The future of computer-driven cars and deliverbots
In June, Cruise had the first crash for an uncrewed robotaxi which caused injuries to 3rd parties, including a passenger and a person in the other vehicle. The Cruise vehicle was partly at fault.
In this article I outline the new details we have learned about the crash, but also discuss what it means for the future, and whether the use of a NHTSA "Recall" for this particular software update is the right idea.
Tesla announced the price for the FSD software add-on will rise to $15K (from $12K) Sept 5. The price is amazingly high for a prepaid pre-order of a product that doesn't exist yet. Yet people only pay $4K for it in the aftermarket, and the take rate keeps going down as they raise the price, negating revenue gains.
So what does it all mean? One unusual option is that at $15K/head, Tesla could fail at producing the FSD software, but buy another company that does succeed (using LIDAR probably) and retrofit the old cars at a profit. At this price it's hard for them to lose.
Some transit agencies want to be in charge of how self-driving cars are deployed in their cities. Otherwise, they say, robocars will compete with transit, as if that would be bad.
Read more about these issues at Will Transit Agencies Fight Or Yield To The Self-Driving Revolution?
There has been lots of buzz over a video made by Tesla Critic Dan O'Dowd of a Tesla allegedly in FSD mode hitting a dummy in the shape of a child. We've seen Tesla fans duplicate it with their own kids, and Tesla asking the original video be taken down, and NHTSA saying not to use your own kids and more.
But it all misses the point. Of course a prototype fails in ways like this. The question is, are people actually getting hurt, and how do we really test these things to get them working? Is it OK to have customers participate in testing?
Baidu has finally moved to having a paid robotaxi service with no employee in the car. While they have remote monitoring and even driving over 5G this is still a big step and a show of internal confidence in the vehicle.
Read more at Baidu starts paid robotaxi service in China
Baidu Apollo has released their own custom robotaxi plan. This one looks more like a regular minivan/custom taxi, but its steering wheel, there only for compliance purposes, is designed to be removed when the law allows, and that opens up the interior. They also say they can make it for about $37,000.
For more details see my Forbes.com story at Custom Robotaxi from Baidu
Several instances have taken place where multiple Cruise robotaxis have all frozen as a group, sometimes blocking intersections.
I discuss reasons for that, and why it's not that big a deal, in this new article on the Forbes site.
I have written a guide of useful hints and tricks for doing an EV road trip and barely spending any time charging. I've done over 10,000 miles of EV road trips and you can to, once you get an EV.
Read this at Forbes.com:
I have two other articles on Forbes.com that I didn't publish here in the blog:
San Francisco Fire complained that a Cruise robotaxi delayed a fire truck (but by less than 25 seconds) when it was stuck behind a stopped garbage truck, and the Cruise couldn't pull over in the oncoming lane enough.
Read the Forbes.com story at Cruise Robotaxi delays SFFD fire truck
A recent big announcement says the Cavnue consortium and Michigant will build a "Connected Autonomous Vehicle" corridor on I-94 outside Detroit. It's the classic "smart road" which special infrastructure and cars communicating with it.
But is it that smart, or is a dumb highway smarter in the end?
I outline the reasons in this Forbes site article at Michigan wants a smart highway on I-94. A dumb highway is better
For the robotaxi business to be worth it, they must get customers who give up car ownership because of the service, and use it regularly. But since robotaxis will have a limited service area, what will they do to make it happen?
I discuss various strategies, including partnering with competitors and linking services areas in a new Forbes site column at What must robotaxis do to make people give up car ownership?
Elon Musk has now teased that Tesla will build its own custom robotaxi, at low cost. This is at odds with their brilliant plan to turn off-lease Teslas into robotaxis, letting somebody else eat up 40% of the depreciation. Will they do both, or do they have a new plan up their sleeve for a small one-person pod?
Read about this in my new Forbes.com story at Tesla teases a Robotaxi, are they crazy to give up off-lease plan?
From the earliest days, one of the most common questions was "What happens when the cops want to pull over a robocar or give it a ticket?" We find out a real answer in a video of SFPD stopping an empty Cruise robotaxi on the streets of San Francisco.
It wasn't actually that much of a mystery, and the major teams all have detailed first responder training and plans in place, and it happened here. This was a very rare case where it actually made sense to pull over this car, which was driving at night without its lights on, which is unsafe.
It is common to see plans for EV charging which are still bound up in "gasoline thinking" where people treat an EV like a car with a tank you empty and then fill up while empty, waiting. In fact, if you do EV charging right, you always do it while you are doing something else, so it takes zero time from your day.
Mercedes has been promoting the new Drive Pilot system in high end models. Equipped with LIDAR, it will do the full driving task on freeways in traffic jams in daytime good weather.
In this new article I discuss whether the so-called "level" 3 (or any of the levels) make sense, and what this product means, good and bad.
Dan O'Dowd is the CEO of Green Hills Software. He recently placed a full page ad in the New York Times protesting the poor quality of Tesla FSD, and has started a project to get the world to secure all critical systems, including cars, using his techniques. He makes the bold claim that only he knows how to make software truly secure and bug free, and warns the world it had better listen. He knows that's an extreme claim, but also says he has proof if the secure systems he has designed for aircraft, fighter jets and the FBI. And he's got the money to make a stir.
It may seem minor that Waymo is going to start charging for robotaxi rides. But this starts the process of learning real facts about robotaxi economics and what a ride will cost, and how that changes the world.
Read the Forbes.com story at Waymo can charge for rides, so it gets interesting