https://www.arcamax.com/business/businessnews/s-3049075?fs

sample quote

At Aurora’s terminal just south of Dallas, a worker cleans sensors on top and at the side of a dark blue Peterbilt truck while a safety driver sits in the cab ready for the truck to pull out. If all goes as planned, the safety driver, whose hands now hover above the wheel without touching it while the truck is in transit, will soon no longer be needed for the 200-mile trek to Houston.
...snip...
Gatik AI, a Mountain View, California-based startup, has already driven trucks without a driver in Arkansas and Canada. The company uses smaller, box trucks and plans to deliver from distribution centers to stores. In 2024, the company expects to deploy driverless trucks in the Dallas area “at scale,” said Gautam Narang, Gatik’s co-founder and CEO, in an interview.
...snip...
Besides saving on trucker pay, the trucks can travel longer than the 11-hour limit now on human drivers.
The sensors scan in all directions several times a second to identify objects, speeding up reaction time.
There are even estimated savings on emissions of 10% or more because
the vehicles will stay just below the speed limit and travel at a steady cadence, the companies say.
...snip...
While driverless trucks haven’t had any at-fault incidents with other vehicles in testing with safety drivers, the FMCSA report suggests that they may not be immune to accidents. Nearly two-thirds of fatal accidents occur when a person, object, animal or other vehicle veers into a truck’s lane.

What could go wrong?

I'm sure they've thought of everything...

Regularly driving the highways of Texas all I can say is the current crop of truck drivers are, for the most part, a bunch of donkeys. They sure aren't professionals. Be glad for whatever takes them off the road.

I keep wondering and waiting for the

I saw a news report on one of these companies. The CEO showed how it works. There's no blind spots. They have done plenty of testing, collected plenty of data but the Federal DOT hasn't asked for it.

Airplanes have had auto-pilot systems for years. Still pilots are required to be at the controls. Airplanes do crash as a result of problems with the auto-pilot system or related sensors. Disasters are frequently avoided as a result of the pilot over-riding the auto-pilot. Given the choice of flying on an airplane controlled by an experienced pilot or an airplane being controlled by a software engineer, I'll choose the pilot every time.

I'd be curious how these autopilot systems react to a mechanical failure. Say a blown steer tire? Assuming the system maintains control of the truck, will it pull to the berm or just stop in a lane of traffic?

What if it loses compression braking going downhill? Like a broken driveshaft. Will the system fry the brakes trying to stop? Will it use a runaway ramp?

What if the truck breaks down in traffic? Will it allow another truck to pull it out of the way?

I was being very sarcastic with my comments.

I know. And I wasn't. These are real life possibilities and knowing the answers can be helpful.

We see this in my line of work, engineers design a system and forget about common variables. Example from a few years ago was an update to a forklift transmission, the factory wanted to use a valve body from another model to save costs. Two very different transmission designs and the valve worked fine in it's original application.

The engineers never considered internal leakage in the receiving transmission and pressures were all over the place. Mechanics had to play with springs and orifice sizes to get it functioning.

A fellow at a fab shop that built fairly large structures mentioned that a new design ended up almost an inch out of spec. The designers had all the parts laser cut to some exact dimensions. But they completely forgot to leave space for welds. Lots of grinding followed on the rest of the first batch while the plans were redrawn for the laser cutting.

My first priority worry would be:
would a malicious passenger car (or motorcycle) driver figure out
how to use his (maybe stolen) vehicle’s movements
to wreck the robot truck without touching it.

You wouldn't have to wreck it.
Just box it in to stop the truck, loot it, and drive away.

YouTube is full of videos of truckers falling asleep, loosing control, wrecking, etc.

Nothing is foolproof but i think that over time driverless technology can become an asset but not before a lot of problems rear their ugly head.

Think about how many people died from poor designs over the first 100 years of cars.
There are thousands of little things designed into cars because people were injured or killed.
Safety glass, brakes on all 4 wheels, crumple zones on hoods, double latch doors and hoods, etc.

Driverless cars are going to kill a few people to find mistakes in the design. Just like people driven cars have done.

Like the airplane example above I prefer a trained person behind the wheel.
Definitely for now but the day of the driverless will be coming soon.

I think Lexus already released a steer by wire car or at least are sampling the waters.
Computers already control brakes, throttle and gears. Complete control of the steering is next.

Driverless vehicles out on public roads? HaHaHaHaHaHaHa.

Until there are slots cut into the roads or tracks, forget it. The "driverless" vehicle concept out on public roads with other vehicles with drivers & passengers is absolutely ABSURD.

I will keep saying this until I am blue in the face, but driverless vehicles WILL NOT WORK UNLESS EVERY SINGLE VEHICLE ON THE ROAD IS DRIVERLESS! Put one human being in the mix and all bets are off...

wait til some bean counter starts getting cheaper parts or cuts back Maintenace then you will see why driverless is not safer.

How will driver-less trucks sense and handle “black ice” ?

Same way humans do

Shades of Demolition Man