How Car Factories Calibrate LIDAR Sensors for Self-Driving Cars

This patent describes a system for car factories to precisely adjust a vehicle's LIDAR sensors using special floor and wall targets, ensuring they accurately map the world for features like self-driving.

What it covers

This calibration system helps car manufacturers accurately set up LIDAR sensors in new vehicles. It uses a horizontal target on the ground and a vertical target, both with distinct patterns of markers, placed in a special 'end of line calibration region' at the factory (Claim 1). When a vehicle is in this spot, its LIDAR sensor detects these markers (Claim 1). By processing the data from the LIDAR, the system figures out the exact positions of these markers relative to the vehicle (Claim 1). Based on these determined locations, the system then calculates any misalignment in the LIDAR sensor's 'roll,' 'pitch,' and 'yaw' (Claim 1). Finally, it adjusts the LIDAR system to correct these misalignments (Claim 1). For example, if a LIDAR sensor is slightly tilted down (pitch) or rotated to the side (yaw), this system can detect that and correct it before the car leaves the factory.

What it doesn't cover

• —Does not cover calibration systems that only use a single target (e.g., only horizontal or only vertical), as it requires both a horizontal and a vertical target (Claim 1).
• —Does not cover calibration performed outside of a vehicle assembly facility's 'end of line calibration region' (Claim 1).
• —Does not cover systems that calibrate LIDAR sensors without determining misalignment across 'roll,' 'pitch,' and 'yaw' axes (Claim 1).
• —Does not cover calibration methods that do not rely on light reflecting targets arranged in predetermined patterns (Claim 1).
• —Does not cover calibration for sensors other than LIDAR, such as cameras or radar, or for LIDAR sensors that do not sense forward of the vehicle (Claim 2).

The clever bit

The novelty lies in using both horizontal and vertical targets with distinct patterns in a factory setting to precisely determine and correct misalignment across all three rotational axes (roll, pitch, and yaw) of a LIDAR sensor, ensuring comprehensive calibration.

Why it matters

Accurate LIDAR sensor calibration is critical for the safety and performance of advanced driver-assistance systems (ADAS) and autonomous vehicles. If a LIDAR sensor is misaligned, it could misinterpret the distance or position of objects, leading to incorrect decisions by the vehicle's computer. This system ensures that LIDAR sensors are precisely aligned before vehicles are delivered to customers, which is essential for reliable operation of features like adaptive cruise control, automatic emergency braking, and self-parking.

Real-world examples

• 1.Vehicle assembly lines for autonomous cars
• 2.Quality control stations for ADAS-equipped vehicles
• 3.Automotive manufacturing plants producing LIDAR-enabled vehicles