depending on the vehicle’s current Autopilot package.

Why anyone would pay $15,000 for dangerous features that also require you to keep your hands on the wheel is a mystery.

Tesla Robotaxi Conclusion

The Tesla robotaxi is nothing more than vaporware and it will remain so as long as Musk remains committed to unsafe, stand-alone technology.

Tesla is now many years behind Waymo.

Cybertruck Failure

Musk’s diversion into the cybertruck is doomed to fail, assuming it hasn’t failed already.

Please consider Tesla Cybertruck Owners’ Forum Is Already Full Of Tales Of Broken, Malfunctioning Cybertrucks

From gunky surfaces t0 embarrassing off-road exploits to malfunctioning after going through water, the stainless steel Tesla Cybertruck seems like an all-around mess. We cover a lot of those concerns here on Jalopnik, but if you want an even clearer picture of just how dire owners of this machine have it, look no further than the Cybertruck owners forums.

We decided to check out the forums for ourselves after we spotted a tweet by @salsadrunkard that featured a screenshot of a Cybertruck owner detailing how their truck broke down a mile from the delivery center. It didn’t take much digging to find the forum and the thread itself on Cybertrucks Owners Club.

Click through some other posts, and you’ll quickly fall down a rabbit hole of malfunctions and issues. One owner, for example, described how their Cybertruck suddenly braked after passing another truck on the side of the highway.

Another owner gave their Cybertruck such a damning review that they were allegedly banned from one owners forum and had to turn to another, Tesla Motors Club, to warn prospective buyers about the “borderline dangerous” visibility issues:

It’s so, so bad. You can’t see the front corners adequately. It’s borderline dangerous! It’s a joke that there’s no rearview camera display where the rear view is, instead, it’s a tiny display on your main screen that shows you behind. So stupid. Cost cutting.

And yet another owner described how, after just two days of ownership, their Cybertruck simply stopped turning on, despite the 40-percent battery charge.

I’ve had my truck for two days, got in this morning, everything was on. Went to press the brake to put it in reverse and everything went black. Power door button wouldn’t even let me out, had to use manual release to get out. I can not get back in either. My battery is at 40%, so no it’s not dead. Has anyone had this problem?

Musk Promises 50,000 EV Semis a Year

That won’t happen because there are 4 Million Semis on the Road, Only 35 Class 8 Truck EV Charging Stations

For the 5th year, Musk is hyping 50,000 electric semis without having a factory to produce them.

Electrek says Tesla’s giga factory is only about 30% complete and Tesla hasn’t expanded the facility for years.

And now Tesla auto sales are falling.

Tesla’s Deliveries Drop for First Time Since 2020, It’s Demand Not Supply

Please note Tesla’s Deliveries Drop for First Time Since 2020, It’s Demand Not Supply

Tesla’s heydays of surging demand growth for Teslas is over. Competition is increasing and relative demand growth, if not absolute demand growth, is falling.

Tesla has a drought of new products and competition is catching up everywhere. It’s autonomous driving features are an outright joke. More importantly, they are a huge safety risk.

---

More By This Author:

Reuters Reports Tesla To Scrap $25,000 Entry EV, Musk Cries Liar
Millennials Rush To Buy $2,300 Gold Bars At Costco
Jobs Up 303,000 Full Time Employment Down 6,000 In March