Even experienced drivers can be forgiven for missing a roundabout exit once or twice, but a disoriented robotaxi in Arizona did it 36 times… in a row. While Waymo taxis are among the most advanced autonomous vehicles on the road today, in a video posted earlier this month on X, a confused AV appears to be quite literally stuck in a loop. (Waymo clarified there was thankfully no passenger on board the high-tech merry-go-round.)
Sorry I’m late, my WAYMO did 37 laps in the roundabout 💁🏻♂️ pic.twitter.com/GSR4sqChV2
— Greggertruck (@greggertruck) December
11, 2024
And yet, even with blunders like these, there were more vehicles driving themselves this year than ever before. Once cordoned off to a few test tracks and small patches of land in Mountain View, AVs are now rearing their sensors-flapping heads in more than a dozen US cities. Tens of millions of drivers, cyclists, and pedestrians are learning how to coexist amongst these machines while their shared roads serve as real-world test-beds for full-scale AV deployment. Sure, it’s not The Jetsons, but 2024 was the year driverless cars hit their stride and flooded the streets–in other words, ready or not, they got real.
Whether or not that’s a good thing depends entirely on who you ask. Several computer science experts and civil engineers told Popular Science that they were confident these early autonomous vehicles (AV’s), slow and confused as they might seem sometimes, are the harbingers of a near-future marked by considerably less deadly damage caused by distracted humans. Others, were less optimistic. The actual on-the-ground performance of these cars in 2024 remains a mixed-bag at best.
One thing seems certain: the AV race is on. Though many companies have faded, some dramatically, a handful of major players like Alphabet’s Waymo, Amazon-backed Zoox, and Aurora are bulking up and scaling up operations. In the process, they are changing the way millions of people interact on the road. And it’s just getting started.
Waymo’s strategy? Slow and steady wins the race
Though this proliferation of AVs can feel sudden to the average person, Carnegie Mellon University Professor Ragunathan Rajkumar told Popular Science the road there was incremental. Rajkumar, who has worked on autonomous research for decades, says the first example of what one could consider a rudimentary autonomous vehicle dates all the way back to 1984 at Carnegie Mellon. Progress in the space leaped forward twenty years later when DARPA, the US Defense Department’s experimental research arm, held a driverless car competition called the “Grand Challenge.” The teams that won attracted the attention of Google co-founders Larry Page and Sergey Brin. That eventually led to the creation of what would become Waymo and helped turbo charge the current era of privatized driverless car development.
“The [2007] competition showed it was actually feasible,” Rajkumar siad. “It made this singular point that autonomous vehicles are only a question of when and not a question of if anymore.”
Waymo was not only one of the earliest entries into the space, but also one of the most uniquely suited to make their product commercially available in 2024. Being tied to Google gave Waymo all the technical expertise and crucially, the deep pockets it needed to slowly build up its testing apparatus over the course of several years without a pressing need for an immediate return on investment. The company started in relatively sparse, Arizona suburbs and has gradually built up to more complex environments over time. Rajkumar said this approach differs from competitors like GM-backed Cruise which started later and may have tried to rush over some steps in a race to make a commercially viable product. That resulted in errors and an immediate degradation of public trust. By the time they were ready to test again this year, Waymo had already gained too much ground. In other words, slow and steady really does win the race. (GM announced it would absorb Cruise and transition operations away from taxis and towards fully autonomous personal vehicles earlier this month.)
Driverless cars are operating in over a dozen US cities
2024 marked a “milestone” in driverless vehicle deployment according to Autonomous Industry Vehicle Association (AIVA) CEO Jeff Farrah. He told Popular Science that 25 states are now on the policy path towards some AV deployment. These new laws will cover around 56% of the US population. Waymo, after testing for the better part of a decade, currently offers fully autonomous commercial rides in San Francisco, Los Angeles, and Phoenix through its app. As of late October 2024, Waymo said it was offering 150,000 paid trips every week. Those numbers will balloon further next year when the company begins offering paid trips in Austin and Atlanta through a partnership with Uber.
Waymo isn’t alone either. Zoox—an Amazon-backed robotaxi service that is working on deploying a purpose-built, bi-direcitonal vehicle with no steering wheel and four seats facing towards each other—is currently testing their fleet in Las Vegas, Seattle, Austin, and Miami. Cruise resumed testing in Dallas, Houston, and Phoenix, earlier this year following a spate of missteps in 2023 that saw it lose its license to operate in California. Tesla, which is controversially trying to achieve full autonomy without using LiDAR technology, charted out plans to begin production of a “Cybercab” robotaxis service sometime before 2027.
[ Related: Why are ‘driverless’ cars still hitting things? ]
Taxi-style ride hailing services also aren’t the only types of self-driving vehicles already filling US roads. Mercedes-Benz, last year, gained approval to sell its “Drive Pilot” feature, a SAE Level 3 system that lets drivers legally operate a vehicle up to 40mph hands free and without their eyes facing the road. Aurora, a company racing to develop autonomous big rig trucks, has been conducting self-driving tests (with a human safety driver present) in Texas for several years. The company recently pushed back plans to let trucks drive free of human operators in the state next year.
“It’s clear that AVs are here to stay and as more consumers experience them, the data shows they are enjoying their experience,” Farrah told Popular Science.
American are learning to live alongside new autonomous neighbors
For cities like San Francisco and Austin, Texas where hundreds of autonomous vehicles are making their way around town at any given time, the presence of cars without a human driver has gone from novel to common. Though they once regularly solicited gawking pedestrians and yells from angry pickup drivers, AVs in 2024 increasingly blend into the urban traffic. That seeming sense of normalcy on the ground runs counter to broader snapshots of American sentiment. Only 37% of US adults surveyed by Pew Research in 2022 said they would want to ride in a driverless vehicle. And yet, in places where driverless options are actually available, early reports show some riders are being won over. Part of that acceptance may be due to their passive behavior.
“They [robotaxis] tend to drive very cautiously,” University of Texas Professor of Transportation Engineering Kara Kockelman told Popular Science. “They look and feel much less dangerous than other vehicles driven by 19-year-olds who want a speed rush.”
Nicole Moore, the President of Rideshare Drivers United and herself a part-time Lyft driver had another way of describing them:
“It’s slow, stoned driving,” Moore told Popular Science.
Driverless cars could cut down on crashes…one day
Proponents of autonomous vehicles say the billions of dollars and investment and countless hours of testing are in the service of two main overarching goals: safety and convenience. On safety, experts told Popular Science a world awash with AVs could meaningfully cut down on the estimated 3,000 auto deaths linked to distracted driving in the US every year. Globally, around 1.3 million people die every year in auto accidents. Kockelman, who has conducted academic research on the cost-benefit analysis of AV fleets on public roads, said she expects driverless tech to potentially cut down on 80-85% of costs associated with auto crashes, a figure she puts close to $1 trillion. That’s not counting other non-economic costs resulting from crashes too like pricey doctors’ visits and long-lasting mental and emotional scars. A recent study conducted by Waymo and insurer Swiss Re analyzing insurance claims data found Waymo vehicles resulted in less property damages and fewer injuries than human drivers.
Aside from the safety argument, Rajkumar said he was optimistic AVs would also empower legally blind individuals and others with physical disabilities to have greater degrees of mobility and autonomy. For others, time spent saved from focusing on driving could be spent completing a meeting, speaking with family, or simply taking a nap.
“The arc of the progress has been happening,” Rajkumar said.
Maybe. But that vision wasn’t what was on offer in 2024.
It doesn’t take long scrolling through social media to find shocking examples of supposedly advanced autonomous vehicles appearing undeniably stupid. There’s these accounts of Cruise driverless vehicles causing traffic jams in San Francisco and Austin. In other cases, a fleet of recharging Waymo vehicles mysteriously started honking at each other, much to the annoyance of nearby apartment residents trying to sleep. A Cruise vehicle operating in San Francisco last year drove into a city paving project and got stuck in wet cement. In Austin, a server at a local taco shop told this reporter about a recent case of a Waymo seemingly getting trapped in the restaurant’s parking lot, apparently mistaking it for a side-street.
“Everyone went outside and started clapping when it drove away,” he said.
Real-world testing isn’t living up to the hype
But there are more serious concerns as well. For starters, research looking into the limited test cases of AVs currently operating on public roads hasn’t met the high-bar set by some of the tech’s most vocal boosters. A recent analysis of self-driving vehicles operating in San Francisco conducted by researchers from George Mason and Duke University concluded many areas of improvement were needed. The paper, which compared crash reporting data from AVs and human drivers, claims AV computer vision systems were at times “very brittle” and can “play an outsized role in crashes.
In some notable cases, self-driving vehicles were more likely to be involved in accidents than humans. The self-driving vehicles reviewed were struck from behind at a rate 1.7 times higher than human drivers. And while AV-makers often blame these incidents on other distracted human drivers, the researchers say AV can surprisingly decelerate or slam on the brakes “for no obvious reason.” The researchers also noted a “substantial” number of crashes that occurred from AVs seemingly unable to properly understand social norms around driving.
Those issues can lead to tragedies when scaled up. In 2018, the first reported human death linked to an autonomous vehicle occurred when an Uber self-driving car fatally struck a 49-year-old woman crossing a road in Temple, Arizona. An investigation following the death found the vehicle’s software wasn’t prepared to expect pedestrians crossing streets outside of crosswalks. Last year, a Cruise vehicle collided with and then dragged a pedestrian for 20 feet. Experts told Popular Science the brutal dragging incident, which contributed to Crusie losing its license to operate in California, was likely the result of the company failing to include sensors underneath its vehicle.
Though fully-autonomous vehicles like those operated by Waymo and Zoox haven’t resulted in human fatalities thus far, they also mostly aren’t yet operating on more dangerous highway speeds. This year, just as more AVs were filling US roads, the National Highway Traffic Administration (NHTSA) announced a pair of investigations into Waymo and Zoox over allegations their vehicles had “potentially violated traffic safety laws.”
A Waymo Spokesperon told Popular Science it currently provides 150,000 paid trips. Weekly “in some of the most challenging and complex environments.”
“We are proud of our performance and safety record over tens of millions of autonomous miles driven, as well as our demonstrated commitment to safety transparency,” the spokesperson said. “NHTSA plays a very important role in road safety and we will continue to work with them as part of our mission to become the world’s most trusted driver.”
‘This is the wild, wild west of deregulated transportation’
There’s also the still unresolved question of what happens to the more than one million human and Uber Lyft drivers who are increasingly finding themselves in competition with driverless cars. Moore, the Lyft driver and Rideshare Drivers United President says she has watched as AVs have proliferated around her in Los Angeles. In the past, her and other drivers would share videos of the cars getting stuck or malfunctioning for a quick laugh. But recently, Moore says she saw a Waymo vehicle pick up a passenger and suddenly had a realization.
“Oh my god,” she recalls. “That [Waymo] is driving somebody that I could be taking!”
That’s a thought many more gig workers may have to come to terms with in the months and years to follow. Uber, the leading ride-hailing service, has plans to let riders request Waymo vehicles from its app in Austin and Atlanta next year. And while Waymo’s still slightly more expensive options and have limited routes, that may soon change. Kockelman and Rajkumar agreed robottaxis may become more appealing to riders but they noted it would likely take time before they mean guly chip away at human gig-worker number. Human drivers will still be needed for the immediate future to operate in inclement weather and high speed situations. Even in markets where robotaxis proliferate, ride-share companies may still rely on humans to meet rider demands during peak periods.
“It’s going to be a slow rollout just like electric vehicles,” Kockelman said.
Those words aren’t all that reassuring for professional drivers like Moore.
“Every Waymo that people get into is costing a driver. There’s no doubt about that,” she said. “It’s the same market. This is the wild, wild west of deregulated transportation.”
And while 2024 saw the realization of driverless vehicles for many, it also previewed a brewing backlash from an impassioned minority. Waymo is currently suing a San Francisco man prosecutors alleged was responsible for slashing the tires of 17 robotaxis between June 24 and June 26, sometimes with riders present. The slashing incidents came several months after a crowd of people, also in San Francisco, reportedly vandalized a Waymo, tossed fireworks into it, and ultimately set it ablaze. People in the crowd cheered and posed for photographs next to the burning wreckage. Anti-driverless car activists previously placed traffic cones on the hoods of cars to trick AV sensors and temporarily disable them.
An autonomous Waymo ride-hailing vehicle is vandalized and set on fire in San Francisco, February 2024. pic.twitter.com/iamQdKN47j
— Future Adam Curtis B-Roll (@adamcurtisbroll) February
18, 2024
Farrah of the AVIA said driverless vehicles are at an “inflection point” in the US. As more states allow them to operate on public roads, there may be increased pressure for new federal legislation mandating a set of standards. Driverless vehicles are present in numerous cities but actually having one drive across the country would be a legal nightmare. Companies like Waymo and Zoox meanwhile, which have had years of generous coddling from wealthy investors to train up their fleets, may increasingly feel pressure to make their products financially viable.
2024 may have been the year driverless cars became “real” for many Ameircans, but 2025 will be the year we find out just how disruptive, both positively and negatively, this technology may really wind up being. And while that transformation can feel gradual in the moment, it can also snowball in what feels like an instant.
“The technology is going to take a while to develop, but when you think about it, AI just popped out and all of a sudden it was there,” Moore said. “We’re dealing with a pop-out [moment] here.”
