The future of computer-driven cars and deliverbots
Two major virtual self-driving conferences scheduled themselves for June 9. So I went to both, of course. Interesting news tidbids came from Argo, Starship, Scale, Chinese robotaxi makers, Zoox and many others. I summarize it here:
Summary of interesting headlines from The Information and TechCrunch simultaneous conferences
Cruise has been granted a permit to begin unmanned robotaxi tests in California. But they aren't allowed to charge, which forbids a lot of useful research, and explains why Waymo hasn't bothered with that permit yet. I discuss why it's so useful to experiment with payment in this Forbes site article at:
For a long time, Tesla refused to implement driver gaze monitoring with Autopilot. Now, quietly, the new release does it, though with no other changes as yet.
But it's quite a shift in their actions.
See the details at this Forbes.com story at
Some buzz has arisen due to photos of a Tesla with a LIDAR on it, and hints that Tesla is dropping radar from their cars in a hurry. Is this a major change of heart, par for the course, or a response to the chip shortage.
See analysis of that at this new Forbes site article:
A recently released video shows Waymo having some real problems when it encounters a construction zone it doesn't understand (that's expected) but then the remote ops team gives the wrong instructions and a comedy of errors follows. I discuss it here:
A VW executive suggested they might sell you a real self-driving function (when they have it) for around $8.50 per hour, as opposed to paying for it in a lump sum when you buy the car as Tesla plans.
What are the economics like of buying self-driving by the hour?
Read about that in my Forbes.com article at VW will sell you self-driving for $8.50/hour some day. Too high or too low?
For the blog series of my client DeepMap, here is an article about issues with mapping the private spaces of the world -- parking lots, garages, driveways, pick-up/drop-off zones and more.
The controversial Tesla crash in Texas is probably not so controversial. The NTSB's preliminary report only releases a few tidbits of information, but they point to Autopilot not being involved.
Read NTSB early report on Tesla crash has just a few more details
The new services keep coming, and now Baidu/Apollo has opened up a robotaxi service in outer Beijing, at an industrial park. While they call it fully driverless, they still have an employee in the passenger seat who is told to do nothing. This is in contrast with AutoX's service near Shenzhen and a few others, as we move closer to a true robotaxi service.
Details, and contrast with AutoX are in my new Forbes site article at:
Tesla released some important new details on the Texas crash that everybody's talking about (but probably shouldn't be talking that much about.)
The new details are not enough, though. Information is now contradictory until we learn more. Was there somebody in the driver's seat or not? We've learned that cruise control did play a role, but are told it brought the car to a stop, which it clearly didn't.
Read more details in this piece at:
My client DeepMap asked me to write an article listing various benefits that can come from having good maps, over and above their obvious use in localization, perception and safety.
You can find this post at Here Be Dragons: Surprising benefits of maps
As before, since this is done for a client, I want to disclose that conflict of interest, but what I put under my own byline does represent my views.
There has been much coverage about a fatal Tesla crash in Texas because police say the car didn't have anybody in the driver's seat. Elon Musk says Autopilot was not engaged, though of course the dead men may have been trying to pull a stunt hoping they could engage it in an area it isn't supposed to work. It didn't work.
So here's my analysis of what we know and don't know and why there probably is no big new story here.
I've been working as a paid advisor to DeepMap, which builds technology to allow self-driving teams and ADAS-Pilot systems to create the maps which allow them to work and get greater safety.
As part of that project, they have invited me to write some blog posts for them. The first, explaining just what high definitions maps are, how they came to be, and why they are valuable is now available to read.
Some announcements of hard ship dates for robotic vehicle deployments -- 2022 in Israel for shuttles, 2023 in USA for delivery vans using MobilEye and 2023 in Dubai for GM/Cruise Origin shuttle.
A few more details in new Forbes site post at MobilEye and GM/Cruise announce production ship dates in 2022 and 2023
Waymo self-driving's CEO, John Krafcik, has stepped down. More interesting is the two new co-CEOs appointed from inside (one from the original founding team that I worked with) have no automotive industry background. Having worked at Waymo and followed it more closely than almost anybody, I have some thoughts on the shift in this Forbes site column:
Read Waymo CEO Krafcik Steps Down; New Co-CEOs Come From Tech, Not Auto
Over a decade ago I started advocating for having a large public library of simulation scenarios to test self-driving cars. Today, Deepen.AI, a company I am an investor/advisor to announces it has built such a library, in cooperation with the World Economic Forum and WMG University of Warwick and with involvement from many companies and government agencies. In the Safety Pool, people can build and contribute test scenarios, and in return get back manyfold from the contributions of other members.
Right now Waymo One only serves suburban Phoenix. How hard is it going to be for self-driving companies to expand to new cities, new countries, new conditions and new rules of the road? Some think very hard -- and it's not trivial. But it's likely they can afford it just fine. I explore why is this new article on Forbes.com:
Will It Be Hard Or Easy For Self-Driving Cars To Expand Their Territory?
A modest change at Waymo -- allowing you to do multi-stop trips where the vehicle waits around for you (physically or virtually) at no charge gives us a taste of the different economics of robotaxis compared to Uber, since no driver has to be paid. I discuss these changes in a new Forbes site article at:
Waymo Lets You Make Multiple Stops On A Robotaxi Trip, But Not Leave Your Stuff
A proposed California bill would require all robocars to be zero emission in under 4 years. As good as going electric is, the government should not be picking the power train so soon, especially when there is no sign that the existing players are bad actors. I detail more in my Forbes site article at
Foolish California Bill demands a robocars be electric by 2025
Everybody is working on making robocars drive more safely and cause fewer accidents. Waymo recently released a paper outlining how they ran a large number of accidents in simulation, and tested what happens if the Waymo system is driving the primary car (which caused the accident) or the secondary car (which didn't) in 2-car collisions. No surprise that they prevented the accidents when being the primary car. More interesting is they prevented almost all the accidents when being the second car.
Copyright © 2025, Brad Ideas
Designed by Zymphonies