"Tesla now has a teleoperations team responsible for “providing remote access to our robotaxis and humanoid robots.” Tesla has apparently discovered what Waymo learned more than six years ago: even the best self-driving software occasionally needs remote assistance."
It's never made sense to me why self-driving companies WOULDN'T rely on remote operators to handle edge cases (<1% of the time). It's important to avoid mishaps like cars driving in circles or holding up traffic, which give people a bad impression of self-driving cars.
As you mentioned in an earlier post, Tesla has an even bigger problem: FSD needs help from the driver very often....as I recall once every 1000 miles. That's probably both because they're behind on software and they still haven't figured out they need lidar. Maybe those things will change, but for now Tesla has a long way to go.
Great write-up. It's not immediately relevant to either firm, but Tesla's insistence on going camera-only, rather than using lidar or even radar, will make it very hard for Tesla to reach unsupervised FSD in winter environments; lidar helps, but radar will be critical for undertaking the driving task under snowy conditions. (More about this next week!)
Waymo hasn’t shown that they have mastered winter driving conditions either though. this will be important since we actually see a spike in ridehail demand when it’s inclement weather, since people don’t want to walk or take transit, etc., but still need to get places.
No, Waymo hasn't mastered it, but their willingness to use more than merely cameras gives them a significant edge; IIRC they were piloting in Buffalo a year ago, which is a useful data point. My forthcoming essay, to be published Tuesday morning, goes into the matter in detail!
Lucky! I don't have radar in my nose, but as a human, I get by in my snow walks just fine.
Of course, I'm not a car weighing 4,000 pounds that will occasionally travel at high speeds in snowy conditions that my training data is relatively unfamiliar with and where my camera vision renders false positives and negatives. If I was, I might want that radar to help me operate safely.
Also your navigational system is the result of hundreds of millions of years of evolution. Waymo and Tesla are hoping to get the job done faster by seven orders of magnitude, and a more “brute force” approach (like extra sensors) might help them do that.
Yet, when it comes to self-driving cars, AI, and humanoid robots, so far it has been marketing in the best case, and outright misleading assertions in the worst, that have also lead to fatalities on the road.
I hope sooner or later reality will catch up with his business approach, and that looks more likely as he's becoming more erratic and disconnected from fundamentals.
That's a stretch. obviously, FSD isn’t true full self driving but the product itself is pretty damn amazing. Especially when you compare it to other OEM’s and their general tech stack.
just not his personality. I think you have to accept his shortcomings, but also understand it’s those shortcomings that allowed him to build Tesla, SpaceX, etc. No one who sets realistic expectations would have ever pushed hard enough to get all of his companies this far.
It is outright lying, to hype up the stock price. People paid for FSD in the promise that rapid improvements are just around the corner and they would not even need to upgrade the car hardware.
good write up and excited to see you at ride ai in LA! If we ignore the elephant in the room of if/when Tesla will have true full self driving...
The remote operations make sense but it’s also a relatively small cost and pretty straightforward. I don’t see this as a big deal or negative for either Tesla or Waymo.
where I disagree though is that Tesla actually won’t need a lot of of the infrastructure (charging depots, repair/maintenance, etc.) since they don’t own a single car much like Uber. So they can offload all of that responsibility to individual owners, who want to put their Tesla vehicle onto the Tesla Robo taxi network in the future.
and there are also a number of existing large fleet owners with anywhere from hundreds to tens of thousands of cars (many are pe or vc backed) that are currently renting out vehicles (lots of tesla vehicles too) to human Uber drivers and they’ve already invested and built a lot of this infrastructure so when Elon hits the FSD switch, they could easily plug all of their Tesla vehicles into the Tesla, Robo taxi network for example.
Tesla owners can do some of the maintenance work, but I don't think they can do all of it. For example if a car gets a flat tire on the other side of town is the owner going to drive over and fix it? That seems unlikely. Offloading some of the work onto owners will probably allow them to scale up faster than Waymo, but there's still going to be significant work required for each market they enter.
"Tesla now has a teleoperations team responsible for “providing remote access to our robotaxis and humanoid robots.” Tesla has apparently discovered what Waymo learned more than six years ago: even the best self-driving software occasionally needs remote assistance."
It's never made sense to me why self-driving companies WOULDN'T rely on remote operators to handle edge cases (<1% of the time). It's important to avoid mishaps like cars driving in circles or holding up traffic, which give people a bad impression of self-driving cars.
As you mentioned in an earlier post, Tesla has an even bigger problem: FSD needs help from the driver very often....as I recall once every 1000 miles. That's probably both because they're behind on software and they still haven't figured out they need lidar. Maybe those things will change, but for now Tesla has a long way to go.
No way Tesla can launch driverless taxi in 2025, not even in 2026 or 2027.
Great write-up. It's not immediately relevant to either firm, but Tesla's insistence on going camera-only, rather than using lidar or even radar, will make it very hard for Tesla to reach unsupervised FSD in winter environments; lidar helps, but radar will be critical for undertaking the driving task under snowy conditions. (More about this next week!)
Waymo hasn’t shown that they have mastered winter driving conditions either though. this will be important since we actually see a spike in ridehail demand when it’s inclement weather, since people don’t want to walk or take transit, etc., but still need to get places.
No, Waymo hasn't mastered it, but their willingness to use more than merely cameras gives them a significant edge; IIRC they were piloting in Buffalo a year ago, which is a useful data point. My forthcoming essay, to be published Tuesday morning, goes into the matter in detail!
When it snows I always turn on the radar in my nose to help me get through it safely. My nosedar is “critical”, otherwise I’d die :(
My nose knows 😉
Lucky! I don't have radar in my nose, but as a human, I get by in my snow walks just fine.
Of course, I'm not a car weighing 4,000 pounds that will occasionally travel at high speeds in snowy conditions that my training data is relatively unfamiliar with and where my camera vision renders false positives and negatives. If I was, I might want that radar to help me operate safely.
Also your navigational system is the result of hundreds of millions of years of evolution. Waymo and Tesla are hoping to get the job done faster by seven orders of magnitude, and a more “brute force” approach (like extra sensors) might help them do that.
Tesla fans are finding the reality is quite different than what they claim when finally the "rubber hits the road".
Musk deserves lots of credit for rocket work.
Yet, when it comes to self-driving cars, AI, and humanoid robots, so far it has been marketing in the best case, and outright misleading assertions in the worst, that have also lead to fatalities on the road.
I hope sooner or later reality will catch up with his business approach, and that looks more likely as he's becoming more erratic and disconnected from fundamentals.
That's a stretch. obviously, FSD isn’t true full self driving but the product itself is pretty damn amazing. Especially when you compare it to other OEM’s and their general tech stack.
If only Musk was honest about limitations rather than promising full self driving was always next year.
just not his personality. I think you have to accept his shortcomings, but also understand it’s those shortcomings that allowed him to build Tesla, SpaceX, etc. No one who sets realistic expectations would have ever pushed hard enough to get all of his companies this far.
It is outright lying, to hype up the stock price. People paid for FSD in the promise that rapid improvements are just around the corner and they would not even need to upgrade the car hardware.
Agree. This constant drum beat that lies and unlawful behavior are to be tolerated because of some perceived genius is so disappointing.
good write up and excited to see you at ride ai in LA! If we ignore the elephant in the room of if/when Tesla will have true full self driving...
The remote operations make sense but it’s also a relatively small cost and pretty straightforward. I don’t see this as a big deal or negative for either Tesla or Waymo.
where I disagree though is that Tesla actually won’t need a lot of of the infrastructure (charging depots, repair/maintenance, etc.) since they don’t own a single car much like Uber. So they can offload all of that responsibility to individual owners, who want to put their Tesla vehicle onto the Tesla Robo taxi network in the future.
and there are also a number of existing large fleet owners with anywhere from hundreds to tens of thousands of cars (many are pe or vc backed) that are currently renting out vehicles (lots of tesla vehicles too) to human Uber drivers and they’ve already invested and built a lot of this infrastructure so when Elon hits the FSD switch, they could easily plug all of their Tesla vehicles into the Tesla, Robo taxi network for example.
Tesla owners can do some of the maintenance work, but I don't think they can do all of it. For example if a car gets a flat tire on the other side of town is the owner going to drive over and fix it? That seems unlikely. Offloading some of the work onto owners will probably allow them to scale up faster than Waymo, but there's still going to be significant work required for each market they enter.
See you in LA!
When it snows I always turn on the radar in my nose to help me get through it safely. My nosedar is “critical”, otherwise I’d die :(