The first
1995 · Mitsubishi Diamante (Preview Distance Control)
Mitsubishi's Preview Distance Control, fitted to the Japanese-market Diamante in 1995, was the first production car system that used a forward-facing sensor (in this case a lidar unit, not radar) to detect the car in front and reduce engine torque to maintain a set following distance. It didn't apply the brakes. It only worked above 25 mph. It cost roughly £1,500-equivalent on a £20,000 car, take-up was negligible, and Mitsubishi pulled the option after two years.
The reason it didn't catch on is unglamorous: the lidar struggled in fog, rain, and bright sunlight, and the system only modulated throttle, so it was effectively a glorified cruise control. If the car in front actually stopped, you had to brake yourself.
What's worth noticing is what Mitsubishi's engineers had already worked out by 1995. The basic recipe of an ACC system — forward sensor, distance setpoint, throttle reduction, integration with cruise — was all there. The next twelve years were spent fixing reliability and adding the missing pieces.
The version that made it stick
1998 · Mercedes-Benz S-Class W220 (Distronic)
Mercedes launched Distronic on the W220 S-Class as a £1,400 option. It used a 76 GHz radar (not lidar) mounted behind the grille, worked from 25 mph upwards, modulated both throttle and the brakes (up to about 20% braking force), and came integrated with the cruise control on a single stalk.
What Distronic actually proved was that the radar was the right answer. Lidar in 1995 was expensive (£800+ per unit), large, and weather-sensitive. The 76 GHz millimetre-wave radar that Mercedes adopted from Continental was £180 in 2000, half the size, and worked through fog, rain, and salt spray. Every other manufacturer that introduced ACC over the next decade used the same radar approach.
Take-up on the S-Class was approximately 8% in the first model year. By the 2002 update Distronic was a £900 option and take-up was over 30%. By 2005 Mercedes had it on the C-Class. The trickle-down accelerated. By 2010, BMW, Audi, Volvo, Lexus, Cadillac, and Infiniti had launched competing systems. By 2015 it was on the Vauxhall Astra.
ACC went from zero production cars to over half the new-car market in seventeen years. It took ABS forty.
The "stop and go" turning point
The killer feature of modern ACC isn't the sensor — it's the ability to stop and restart. The first generation of ACC systems had a 25-30 mph minimum speed because the radar (and the brake hardware behind it) couldn't reliably bring the car to a complete halt. Below the threshold, the system disengaged and beeped at you to take over.
Mercedes added "Stop & Go" capability to Distronic Plus on the 2005 S-Class W221. It was the first system that could hold the car stationary in traffic, then resume from a complete stop when the car in front moved. The software was harder than it looked: predicting whether a stopped car was actually going to move again, distinguishing between "this is a queue, follow it" and "this is a parked car, change lanes."
Ford brought stop-and-go ACC to mainstream cars on the Focus Mk3 facelift (2014). Volvo's Pilot Assist (2015), launched on the V60, added lane-keeping to stop-and-go ACC and rebranded the combined system as a hands-on driver assistance package. By 2018 every premium German manufacturer had a comparable system, and most of them were marketed as "almost autonomous" — language the regulators were already starting to push back against.
The marketing problem
The technology worked. The marketing didn't.
Tesla's "Autopilot" (2014, on the Model S) was the most aggressive language any manufacturer used. It was, in regulatory terms, an SAE Level 2 system — the driver had to keep their hands on the wheel, monitor the road, and take over instantly when needed. Calling it "Autopilot" implied otherwise. Within five years a series of well-publicised crashes had triggered investigations from the NTSB (US), KBA (Germany), and the UK Department for Transport. Mercedes responded by being explicit that their system was "Drive Pilot." Volvo branded theirs "Pilot Assist," carefully avoiding the word "auto." Ford settled on "BlueCruise."
The technology underneath didn't actually differ much between systems. They were all SAE Level 2 ACC + lane-keeping + a torque sensor on the steering column to check the driver had hands on the wheel. The differences were in calibration: Tesla allowed the system to engage on more roads, Mercedes was more restrictive, Volvo more conservative still. The German consumer organisation ADAC tested all of them in 2019 and found Tesla had the best lateral control, Mercedes the smoothest deceleration, Volvo the most cautious — but the differences were nuance, not capability.
What did differ was the willingness of drivers to trust them. A 2022 IIHS survey found Tesla owners were nearly four times more likely to take their hands off the wheel for extended periods than Mercedes or BMW owners using the same SAE Level 2 capability. The brand was apparently more important than the actual system behaviour.
The Level 3 distinction
Mercedes-Benz Drive Pilot, on the 2022 S-Class W223, became the first SAE Level 3 system approved for use on public roads — first in Germany, then in California and Nevada. The distinction matters: in Level 3, the car is legally responsible for what happens while the system is engaged, not the driver. The driver can legally watch a film or check email. The system is restricted to motorways below 60 km/h (37 mph) — i.e. only useful in heavy traffic — but it's the first time any manufacturer has accepted formal liability for a car driving itself.
Honda Sensing Elite on the Legend (Japan, 2021) was actually first to ship Level 3 in regulatory terms, but only 100 units were ever leased and the system was geofenced to specific Tokyo motorways. It barely counts.
The interesting thing about Level 3 is that the speed restriction is real. Above 60 km/h, all currently-shipping ACC systems are still Level 2 — driver still legally in charge, even if the car is doing all the work. Tesla's "Full Self-Driving" (Beta), in spite of the name, is also Level 2 in the US and the UK. Tesla has never accepted liability for what the system does.
What changed for everyone else
The flow-down effects of ACC have been broader than the systems themselves.
Headway-aware AEB. Once you have a forward radar, you can use it to detect a stopped car ahead even when ACC isn't engaged. Autonomous emergency braking — required in EU since 2024 on all new car types — is essentially the panic-stop branch of the same software stack.
Adaptive headlights. Forward radar plus a camera can detect oncoming traffic before it's in the headlight beam, allowing the matrix-LED system to dim individual segments rather than the whole high-beam. Most premium cars have used this since 2017.
Speed-limit recognition. The same forward camera that's reading lane markings can read speed-limit signs. Some 2024-2026 cars (Audi A4 latest, BMW 5 Series G60) use this to pre-set the cruise speed automatically, defaulting to whatever sign the camera last read.
The 1995 Mitsubishi Diamante had one feature you could buy. The 2026 Mercedes EQS has at least eight, all running on extensions of the same hardware: forward radar, steering torque sensor, brake pressure sensor, forward camera. ACC was the toehold.
The next step that probably won't come
Level 4 — meaning the car actually drives itself, with no driver supervision required, in defined areas — has been "five years away" since 2014. As of 2026 it's still five years away. Waymo runs Level 4 robotaxis in a handful of US cities, but no major manufacturer has shipped a Level 4 capability in a private car.
The reasons are partly technical (edge cases — pedestrians, weather, construction zones — break almost every system) and partly legal (no jurisdiction has yet decided the liability rules for a privately-owned Level 4 car). The likely outcome is that Level 3 stays the high-water mark for private cars, and Level 4 lives only in commercial fleets where the operator can geofence the system, control the maintenance, and accept the liability.
Which means the ACC road-map for the next decade is not "the car drives itself" but "the car you already have gets quietly better." That's probably fine. ACC has already delivered most of what it promised in 1995 — fewer rear-end collisions, less driver fatigue on long motorway runs, almost no down side. It just doesn't make for exciting marketing.