How well do driving assistance systems work?

Accelerating past 50mph, I pressed a button on the steering wheel of the BMW 3 Series I was driving, let go of the wheel and took my foot off the accelerator. The car took over the driving, keeping itself in its lane and at a constant speed while it travelled along a quarter-mile straight and then into an S-bend. 

Although the car could have slowed itself down to take the corners, it calculated that it didn’t need to and neatly rounded the right and left bends, keeping the car perfectly in the centre of the lane. 

All the time, while the car was doing the driving, a display in the digital instrument panel showed the car in its lane and any other vehicles in real time in front and on both sides of it. A green steering wheel symbol and road markings were also displayed to indicate that the system was operational, and only when the white lines on the road came to an end did these icons change first to yellow (to tell me that I should take back control of the car) and then to red, to tell me that I really did need to take over or the system would slowly bring the car to a halt. 

This was my first trial of BMW’s latest assisted driving system, carried out on a closed test track so that I could legally and safely let the car take over the driving. I was replicating some of the tests carried out by Thatcham Research on behalf of Euro NCAP to evaluate how well these systems work. Their test results are being used to create the first ratings for assisted driving systems. 

The second challenge for the 3 Series was to see how the system reacted to driver intervention. So I activated the Driver Assistant again and drove towards a pothole. As I approached it, I gently guided the car around the hole and back into the centre of the lane. The car let me make the manoeuvre and carried on providing driving assistance afterwards. 

The final test was to see how well the system coped with a stationary vehicle ahead. The 3 Series’ adaptive cruise control was able to decelerate the car enough to avoid a collision at most speeds, and when it couldn’t, the automatic emergency braking system cut in and stopped the car to prevent a crash. 

Next, I tried the same tests in a Tesla Model 3. Its Autopilot driver assistance system took over the accelerator and steering wheel and guided the car neatly around the S-bend, but when I intervened to avoid the pothole, the steering wheel felt like it was fighting against me. And once I’d taken over the steering, it reacted by switching off all the automated driving functions, making me feel like I shouldn’t have interfered with it. 

The Model 3 excelled at avoiding a collision with a stopped car, though, gently bringing itself to a stop a safe distance from it. 

Finally, I tried the Highway and Traffic Jam Companion on the latest Renault Clio. This system kept the Clio in the centre of the lane in the first part of the S-bend but couldn’t slow the car down to an appropriate speed for the second bend, drifting towards the edge of the lane. 

The Clio performed better when it came to avoiding the pothole, letting me temporarily take over the steering to swerve around it and then continuing to operate afterwards.

I didn’t attempt the final test in the Clio, because Euro NCAP had already confirmed that its driver assistance system wouldn’t be able to prevent the car from hitting a stationary vehicle in our test scenario. 

The differences between these three assisted driving systems highlight the potential gulf between the best and worst. There is also a concern that some motorists rely on them to take charge of driving too much, resulting in collisions and, in some cases, deaths. Both of these factors have prompted Euro NCAP to introduce the new assisted driving rating so motorists can see how well the systems in different models perform. 

Next: Results of the first official test of driver assistance systems >>

Page 1 of 2