Young Canadian robotics geeks’ Embark on a self-driving journey with a truckload of confidence

Open this photo in gallery:

Beverly Morrison loves to drive – the larger the vehicle, the better – and for nearly 30 years, she’s made it her career. Ms. Morrison started piloting a Greyhound bus when she was 24, and spent time in the U.S. military, where she drove trucks. But it was big rigs she really fell in love with. She hauled gasoline from refineries to fuel stations around California, and later delivered to outposts in Nevada, Oregon, Idaho and Washington State.

Ms. Morrison became accustomed to long, solitary journeys. “I love being alone,” she says, adding that’s why she stopped driving buses and switched to trucks. “I just wanted to be alone in the vehicle.” To pass the time, she listened to National Public Radio and devoured audio books in 10-hour stretches. She found something exhilarating in manoeuvring such massive objects through physical space. “It’s in my blood,” she says. “That’s the best way I can describe it.”

She’s not doing much driving any more, at least not in the traditional sense. Ms. Morrison works as a safety driver for Embark Trucks Inc., a San Francisco startup developing autonomous long-haul rigs. She sits in the bright blue cab, foot poised over the pedal, hands at the wheel, while the truck handles steering, braking, acceleration and staying in its lane. Ms. Morrison drives on public highways and on a test track outside the Bay Area, taking over the controls when necessary, such as when other drivers are behaving erratically. (Embark does not disclose how often its safety drivers intervene.)

An outsider might think there’s something strange or even demoralizing for a lifelong trucker to just sit there. But there isn’t, Ms. Morrison says. She emphasizes she always has control. “That’s why it’s not quite as scary as it sounds,” she says. By taking the truck out into the world, she is effectively teaching its electronic brain to become smarter, so that someday soon it can handle itself with no one at the wheel.

Here’s where the job seems to turn from unusual to comically dystopian. Isn’t she training the very technology that could put her and countless others out of work? Again, that’s not how she sees it. Ms. Morrison took the job with Embark because she says she believes it has the potential to make trucking safer and more efficient. “To be able to keep driving a truck, and work on this technology, is the best thing in the world to me,” she says.

Valuation of autonomous-vehicle

companies

In billions of U.S. dollars

$30.8

Waymo

(May 2020)

$10

Aurora

(Feb. 2021)

$8.5

TuSimple

(May 2021)

$5.2

Embark

$3.3

Plus

(May 2021)

THE GLOBE AND MAIL, SOURCE: embark

Valuation of autonomous-vehicle companies

In billions of U.S. dollars

$30.8

Waymo

(May 2020)

$10

Aurora

(Feb. 2021)

$8.5

TuSimple

(May 2021)

$5.2

Embark

$3.3

Plus

(May 2021)

THE GLOBE AND MAIL, SOURCE: embark

Valuation of autonomous-vehicle companies

In billions of U.S. dollars

$30.8

$10

$8.5

$5.2

$3.3

Waymo

(May 2020)

Aurora

(Feb. 2021)

TuSimple

(May 2021)

Embark

Plus

(May 2021)

THE GLOBE AND MAIL, SOURCE: embark

The developers behind that technology are two Canadians, Alex Rodrigues and Brandon Moak, Embark’s chief executive and chief technology officer, respectively. Both dropped out of the University of Waterloo, moved to California and founded the company in 2016 to reimagine the US$700-billion trucking industry. Their test vehicles have since accumulated more than one million miles of autonomous road time (with safety drivers), and the company, which will list on the Nasdaq this year, is piloting deliveries for a handful of partners. By 2024, it plans to run autonomous trucks commercially in the southern U.S. and start expanding to the rest of the country two years later. (That requires dealing with snow, a capability Embark’s trucks do not have.)

The two of them still have thorny technical hurdles to overcome. “They’re tough enough that nobody has really tried to tackle them yet,” says Mr. Rodrigues, “but it’s something Embark feels pretty confident about.” Technology isn’t the only roadblock, either: Regulators, insurance companies and carriers have to be onboard, too.

The autonomous driving field is replete with bold claims and unrealized promises, as developing vehicles that can safely navigate streets without human intervention has proved to be far more difficult than initially believed. Uber threw gobs of money at self-driving taxis, but sold the division last year in defeat. Long-haul trucking, though, involves long stretches of monotonous highway and is generally thought to be more manageable. There’s a business case, too, as the technology could drive down costs and alleviate a chronic shortage of drivers.

Investors are now pouring billions of dollars into automated trucking startups, and Embark is one of a handful of players striving to bring the technology to market in the next two or three years. The message, both from investors and trucking startups, is that after years of false starts, autonomous driving is finally near at hand. Embark’s valuation is a testament to that hope: US$5.2-billion, despite no revenue, substantial losses and an unproven business model.

For anyone to make money with self-driving trucks in the next few years would be a huge achievement, but all the more so for Mr. Rodrigues and Mr. Moak: the former is only 25, and the latter only recently turned 26. What they’ve built so far has amazed Ms. Morrison, the safety driver. She gives demos to employees and job candidates, and hears a common refrain: “It’s so human.”

Autonomous future

Embark’s vision is for autonomous vehicles and human

drivers to work together. Self-driving trucks will handle

the highways, while people tackle the more complicated

urban portions of a delivery.

Embark transfer departure hub

Local drivers

deliver freight

to be loaded

Embark’s automated

software engages for

the long-haul delivery

Its Vision Map Fusion program

relies on maps and the truck’s

sensors to analyze and

respond to real-world

conditions

Hazard

avoidance

Lane

detection

GPS

antenna

Lidar*

Cameras

Sensor

array

Rear

radar

Front radar

Embark transfer departure hub

Upon arrival at

Embark hub,

the freight is

unloaded.

Final destination

Local drivers then

deliver it to final

destination.

*Laser imaging, detection, and ranging

JOHN SOPINSKI/THE GLOBE AND MAIl, SOURCE: embark

Autonomous future

Embark’s vision is for autonomous vehicles and human

drivers to work together. Self-driving trucks will handle

the highways, while people tackle the more complicated

urban portions of a delivery.

Embark transfer departure hub

Local drivers

deliver freight

to be loaded

Embark’s automated

software engages for

the long-haul delivery

Its Vision Map Fusion program

relies on maps and the truck’s

sensors to analyze and

respond to real-world

conditions

Hazard

avoidance

Lane

detection

GPS

antenna

Lidar*

Cameras

Sensor

array

Rear

radar

Front radar

Embark transfer departure hub

Upon arrival at

Embark hub,

the freight is

unloaded.

Final destination

Local drivers then

deliver it to final

destination.

*Laser imaging, detection, and ranging

JOHN SOPINSKI/THE GLOBE AND MAIl, SOURCE: embark

Autonomous future

Embark’s vision is for autonomous vehicles and human drivers to work together.

Self-driving trucks will handle the highways, while people tackle the more complicated

urban portions of a delivery.

Its Vision Map Fusion program

relies on maps and the truck’s

sensors to analyze and

respond to real-world

conditions

Embark transfer departure hub

Embark’s automated

software engages for

the long-haul delivery

Local drivers

deliver freight

to be loaded

Hazard

avoidance

Lane

detection

Final

destination

Embark transfer

arrival hub

Sensor

array

Local drivers then

deliver it to final

destination.

GPS

antenna

Lidar*

Cameras

Upon arrival at

Embark hub,

the freight is

unloaded.

Rear

radar

*Laser imaging, detection, and ranging

Front radar

JOHN SOPINSKI/THE GLOBE AND MAIl

SOURCE: embark

Embark maintains a fleet of about 20 trucks, each customized with multiple lidar devices, cameras and GPS antennas. Today, Embark pilots its trucks along Interstate 10 between Los Angeles and Phoenix, a roughly 600-kilometre journey one-way. As the truck drives, the artificial intelligence underlying its operations runs through 1,200 scenarios a second, planning moves up to one minute in advance, navigating construction, lane closures and other situations in real time.

The industry is intensely competitive, and Mr. Rodrigues is more than ready to rhapsodize about what he says is Embark’s superior technology. He also emphasizes Embark has been focused on trucks for longer than its competitors, some of whom first started working on cars. Even so, Waymo, which is owned by Google parent Alphabet, started developing autonomous vehicle technology back in 2009. In some respects, Embark is an underdog; competitors such as TuSimple and Aurora (also co-founded by a Canadian) are worth billions more today.

Still, Mr. Rodrigues is adamant Embark’s singular focus on trucking is an advantage. The company has no plans to build vehicles itself, but is instead developing a self-driving software platform compatible with four of the biggest truck manufacturers in the U.S. “It allows us to be far more scalable,” Mr. Rodrigues says. Embark will then charge carriers using its software on a per-mile basis.

Under Embark’s model, which some competitors are pursuing as well, autonomous and human drivers will work together. Self-driving trucks will traverse the highways while people handle the more complicated urban portions of a delivery. Mr. Rodrigues first foresees a human driver picking up cargo from a factory or distribution centre and taking it to a transfer hub. There, the cargo would be switched to an autonomous Embark truck, which would then drive itself to a highway and complete the longest portion of the journey, ending at a different transfer hub close to the final destination. Another human driver would pick up the load to finish the last leg.

It’s an approach the company says will save money and improve safety without decimating employment. “You end up with there being more jobs and more drivers that work locally within a given city,” Mr. Rodrigues says, speaking over Zoom from his parents’ place in Kelowna, B.C., where he was visiting. At 25, sporting a button-down shirt, he looks like he just bounded out of his first Engineering 101 class, but he fields questions like a seasoned executive. He has a few well-honed lines, deftly bats away skepticism, and perks up when asked a question he hasn’t had to answer before. In interviews, he has noted he’s spent some 15 years building robots – which is confusing until you realize he’s counting from the time he was 11.

He grew up in Calgary, and his dad, Claudio, started his own company based on a very low-tech concept: grocery store checkout dividers that contained advertising inserts. He and his wife, Jill, patented the device, which they dubbed the AdBar. As a kid, he watched his mom and some friends slip ads into the dividers and deliver them to grocery stores. The company has since expanded into other advertising and marketing services, and Jill served as executive vice-president until recently. She now runs a small business selling anti-fog face masks.

Mr. Rodrigues was a precocious kid who kept a journal for business ideas, which he and his father discussed in lieu of bedtime stories. Claudio would prod his son with questions about cost structures and staffing needs. “They weren’t the most realistic business plans,” Mr. Rodrigues says, “but they weren’t the worst.” (His sister, meanwhile, attends medical school.)

He found his calling in seventh grade, when he joined his school’s competitive robotics team. Each year, teams are challenged with constructing a robot to perform certain tasks, such as shooting basketballs. Mr. Rodrigues has a competitive streak (he also joined the debate team) and took these tournaments very seriously. The next year, his team won a world championship, and two years later, he opted to attend a different high school than his friends just to join its robotics crew. He was unimpressed with his teammates, though. “They just didn’t have the desire to be great,” he says flatly. So he changed schools, instituted its first robotics team, and carried it to the world finals.

Mr. Rodrigues brightens when talking about the robots he used to build. During our Zoom interview, he shares a video of a boxy red robot the size of a dishwasher stacking crates, built by a team he was mentoring at the time. The competitions taught him how to focus and winnow down a problem into something manageable. “When we built something super complicated, we got destroyed,” he says. “It doesn’t matter how good you are at this, you need to pick something with a narrow scope.”

After graduation, he enrolled in mechatronics engineering at the University of Waterloo in 2013, a demanding program that involves robotics, machine learning and computer science. While there, he got to know another mechatronics student named Michael Skupien. They’d actually met before, during a high-school robotics match; their teams joined forces to win a regional competition. At the time, tech companies were investing heavily in autonomous driving, and the two students, brimming with youthful confidence, wondered if they could do it, too.

They slapped together a small contraption with Lego and a cheap motion sensor, which did little more than navigate by following red objects. They decided to try something more ambitious and purchased a used golf cart online, working various connections to procure equipment. Outwardly, they presented the idea as a business venture, but in reality, they simply thought it was cool.

The summer of 2015, they assembled the first version in the garage of Mr. Rodrigues’s parents’ home in Calgary, programming it to follow a particular route. Usually, someone sat inside as a safety measure, but once Mr. Rodrigues stepped out and watched in awe as the golf cart circled around a field, completely unmanned. There was no way to stop it remotely. “The plan was, like, I was just going to run after it,” he says.

Open this photo in gallery:

When they returned to university, they recruited Brandon Moak, another mechatronics student, hailing from Nova Scotia, who took over programming efforts. (They had already consulted him on some tech questions over the summer.) All three were part of a university-backed incubator called Velocity, having enrolled during their first year. Velocity typically admitted older students, but made an exception. “They were just very mature, very focused and very friendly, and checked every box that we would want,” says Nancy Heide, who handled admissions at the time. Mr. Rodrigues struck her as a natural leader, who combined a technical mind with an outgoing personality. “He also just got away with things,” she says.

The crew, for example, needed somewhere to store their golf cart. So they erected a tent next to the Velocity building and parked it there without bothering to ask for permission. (Mr. Moak recalls they at least alerted a facilities manager.) The golf cart became a novel marketing tool for the university. “The president loved putting people in that thing and driving them around,” says Mike Kirkup, Velocity’s former director. He also recalls how confident the three of them were about the future of autonomous driving. “To them, this was so obvious,” he says. “Like, ‘I don’t understand why all of you are living in the past.’ ”

In 2016, they paused their studies and decamped to the Y Combinator incubator in San Francisco. By then, they’d incorporated a company called Varden Labs and planned to build autonomous shuttles to ferry students around college campuses. They had a few pilot projects, but it was soon obvious the business was a dud. Students had no trouble getting to class on foot, for one thing. They considered pivoting to mining vehicles and delivery robots, but settled on trucking. Mr. Rodrigues had already fielded calls from curious transportation companies.

He also heard from the administration at Waterloo, who wanted to know when he would return to finish his degree. Mr. Rodrigues always told them he wasn’t coming back. One time, an official gave him an ultimatum and Mr. Rodrigues brushed it off, explaining he had eight employees under his watch. The official switched gears and offered him the chance to return next semester. “I was like, ‘I don’t think you heard me right.’ ”

Open this photo in gallery:

That single-minded focus helps explain how, after leaving golf carts behind, Mr. Rodrigues and his small team turned Embark into a billion-dollar enterprise. They started working out of an old bicycle repair shop, and their first task was to reverse engineer a long-haul truck to take control of its steering, acceleration and braking capabilities. “That’s one of the more complex tasks that people don’t necessarily think about,” Mr. Moak says. “How do you control this system for the first time? We didn’t know if we could.”

The process took a couple of months, and the team first tested the semi-truck at a decommissioned airstrip. Mr. Skupien punched some buttons on a keyboard and watched the truck lurch forward. “It was pretty surreal, especially when this thing is barrelling at 120 kilometres per hour,” he says. It was the first of many milestones for Embark, but one of Mr. Skupien’s last acts at the company. He left Embark in 2017 and returned to Canada. Neither he nor the company can comment on his departure. All he allows is that he no longer considers Mr. Rodrigues and Mr. Moak friends.

Since then, Embark has steadily hit its milestones, honing its technology, raising money and building connections with trucking companies. But if it is to run autonomous vehicles on highways in 2024, Embark has a lot more to do. First, there are five capabilities Embark needs to develop and perfect, such as making emergency stops, evasive manoeuvres, and dealing with law enforcement, inspectors and mechanical failures – problems nobody has yet to crack. They’re not just tech issues, either, Mr. Moak points out. Conducting inspections on autonomous vehicles, for example, requires working with regulators to rethink the process.

Tackling those issues also requires funding, which is why Embark earlier retained investment bankers to explore options. Last March, those bankers contacted Ian Robertson in Oakville, Ont., the co-founder of Algonquin Power and Utilities, which he and his partner built into a $12-billion business over three decades. He left last year to dive into the world of special purpose acquisition companies (SPACs), publicly traded blank-cheque firms. Under the banner of Northern Genesis, Mr. Robertson and his partners first backed Lion Electric, a Canadian manufacturer of zero-emission commercial vehicles.

Mr. Robertson was agnostic when pitched on Embark, especially because it had zero revenue. “Generally, that was a no-go from my perspective,” he says. But like Mr. Rodrigues, he saw autonomous driving as inevitable.

Early in the talks, he arranged a trip to San Francisco. COVID-19 restrictions were still in place at the border, so Mr. Robertson, who has a pilot’s licence, flew himself south. He met with Mr. Rodrigues and climbed into an Embark truck. “It’s a weird thing to say about a computer, but there was a thoughtfulness about its driving,” Mr. Robertson says. The truck accelerated to allow a car in behind it and smoothly moved over when it came upon a temporary lane closing. When Mr. Robertson finished his ride an hour and a half later, he was convinced the technology was real and not that far off.

In June, Embark announced a merger with a Northern Genesis SPAC, which is slated to close this fall. The deal gives Embark access to more than US$600-million in cash, and includes investment from Tiger Global Management and the Canada Pension Plan Investment Board. Mr. Rodrigues will retain 39.6 per cent of the voting power, while Mr. Moak will have 29.3 per cent. The additional capital sets up Embark to achieve its goals, Mr. Robertson says, but he sounds a note of caution: “It’s a pre-revenue story, and we can’t lose sight of that.”

Embark lost nearly $37-million in the past two calendar years, plus another $29-million in the first half of 2021. But it anticipates making money very rapidly. In 2024, Embark projects it will bring in US$867-million in revenue and turn a profit of US$584-million – a 67-per-cent margin. Embark’s estimates leave zero room for error, says Richard Windsor, founder of Radio Free Mobile, a technology research firm. “It’s priced to perfection,” he says. “If they make those numbers, I can get behind a $5-billion valuation. But it’s asking an awful lot to pay up for it at that price now.”

Open this photo in gallery:

Talking to Mr. Rodrigues, it’s easy to become convinced he really will have autonomous trucks on the highways in 2024. He’s so breezily certain, so measured in his approach, so accomplished at a young age that surely he’ll pull it off, right?

“Overly optimistic,” is the blunt assessment of Missy Cummings, a professor and director of the Humans and Autonomy Lab at Duke University. Machine-learning systems lack top-down reasoning, she says, meaning they can’t adequately decipher contextual clues to make sense of the world. Such systems can be flummoxed by tree branches obscuring road signs and have been unable to recognize familiar objects in unfamiliar poses, such as a motorcycle lying on the road.

The solution engineers have proposed is simply to feed AI systems more data. “It begs the question as to how much of this finger-in-the-dyke engineering is practical or possible?” Prof. Cummings wrote in a paper. Already, Tesla drivers have been killed while using that company’s driving assist feature, and a self-driving Uber killed a pedestrian in Arizona in 2018, even with a safety driver onboard. While highway driving might be less complex than navigating congested downtown streets, tractor trailers are huge objects that can do catastrophic damage. Prof. Cummings also faults autonomous vehicle developers for not disclosing detailed safety and hazard analyses reports. “Until we can see those kinds of data, no one can have any certainty about whether these vehicles are safe,” she says.

A U.S. Department of Transportation database shows Embark trucks have not been involved in any collisions, and the company says it has not even experienced a minor accident. Safety drivers have been present at all times, though, so record doesn’t provide much indication of how the truck will perform with no driver at all. (The company will only start testing with no safety drivers in 2023.) Embark maintains trucking is simply an easier problem to tackle and thus achievable. “We have this opportunity to pick the routes that we understand to be good from a construction, weather pattern and traffic perspective,” Mr. Moak says. “You don’t need this higher-level order of reasoning, just because the environment itself ends up being inherently simpler.” Still, unpredictable events can happen on highways, he acknowledges. “Again, it’s a game of what is feasible. What is the frequency of these things?”

Success also depends on how quickly carriers adopt autonomous vehicles. Embark, like its peers, is selling its technology as a way to deal with labour shortages and save money. (Autonomous drivers never sleep, of course.) Embark estimates it can save carriers 80 cents a mile, allowing customers to nearly double their profit margins. It’s already running pilot projects to transport cargo between Los Angeles and Phoenix for a variety of companies, including HP Inc. and Werner Enterprises, a US$3-billion transportation company headquartered in Nebraska.

Carriers are interested, but generally in no rush. Chad Dittberner, a senior vice-president at Werner, says the company would only use such vehicles if they are proven to be safer than a human driver. “There’s going to have to be miles put on the roadways,” he says, “and we’re talking lots of miles.” The rate of industry adoption will be gradual, he adds, in part because the lifespan of a long-haul truck is around nine years. “Do I think there’s a light switch in 2025 or 2026 and all trucks being purchased go to autonomous?” he says. “No, that’s not going to happen.”

Open this photo in gallery:

Automation also brings long-term consequences. Embark describes the implications in almost utopian terms: cost savings, safer roads and jobs for truckers. Even as long-haul routes are automated, drivers will be needed for short routes, it claims. During our interview, Mr. Rodrigues pointed to a recent study funded by the U.S. Department of Transportation that found autonomous trucks could increase annual earnings for all workers by at least US$203 a year through productivity improvements and that long-haul drivers would not suffer mass layoffs.

Others aren’t so sanguine. Steve Viscelli, a professor at the University of Pennsylvania who has extensively studied trucking, estimates some 294,000 long-haul jobs are at risk in the U.S., including some of the highest-paying positions in the industry. Any short-haul jobs created as a result of increased demand are likely to be far worse than those that are lost, he says. Short-haul drivers already deal with low wages and long periods of unpaid waiting. Absent public-policy intervention, Prof. Viscelli says, there is little reason to believe conditions will improve. “We have essentially 40 years of history that tell us that new technology in truck driving makes jobs worse,” he says. “The burden of proof is definitely on the developers to show why this particular technology is going to do something completely different.”

Embark’s own filings note the unpredictability of the endeavour. “There is, therefore, substantial uncertainty as to whether Embark’s business plan will prove successful,” reads the company’s 40-page risk section. “I’m not going to reply directly to that,” Mr. Rodrigues says when asked about it. “Embark has a track record of being very successful at delivering this technology and being a leader – if not the leader – at hitting these milestones. At the same time, this is a thing no one has ever done before.”

It’s hard to fault Mr. Rodrigues for such boundless confidence. After all, his story so far is one of success after success, from building a rudimentary self-driving Lego vehicle to helming a multibillion-dollar autonomous trucking firm, all before turning 30. Rather than be discouraged by the enormous task that remains, he and Mr. Moak have already seen the arc technology follows. Something that once seemed like science fiction quickly becomes mundane. Back when they were shuttling students around campuses, the passengers were initially amazed to see a vehicle drive itself. After a few minutes, their heads were bowed, lost to their smartphones, like it was just another ride.

Your time is valuable. Have the Top Business Headlines newsletter conveniently delivered to your inbox in the morning or evening. Sign up today.