Self Driving Cars 101
← Back to Learn

Foundations

ADAS: Driver Assistance, Explained

ADAS, short for advanced driver assistance systems, are features that help a human drive, such as automatic emergency braking, lane keeping and adaptive cruise control. They sit at SAE Levels 0 to 2: however capable they feel, the person behind the wheel is still the driver and must stay attentive at all times.

What counts as ADAS?

NHTSA describes the driver assistance features on today's cars as helping the driver but not self-driving: the driver must stay fully engaged. The features fall into three SAE levels. Warnings and brief emergency interventions are Level 0, a single continuous assist with steering or speed is Level 1, and combined steering and speed control is Level 2.

Regulators are making the basics mandatory. In the European Union, all new cars and vans have required automatic emergency braking, emergency lane keeping, intelligent speed assistance and drowsiness warning since July 2024. In the United States, 20 automakers voluntarily made automatic emergency braking standard on nearly all their light vehicles by 2022, and a 2024 rule currently requires it on new light vehicles by September 2029, although that deadline is under reconsideration.

FeatureWhat it doesSAE level
Automatic emergency braking (AEB)Brakes on its own if a collision is imminent and the driver does not react0
Lane departure and blind spot warningAlerts the driver0
Adaptive cruise controlKeeps a set speed and gap to the car ahead1
Lane keeping assistSteers to keep the car in its lane1
Adaptive cruise with lane centeringControls speed and steering together, with the driver supervising2
Hands-free highway systems (GM Super Cruise, Ford BlueCruise)Level 2 control on mapped highways, with the driver's eyes monitored2

Read: The SAE levels of driving automation →

Where did ADAS come from?

Its precursors are older than any of today's sensors. Bosch and Mercedes-Benz brought anti-lock brakes to the S-Class in 1978 and electronic stability control in 1995. Neither senses the world around the car, so they are usually classed as vehicle safety systems rather than ADAS, but they made the car's brakes something a computer could command.

Driver assistance that watches the road arrived in the mid 1990s, in the same years as the research cars that first drove across countries. Mitsubishi offered the first production adaptive cruise control in Japan in 1995: a laser sensor that eased off the throttle to keep a gap, but could not brake. Toyota followed in 1997, and Mercedes-Benz launched Distronic, widely cited as the first radar-based adaptive cruise control, with its 1998 S-Class. Warnings and emergency braking followed in the 2000s, from truck lane departure warning to Volvo's City Safety, the first low-speed automatic braking fitted as standard.

YearSystemWhat it did
1995Mitsubishi adaptive cruise control (Japan)Kept a gap with a laser sensor, without braking
1997Toyota laser adaptive cruise control (Celsior, Japan)Distance-keeping cruise at 40 to 100 km/h
1998Mercedes-Benz Distronic (S-Class)Radar-based adaptive cruise control
2000Iteris lane departure warning (Mercedes Actros trucks)Warned when a truck drifted out of its lane
2003Honda and Toyota collision systems (Japan)Radar that warned of a crash and could apply the brakes
2008Volvo City Safety (XC60)Low-speed automatic braking, fitted as standard

Read: Meanwhile in the labs: the first self-driving cars →

Who builds ADAS today?

ADAS is built in layers. Carmakers decide what goes into a car and brand the features, such as GM's Super Cruise or BYD's God's Eye, which BYD made standard on 21 models in 2025 down to its cheapest car. Tier 1 suppliers integrate the cameras, radars, control units and software. Chipmakers supply the processors that run the perception software, and some carmakers build that part themselves.

The layers often combine: the supplier Valeo, for example, has built more than 20 million front camera systems around Mobileye's EyeQ chip. Mobileye is the long-time leader in camera processors, with about 52 percent of ADAS and automated driving processor revenue in 2022 by Yole Group's estimate, but NVIDIA, Qualcomm and Horizon Robotics have grown fast, and Tesla, Xpeng and Huawei develop their own.

LayerExamplesRole
CarmakersGM, Ford, Mercedes-Benz, Toyota, BYD, TeslaChoose, brand and sell the features
Tier 1 suppliersBosch, Aumovio (Continental's former automotive group), Denso, ZF, Aptiv, Valeo, Magna, Hyundai MobisBuild and integrate sensors, control units and complete systems
Chips and computeMobileye, NVIDIA, Qualcomm, Horizon RoboticsProcessors and software platforms for perception and driving
In-houseTesla, Xpeng, Huawei (for partner brands)Carmakers or tech firms designing their own chips and software

From ADAS toward autonomy: the incremental path

Several companies are trying to climb from driver assistance to full autonomy one step at a time, the opposite of Google's decision to skip the supervised levels. Mobileye is the clearest example. Founded in Israel in 1999 by Amnon Shashua and Ziv Aviram, it put camera-based lane departure warning into GM and BMW models in 2007 and has since reached more than 230 million vehicles. Its product line now runs from basic ADAS to a robotaxi system, each step taking on more of the driving.

Others are on similar ladders. NVIDIA's DRIVE platform powers both Level 2 assistance and robotaxi programs. Qualcomm bought Veoneer's Arriver software in 2022 and launched hands-free highway assist with BMW in 2025. Horizon Robotics moved from front camera chips to a city driving system in 2025. Tesla went from Mobileye chips in its first Autopilot to its own chips and FSD (Supervised).

Mobileye productWhat the driver doesStatus (September 2026)
EyeQ-based ADASDrives, with warnings and assistsIn production cars since 2007
SuperVisionTakes hands off on supported roads, keeps eyes on the roadSince 2021, first in the Zeekr 001
ChauffeurTakes eyes off in defined conditions (Level 3)Planned, not yet shipping
DriveNothing; there is no driver (Level 4)Robotaxi testing with safety drivers, including VW ID. Buzz with MOIA

Read: The opposite bet: Google's all-or-nothing approach →

Why every dollar matters: ADAS versus a robotaxi

ADAS is sold inside privately owned cars. Every sensor and chip adds to the sticker price of millions of vehicles, and buyers compare prices, which is why most systems rely on cameras and radar. Few suppliers publish prices, but Mobileye does: its average price per system in late 2025 was about $51, and in the filing for its first stock offering it wrote that consumers are interested in safety but are also very sensitive to costs.

A robotaxi works differently. The operator pays for the vehicle once and recovers the cost across thousands of paid rides, so it can carry a much richer sensor suite, including lidar. A Mobileye executive said in August 2026 that its robotaxi system costs about $40,000 per vehicle, against roughly $1,300 for its hands-off highway system. That gap is why lidar has mostly stayed in robotaxis.

Consumer ADASRobotaxi
Who paysThe car buyer, in the sticker priceThe fleet operator
How the cost is recoveredIt is not; it is part of the carAcross thousands of paid rides
Typical sensorsCameras and radarCameras, radar and lidar
Price pressureEvery dollar, times millions of carsHardware cost per mile driven

Level 2 hands-free, and the Level 3 question

GM's Super Cruise, launched on the 2018 Cadillac CT6, was the first hands-free highway system, and Ford's BlueCruise followed in 2021. Both are Level 2: the car steers and controls speed on mapped highways, and a camera checks that the driver is watching the road.

Level 3 is where the system becomes the driver in defined conditions. Mercedes-Benz launched Drive Pilot in Germany and later Nevada and California, the first Level 3 system approved in the United States. In January 2026, Mercedes paused it in favor of a more advanced Level 2 system, and BMW was reported to have stepped back from Level 3 as well. The handoff between car and human remains the hardest part of the middle levels.

Read: Robotaxis: skipping the handoff entirely →

Watch

Frequently asked

What is ADAS?
ADAS stands for advanced driver assistance systems: features such as automatic emergency braking, lane keeping assist and adaptive cruise control that warn the driver or briefly act to help avoid a crash. The driver stays responsible at all times.
Is ADAS the same as self-driving?
No. NHTSA says the driver assistance features sold on cars today are not self-driving and need a fully attentive driver. Self-driving means the system is responsible for driving, which starts at Level 3.
What level is ADAS?
Between Level 0 and Level 2. Warnings and emergency braking are Level 0, a single steering or speed assist is Level 1, and combined steering and speed control, including hands-free highway systems, is Level 2.
When did ADAS start?
Sensing driver assistance began with adaptive cruise control in Japan in 1995, followed by radar cruise control, lane departure warning and collision braking in the late 1990s and 2000s. Anti-lock brakes (1978) and stability control (1995) were earlier precursors.
Who makes ADAS systems?
Carmakers, working with suppliers such as Bosch, Aumovio, Denso, ZF, Valeo and Magna, and chipmakers such as Mobileye, NVIDIA, Qualcomm and Horizon Robotics. Some carmakers, including Tesla, build their own.
Why don't most consumer cars use lidar?
Driver assistance is sold in privately owned cars, where every component adds to the price. Robotaxi fleets can spread the cost of lidar-rich sensor suites across thousands of paid rides, which is why lidar has mostly stayed in robotaxis.

Keep learning, once a week

The weekly digest: the stories worth your time from the autonomous vehicle industry, hand-picked by people who work in it. No marketing gloss.

Related