Autonomous vehicles are learning how to predict pedestrians’ movements. How soon before drivers are no longer needed?

Autonomous vehicles can already identify people in and near roadways. But human behavior is complicated.

Is that pedestrian about to step off the curb and into traffic? Does that person in the crosswalk up ahead know a vehicle is approaching? Is the human distracted by another human or, worse yet, one of those little devices humans spend so much time looking at?

New software recently demonstrated in Ontario aims to answer those questions and prevent self-driving cars and trucks from the kind of robotic panic attacks that cause excessive braking, interrupting the flow of traffic, as the computers manning the vehicles attempt to predict what a nearby human will do.

“I’m looking forward to not having to depend on drivers,” Ron Massman, CEO of Dependable Supply Chain Services, grinned to a crowd of journalists and automotive industry professionals. “Just kidding. That’s not going to happen for a while, anyway.”

But the June 11 demonstration at the company’s subsidiary DHE suggested a future where autonomous delivery trucks roam the nation’s highways may not be that far off.

Boston-based Perceptive Automata showed off its software, using two months of data from Los Angeles and Inland Empire roads. Although the software was enabled on one of DHE’s big rigs, which analyzed every human that crossed in front of its camera, the road tests were conducted by taking human-driven Volvo SUVs to Ontario International Airport, with its arriving and departing — and often distracted — travelers.

If autonomous vehicles can reliably identify whether humans are aware of vehicles and judge how likely they are to enter the roadway, it can allow cars and trucks to not stop every time they spot a human. That means less herky-jerky driving, a more comfortable ride for passengers, less wear and tear on vehicles, and less chance of being rear-ended by surprised human drivers behind them. And, of course, it could save pedestrian lives.

“The physics are a solved problem, almost,” said James Gowers, vice president of Strategy and Business Development for Perceptive Automata. “This is the final piece.”

At Ontario International, the software that identifies humans — provided by a third party, Perceptive Automata employees noted — had a few false positives, mistakenly identifying fire plugs and palm trees as humans. But Perceptive Automata’s software correctly evaluated in real time whether pedestrians at the Ontario airport were aware the car was coming and whether they were about to step into the roadway. (The company’s software also correctly identified the fire plugs and palm trees as not intending to step into the roadway, although it judged that only the fire plugs were aware of the car’s presence.)

“Driverless vehicles are going to be safer than human-driven vehicles. We have more than 33,000 traffic fatalities every year, and 90 percent of them are due to human error,” said Jeff Miller, a computer science professor at the University of Southern California’s Viterbi School of Engineering. “Once you take humans out of the equation, you’re going to have far fewer collisions.”

In March 2018, a self-driving Uber car struck and killed a pedestrian in Tempe, Ariz.

“But what’s not being talked about is the hundreds of millions of miles that have been driven with no collisions,” Miller said.

DHE already is investing in autonomous vehicles, putting down a deposit on 10 of Tesla’s self-driving electric semi trucks the first day they were available, according to Finney. But it won’t come cheap: It’ll cost up to $600,000 to electrify one of DHE’s locations, and the company would need to do so at a dozen or more sites to support an electric fleet.

Volvo, which has partnered with Uber, is betting big on autonomous vehicles. And no wonder: Sixty-five percent of its revenue is from trucking and another 14% is from buses. Moving away from human drivers could reduce liability and increase the time vehicles can be on the road.

“If something goes wrong, it’s easier to replace an autonomous truck than a human life,” said Aravind Kailas, manager of  Research and Innovation for the Volvo Group.

Last year, the company demonstrated a yacht capable of autonomously backing itself into a dock. A half-dozen autonomous Volvo trucks are transporting limestone in a Norwegian mine. And a driverless electric city bus is being evaluated for use in Sinagpore.

But these are mostly test runs — for now.

“Automation will come, but it’ll be gradually introduced into society,” starting with drivers being assisted by the software, not replaced by it, said Kailas.

Miller agreed, in part.

“There are going to be a couple of phases before we completely remove truck drivers,” he said. “I think we’re still going to have our daily, short-haul drivers.”

That’s good news for the armies of short-haul truckers who serve at ports in Southern California and nationally.

But changes in long-haul trucking, where there’s fewer distractions for an autonomous vehicle to grapple with, could be more immediate.

“Long-haul, you could have an operator program in a truck’s destination, and when it gets there, someone receives it and transfers the cargo to its local destination,” Miller said. “I think we could have that in the next five to 10 years.”

So, not all of the nation’s truck drivers will be out of work a decade from now — but some will be. And not everyone is excited about this vision of the future.

“Is the motoring public going to be the guinea pig for testing this?” said Kara Deniz, a spokeswoman for the International Brotherhood of Teamsters. “We’re not against innovation — there’s lots of innovation that we’ve supported and embraced — but this has to be done the right way. There’s a mad rush to be the first. And it needs to be done transparently and safely.”

The Teamsters represent 1.4 million members in the U.S., Canada and Puerto Rico, about 600,000 of whom are behind the wheel of a school bus, garbage truck, delivery truck or other vehicle.

Although commercial driving requires training to get a commercial driver license, it doesn’t need a college degree. And that makes it an easier and more affordable path to the middle class. Forty-five percent of adults in the Inland Empire have a high school diploma or less education, according to John Husing, an economist with the Inland Empire Economic Partnership. And the region has 600 million square feet of industrial space, most of it in logistics, meaning there’s a lot of truckers living and working in the region.

“Talking about this as ‘well, this is going to happen, you just have to accept it,’ it hogs the conversation around it. And it serves to remove the workers from the equation,” Deniz said. “With truck driving in particular, you’ve got a good wage-earning profession and you can’t write off these people.”

There’s no immediate danger of that, according to Joe Finney, COO of Dependable Supply Chain Services. There are things that autonomous vehicles can’t yet do for themselves.

“When you’re going over the Sierras, you have to stop and chain up,” Finney said. “I don’t think that’s going to happen autonomously.”

Similarly, an autonomous vehicle can’t change its own flat tires.

Even if they could, according to Finney, the public’s not likely to embrace a “Black Mirror”-like future where autonomous trucks roam without a human driver on board.

“Last time I got on a plane, there’s still a couple of pilots up there,” he said. “Even though the planes can basically fly themselves, I don’t think people would be comfortable without them.”