Robocars 2020 In Review: Winter To Spring

2020 began with fear of the “Robocar Winter” — a slide into the “trough of disillusionment” for the self-driving car industry that began in 2019. In spite of the world’s most chaotic year in recent memory, we saw that winter already turn into Spring. Here are the top stories of the year.

Real revolutions can get even more hype than false ones, so it’s not surprising that when the early promises that everybody would deploy by 2020 didn’t come to fruition that there was disappointment and pullback. Indeed, big car OEMs were quite happy to slow down, to not see their own industry turned upside-down at the speed that Silicon Valley startups like to operate. Most of those tech companies didn’t get the memo, and the year concluded with both real products and more insane valuations. Let’s go over the most important stories of the year, from #13 all the way to #1 — which (spoiler alert) once again involves Waymo, as does #2.

This review is also available in video form for those who prefer it:

#13 Startups die, OEMs pull back

The year began with two well-funded startup facing failure, including Starsky Trucking (which just shut down) and Drive.AI (which sold its team to Apple
.) At the same time, many Auto OEMs continued their pullback from ambitious self-driving plans to work on more driver-assist “pilot” style offerings to compete with Tesla. The pullback of the auto OEMs was bad news for all the startups founded with the idea of selling to those car companies, who need a bit of innovation assistance.

Mercedes Drops Partnership With BMW, Hooks Up With NVIDIA On Self-Driving: Robocar Roundup

#12 Cruise reveals the Origin

GM/Cruise has mostly been doing tests and demonstrations in modified Chevy Bolts but they showed they also want to make a fully custom vehicle, and showed off the Cruise Origin, a fairly boxy Robezium meant to be used as a shared vehicle. Cruise says this is not simply a concept car — OEMs have shown off lots of concept robotaxis — but one in active development.

Cruise’s new robotaxi presages a battle with Zoox

#11 NTSB Slams Uber and Tesla, Uber driver gets charges

At the start of the year, the National Transportation Safety Board had hearings on both the Uber Fatality and a Tesla Autopilot fatality. They came down hard on both companies, but also on NHTSA for not regulating enough. While pointing out many flaws at Uber, in the end the primary cause of the fatality was the negligent safety driver, who was eventually charged with negligent homicide by Arizona authorites.

It is noteworthy that the NTSB seemed pretty pleased with how Uber had turned itself around and devoted itself to safety. They like companies who do what they say, and were quite bothered by an unnamed company whose CEO (whose name probably rhymes with dusk) hung up the phone on them.

Uber Robocar Safety Driver Charged With Negligent Homicide

#10 Cruise, Yandex, AutoX, Nuro reach no-driver milestones

We’ll see the biggest milestones later in this list, but several other companies crossed important lines in putting cars out on the road without a human being behind the wheel able to quickly take the controls. While any good team has reached a level where they can have the car run for long periods without needing an intervention, reducing the ability to intervene is a strong sign of confidence in the system — the strongest sign that the public can be shown. It means the engineers tested things extensively and made a presentation which impressed the lawyers and the board enough to allow the project to put itself on the line.

AutoX took the driver out entirely in Shenzen, though on simpler streets. Nuro’s vehicle never had the ability to carry a driver, so they have always needed to work this way, and they did more of it. Yandex has, for some time, done tests with a human in the passenger seat. This person in the passenger seat still has a kill switch and can physically grab the wheel, so this is a much less bold step, but it’s still a step.

Cruise made a big deal of this as well, but also kept a person in the passenger seat and only showed off driving on deserted streets on the outskirts of San Francisco at night — so a small step, but at least a step.

Zoox also has no wheel in their vehicle, and it turns out their drive in San Francisco was a closed-street photo-shoot with staff pedestrians, not a real demo. However, their new vehicle can’t really have a safety driver so they soon will need to move to the real deal.

GM/Cruise Runs Test Vehicles In San Francisco With Nobody In Driver’s Seat

Chinese Robotaxi Firm AutoX Starts Operation In Shenzen With No Human Driver

#9 Zoox sells to Amazon

This was a big year of news for Zoox, but not all of it was good. Early in the year, with the winter in full force and the stock market in Covid turmoil, they needed to raise more money, and could not do it. Instead, they sold themselves to Amazon for just over a billion dollars — which is only a touch more than they raised in previous rounds, so not exactly a success. On the other hand, they now have an ultra-rich parent company who says that they plan to continue the Zoox vision. Many wondered if Amazon bought it just to get the team or to devote their tech to Amazon’s massive logistics business. For now, they are keen to do the robotaxi business as well.

Amazon Buys Self-Driving Company Zoox For $1.2B And May Rule The World

#8 Honda promises Traffic Jam Pilot

This relatively minor news was one of the few bright spots for car OEMs. Honda’s Legend will feature a “traffic jam pilot” that lets you take your eyes off the road during the traffic jam. You need to take control again when the jam clears. While other car companies have promised this but held back, this is a car OEM putting out a very limited-domain real robocar, and taking the liability for it.

Honda ‘Traffic Jam Pilot’ Offers Standby Self-Driving By March 2021

#7 LIDAR and sensor developments and public valuations

A sure sign of the return of enthusiasm has been the explosion in the LIDAR space, including some recent SPAC based public offerings by companies like Velodyne and Luminar with valuations of $4.5B and $8B respectively for their sensors. More are expected to come soon from Ouster and perhaps others.

We saw new forms of LIDAR and entirely different depth sensors, including imaging radar (with rumours that Tesla may use ARBE’s imaging radar chip) and a new multiview camera depth system from “Light” a company that began making consumer cameras but changed direction with a Softbank investment.

Light’s ‘Clarity’ Depth Camera Could Be A Game Changer

#6 Covid

The top story in the mainstream news of 2020 had much less effect on robocars than might be expected. Most teams that were testing with two safety drivers shut down testing for a while, but many companies adapted easily to work-from-home. Covid may have accelerated Waymo’s push (seen below) to do more operations with no safety driver on board. Many companies did Covid gimmicks but everybody wondered what the virus meant for the long term future of shared-ride transporation. Uber and Lyft
saw business plummet and rebound to half, and transit ridership is still way down and may take years to recover. On the other hand, several cities took the chance to experiment with closing some streets to cars to promote outdoor dining and walking.

The delivery business exploded (saving companies like Uber) but brought new attention to robot delivery firms like Nuro and Starship, which switched from campus deliveries to neighborhood ones. While contactless delivery is quite doable with a human driver, people feel better when it’s just a robot.

Covid also caused a great boom in video meetings and virtual conferences, and may create a long term decline in how much travel and commuting people do even after it’s over.

#5 Zoox reveals its secret vehicle

Zoox has been guarding its custom vehicle design since not long after its founding. Their mission was to go beyond all the other teams who put prime focus on the self-driving stack to also design the vehicle and build a robotaxi service. They finally let the world see the vehicle, which takes a similar “Robezium” form with face-to-face seats and sliding doors, but some special Zoox touches, like sensors with no blind spots, 4-wheel steering, symmetrical design with big batteries under the seats and good crash test results.

Amazon-Owned Self-Driving Taxi Zoox Reveals Its Secret Vehicle

#4 Uber ATG Merges with Aurora

After all of Uber ATG’s troubles, Uber decided to get rid of it (and its Uber Elevate e-VTOL project.) It “sold” it to Aurora, in a deal which merged the two companies and invested an extra $800M in Aurora, pushing its valuation towards an incredible $10B. It was widely felt that Uber’s self-driving software stack was nowhere near the top of the pack in quality, and while Aurora has a better pedigree, it has made few demonstrations of the quality of its stack as well. The research part of Uber ATG was let go. Aurora (like Waymo and many startups) has moved a lot of focus to trucking.

Uber ATG And Aurora Merge To Staggering $10B Valuation

#3 Tesla releases beta of city-street Autopilot (”FSD”)

The most attention of the year may have gone to Tesla releasing a very limited beta of what they misleading call their “full self drive” product. This is a driver-assist product which can handle city streets, and that’s pretty impressive considering it has to be done with the limited camera/radar/ultrasound sensor suite of a Tesla, and also works from limited maps which are mostly navigation maps augmented with extra data on certain trouble spots.

Tesla definitely leads the pack when it comes to driver assist, but it still requires a significant breakthrough to turn this into an actual real true working full self-driving system. Youtube is loaded with videos of Tesla’s completing decent length single trips without mistakes — and plenty of videos of scary mistakes — but it’s still a very huge distance from being able to do several lifetimes of driving without mistakes. If Tesla succeeds with this huge bet, it will be a major accomplishment, but it is still a bet. On the other hand…

Tesla’s ‘Full Self-Driving’ Is 99.9% There, Just 1,000 Times Further To Go

#2 Waymo safety report shows incredible progress

Waymo published a deeply-detailed report on all safety incidents from over 6 million miles of driving in their pilot zone in suburban Phoenix. The result — over 8 human lifetimes without ever being at fault in an accident. That’s a superhuman level of performance, and the detail in their report throws down a gauntlet in front of all other teams who want to convince us they have done something impressive.

We should expect to see Waymo to now expand to other cities — suggestions are that San Francisco and Silicon Valley are next.

Waymo Data Shows Superhuman Safety Record. They Should Deploy Today

#1 Waymo robotaxi service open to the public

As a result of having pulled of the safety record at #2, the top story is the release of the world’s first real robotaxi service, open to the public within its service area. Like Uber, anybody can download the Waymo app and take a ride in the Chandler, AZ area. For a time, it was assured to be in a Waymo with no safety driver aboard, though they later starting bringing safety drivers back behind plastic barriers to expand the service area.

Yes, driving in Chandler is easier than most of the world, so Waymo isn’t finished. But they’ve hit the real milestone with this service, and while there is hard work to do, there’s no major breakthroughs needed on the way to bringing robocars to a lot of the world, and eventually to most of it.

Without Fanfare, Waymo Opens Robotaxi Ride Service To The Public

Coming in 2021

Next year, I think we’ll see Waymo expand, and others get more milestones. You’ll see Tesla expand their FSD offering to more customers, and possibly even sell it as a monthly subscription, though at a fairly high price like $200/month. We’ll see more happen in trucking, which is the first thing that will get really commercialized, and we’ll also see more news come out of China. The winter is over, and spring is roaring ahead.

Source link

Leave a Reply

Your email address will not be published. Required fields are marked *