Beyond the Proving Ground: Independent Public-Road Testing of Assisted Lane Change Systems using LiDAR

Imagine driving on a highway, and you decide to change lanes. You signal, check your mirrors, and when the coast is clear, you steer over. For decades, this has been a human decision, a split-second judgment of speed and distance. Today, however, many new cars can do this for you. These systems, known as assisted lane change, use sensors to watch the road, calculate if a gap is safe, and then move the car sideways without the driver touching the wheel. But how do we know these machines are making the right call? Currently, car manufacturers test these systems in closed, controlled environments like proving grounds, running the same scenarios over and over. While this ensures the tests are repeatable, it often misses the messy, unpredictable reality of actual traffic. Furthermore, many of these systems are "geo-fenced," meaning they only work on specific highways, and manufacturers often control access to the test data, making it hard for independent experts to verify safety on their own.

A team of researchers from the European Commission's Joint Research Centre decided to take this testing out of the closed course and onto the open road. They wanted to see if commercial cars, which have already been officially approved for sale, would actually follow the strict safety rules designed to prevent accidents during a lane change. Specifically, they looked at a rule that says a car must not change lanes if another vehicle is coming up too fast and too close in the next lane. To do this, they drove on the A31 motorway in France, between Dijon and Nancy, using a special setup that allowed them to measure distances with high precision without needing the car manufacturer's help or relying on perfect satellite signals, which can be unreliable in tunnels or rural areas.

The researchers used a test vehicle equipped with a laser scanner, a device that creates a detailed 3D map of the surroundings by bouncing light off objects. This scanner was mounted on a second car that acted as the "approaching vehicle," driving faster in the lane next to the test car. A third car followed behind to keep the test car from being cut off by other traffic. The team set up a series of tests where they triggered the lane-change system at different speeds and distances. They wanted to see if the system would correctly stop the lane change when the approaching car was too close, and if it would proceed when there was enough space. They ran 27 of these specific maneuvers, covering a range of speeds from 100 to 130 kilometers per hour and distances between 20 and 60 meters.

The results revealed that the system was not always perfect. Out of the 27 attempts, the car successfully completed 18 lane changes and correctly stopped 9 of them. However, in six of the completed maneuvers, the car changed lanes even though the approaching vehicle was closer than the safety rules allowed. When the researchers analyzed these six cases more closely, accounting for the tiny margin of error in their laser measurements, they found that in three of them, the car had definitely violated the minimum safety distance. In these instances, the approaching car was so close that, according to the official safety regulations, the lane change should have been blocked. Interestingly, these violations happened when the speed difference between the two cars was small, a specific situation that the standard safety tests used by manufacturers do not always cover.

This study demonstrates that it is possible to independently verify the safety of advanced driving systems on public roads using laser technology, bypassing the need for manufacturer cooperation or perfect satellite positioning. The findings suggest that while these systems work well in many situations, they may still miss critical safety limits in specific real-world conditions that are not fully captured by current testing methods. The researchers conclude that their approach provides a viable way for authorities to monitor the safety of these vehicles as they are used every day, ensuring that the technology remains safe even when the road conditions are more complex than a controlled test track.