- Car companies are rapidly expanding technologies that can control the acceleration, braking and steering of a vehicle. In some cases, allowing drivers to not control the car for miles at a time.
- CNBC's Michael Wayland recently tested ADAS from Tesla, GM and Ford. Their systems are among the most readily available and dynamic on the market.
- To be clear, no vehicle on sale today is self-driving or autonomous. Drivers always need to pay attention.
DETROIT – Letting go is hard. Even if major automakers want to make it easier.
Car companies are rapidly expanding technologies that can control the acceleration, braking and steering of a vehicle. In some cases, allowing drivers to ease off the steering wheel or pedals for miles at a time.
The systems – formally known as advanced driver assistance systems (ADAS) – have the potential to unlock new revenue streams for companies while easing driver fatigue and improving safety on the road. But automakers have largely built their systems independent of one another, without industry-standard guidelines by federal regulators. That means years into development, "hands-free" or "semi-autonomous" can mean something very different in the hands of rival automakers.
To be clear, no vehicle on sale today is self-driving or autonomous. Drivers always need to pay attention. Current ADAS mostly use a suite of cameras, sensors and mapping data to assist the driver and also monitor the driver's attentiveness.
The automaker most often discussed alongside ADAS is Tesla, which has a range of technologies that it haphazardly calls "Autopilot" and "Full Self-Driving Capability," among other names. (The vehicles do not fully drive themselves.) But General Motors, Ford Motor and others are quickly releasing or improving their own systems and expanding them to new vehicles.
I recently tested ADAS from Tesla, GM and Ford. Their systems are among the most readily available and dynamic on the market. However, none of them were close to flawless during my time behind the wheel.
And even small differences across the systems can have a big impact on driver safety and confidence.
I initially tested GM's system a decade ago on a closed track, and the automaker's years developing Super Cruise have clearly paid off in overall performance, safety and clear communication with the driver. It's the best-performing and most consistent system.
GM initially released Super Cruise on a Cadillac sedan in 2017 – two years after Tesla's Autopilot – before expanding it to 12 vehicles in recent years. It aims to make Super Cruise available on 22 cars, trucks and SUVs globally by the end of 2023.
The system allows drivers to operate "hands-free" when driving on more than 400,000 miles of pre-mapped divided highways in the U.S. and Canada. (Ford has mapped 150,000 miles, and Tesla's system hypothetically operates on any highway.)
Super Cruise is the front-runner when it comes to highway driving and can handle most challenges, including curves and many construction zones. Its newest updates also added automatic lane changes that work quite well to maintain a set speed by avoiding slower vehicles.
Over hundreds of miles driving the system, I was able to regularly engage Super Cruise for upward of 30 minutes, even stretching one stint to more than an hour without ever having to take control of the vehicle. When Super Cruise did disengage, it would typically be available again minutes, if not seconds, later.
The majority of problems I experienced were likely due to outdated mapping data that the system requires to operate, according to GM. When there's newly finished construction or heavier temporary work being done, GM's system defaults to returning control back to the driver until the road is properly pre-mapped.
GM says it has produced more than 40,000 vehicles equipped with Super Cruise, though not all of those represent active users, and has racked up more 45 million hands-free miles.
Pricing for the system varies based on vehicle and brand — $2,500 for a Cadillac, for example — and carries a subscription cost of $25 per month or $250 per year after a free trial period.
Ford's system is the newest of the three automakers and is similar to GM's. Besides pre-mapping and stated capabilities, both systems feature in-vehicle infrared cameras to ensure drivers are paying attention. But if GM's system is a capable and confident "driver," Ford's is still a teenager learning, albeit very quickly.
Ford's system – marketed as Ford BlueCruise and ActiveGlide for Lincoln – first became available in July 2021, though the company has already expanded the systems to more than 109,000 enrolled vehicles with more than 35 million hands-free driving miles through the end of November.
Pricing for Ford's system varies based on the brand and vehicle. It can be part of optional packages that run roughly $2,000 and include other features for the 2023 Ford F-150 and Mustang Mach-E. Like GM, it requires a subscription after trial periods.
Also like GM, Ford's system functions well on highways … that is until it doesn't. It's less predictable and specifically struggles with larger or sharper curves, construction zones and under other circumstances a human driver would easily be able to handle.
The longest I was able to go hands-free with Ford's system during my test drives, which largely took place on I-75 and a construction-laden I-94 in rural and urban areas of Michigan, was 20 minutes and about 25 miles.
That's a problem when you're attempting to ease driver fatigue and increase drivers' confidence in such systems.
"Having it randomly disengage when you're approaching curves in the road, it's not good enough," said Sam Abuelsamid, a principal analyst at Guidehouse Insights, who specializes in advanced and emerging automotive technologies.
Chris Billman, chief engineer of ADAS vehicle systems integration at Ford, stressed that the company is being overly cautious with its system at this stage. Despite the warnings to retake control, the system is designed to remain in operation until the driver takes over.
Billman said the system disengages on most large highway curves because it's not currently designed to slow the vehicle down ahead of a curve – something Super Cruise launched with in 2017. That's expected to be improved with the system's next major update, beginning early next year.
Ford could also improve its system's interactions with the driver. GM uses a lightbar on the steering wheel and communications in the driver cluster — the best communication features among the three current systems.
That's not to say Super Cruise isn't still learning.
Both Ford and GM systems would have likely hit a temporary concrete construction barrier if I hadn't taken over and disengaged on a large S-curve roadway near Detroit.
Super Cruise and BlueCruise both disengaged several times for what seemed like no reason, only to reengage quickly after. Super Cruise also attempted to merge into a breakdown lane or median in a newly finished construction zone, while Ford's did a similar maneuver halfway through a curve.
And of course, neither system operates on city streets like Tesla's.
Tesla's technology is by far the most ambitious of the three and operates well on the highway. But it can be nerve-wracking, if not dangerous, on city streets, specifically turning into traffic.
Tesla vehicles come standard with an ADAS known as Autopilot. However, owners can upgrade the system with additional features, for a cost. The Full Self-Driving (FSD) upgrade currently costs $15,000 at the time you purchase a vehicle, or a monthly subscription opted into later costs between $99 and $199 depending on the vehicle, according to Tesla's website.
I was able to use three Tesla levels of the system with varying functionality in a Tesla Model 3 built in 2019. Driving with the FSD Beta (version 10.69.3.1) was among the most stressful driving moments in my life (and I've had a lot!).
During a limited test on the highway, Tesla's systems functioned very well. The trip included automatic lane changes and navigation-based exiting, although it did overshoot one exit ramp due to traffic. GM and Ford don't currently link navigation to ADAS.
Tesla's ADAS is also able to identify traffic lights on city streets and act accordingly, which was very impressive.
One of my biggest problems with Tesla's system on the highway was how frequently it asked me to "check in" – an action that requires tugging on the steering wheel to prove the driver is physically in the driver's seat and paying attention. The "check-ins" take some getting used to so the system doesn't disengage.
I also struggled with the car's communication about when the system was engaged.
Unlike Ford and GM that prominently show when the system is engaged, the only indication that Tesla's ADAS is engaged is a tiny steering wheel icon – smaller than a dime – in the top left of the vehicle's center screen. (The Tesla Model 3 doesn't have display screens in front of the driver.)
That means to confirm whether the system is engaged, the driver has to actually look away from the road. And if the system disengages it doesn't communicate that very well, leaving the driver unaware when the system is operating and anxious.
Such problems were even more striking while FSD Beta was operating on surface streets. In addition to the highway problems, the system – as documented in countless YouTube videos – has difficulties with some turns.
Add in what's known locally as a "Michigan left" – a median U-turn crossover – and the system turns into the equivalent of a young, if not dangerous, student driver. At one point while performing such a maneuver, the Tesla stopped across not one, but three lanes of traffic as it attempted to make the turn before I overtook the system.
On straight, crowded streets of suburban Detroit, Tesla's system largely worked well. But it lacked the experience to recognize human driver nuances such as stopping to allow others into a lane. It also had some difficulties with lane changes and seemed to be lost when lane markings were not available.
All of these concerns are why no other company has released a system like Tesla's FSD Beta, which has been criticized for using its customers as test mules. Tesla did not respond to a request for comment on this article.
CEO Elon Musk for several years has promised the vehicles would be capable of fully driving themselves. In a recent argument in response to a lawsuit filed in California, Tesla said that its "failure" to realize such a "long-term, aspirational goal" didn't amount to fraud and that it would only achieve full autonomous driving "through constant and rigorous improvements."