How Car Safety Systems Turn Off When Sensors Disagree

This patent describes a car safety system that uses both a camera and a radar sensor to detect objects, and if their detections don't match up, it can disable parts of the driving assistance for safety.

What it covers

This system helps a vehicle stay safe by using two different types of sensors: a camera and a radar sensor, both looking forward. An electronic control unit (ECU) processes the information from both sensors to detect objects, like other cars or pedestrians, in front of the vehicle. The clever part is that the ECU then compares what the camera sees with what the radar sees. If there's a big difference, meaning one sensor detects an object and the other doesn't (as described in claims 8 and 10), or if the sensors are misaligned (claim 12), the ECU determines there's an "error in object detection." If this error is too large, the system will disable at least part of the driving assist function, such as automatic emergency braking or forward collision warning (claims 2 and 3), and can alert the driver (claim 5).

What it doesn't cover

• —Does not cover driving assist systems that rely on only a single type of sensor, such as only a camera or only a radar.
• —Does not cover systems that use sensor disagreement primarily to correct or fuse data, rather than to disable a driving assist function.
• —Does not cover systems that use different combinations of sensors, like lidar and cameras, without a radar sensor.
• —Does not cover systems where the 'error in object detection' is determined by internal sensor diagnostics rather than a direct comparison of object detections between a camera and a radar.
• —Does not cover systems that continue to provide full driving assist functionality even when there is a significant discrepancy between camera and radar object detections.

The clever bit

The truly clever part is using the *discrepancy* between two different types of object detection – visual from a camera and range-based from radar – as a direct trigger to *disable* or reduce the functionality of a driving assist system, rather than just trying to reconcile or fuse the conflicting data. This prioritizes safety by defaulting to a more cautious state when sensor reliability is in doubt.

Why it matters

Modern vehicles increasingly rely on advanced driver-assistance systems (ADAS) for safety, but these systems need to be reliable. This patent addresses a critical safety concern: what happens when sensors provide conflicting information. By disabling functions when there's a significant disagreement between a camera and a radar, the system aims to prevent potentially dangerous actions based on unreliable data, enhancing overall vehicle safety.

Real-world examples

• 1.Automatic Emergency Braking systems in cars
• 2.Forward Collision Warning systems
• 3.Pedestrian detection systems in vehicles
• 4.Adaptive Cruise Control systems that rely on multi-sensor input