Researchers from Duke University in North Carolina have developed a technique that targets a vehicle’s radar system. The method enables manipulation of the radio signals that the car’s sensors receive. The work was published on a platform hosting scientific materials for open access and collaboration.
The technology is called MadRadar. The creators say that with this tool it is possible for a car’s radar to perceive ghost vehicles in seconds or, conversely, for real approaching cars to be hidden from the vehicle’s radar sensors. Experts describe the invention as a serious risk to the radar security of modern vehicles.
Radar plays a key role in many modern vehicles, especially those equipped with assistive driving features and autonomous driving capabilities. It helps detect vehicles nearby and in front of the car, and it also complements other sensing modalities such as cameras and lidar to identify objects ahead or behind the vehicle.
In the MadRadar study, the team demonstrated a radar spoofing system capable of quickly learning a target vehicle’s radar parameters in under a quarter of a second. Once these parameters are identified, the system emits signals that mislead the target’s radar into producing false readings.
In one demonstration, MadRadar sent signals to a target car that caused it to believe another car existed when it did not. This effect is achieved by altering the signal’s characteristics over time and speed to mimic a plausible real contact on the road.
A second, more intricate example showed the device making the target radar think that a vehicle was present in a location where there is none. This is accomplished by precisely injecting masking signals around the real location, creating a bright spot that confuses the radar’s interpretation of the scene.
A third scenario combines the two approaches, making it appear as though a nearby vehicle is changing direction suddenly, even when it is not.
The researchers suggest that car makers introduce randomization in radar operating parameters over time and strengthen processing algorithms to detect such spoofing attempts. These defensive measures would help improve resilience against spoofing and safeguard sensor fusion systems that depend on radar data.
Previous industry incidents include recalls related to autonomous driving concerns, underscoring the importance of ongoing attention to sensor security and robust validation of driver-assistance technologies.