High precision multi-sensor extrinsic calibration via production line and mobile station

1. A method for providing multi-sensor extrinsic calibration in a vehicle, the method comprising:

providing the vehicle within an extrinsic calibration station having a two-dimensional (2D) calibration target;

receiving, by one or more processors of the vehicle, sensor data from a first sensor and a second sensor of the vehicle, the first sensor being a LIDAR sensor and the second sensor being an object-detection sensor or an imaging sensor, the sensor data comprising:

a three-dimensional (3D) point cloud including a first representation of the 2D calibration target as sensed by the first sensor, the first representation corresponding to a partial representation of the 2D calibration target sensed by the first sensor; and

other image data including a second representation of the 2D calibration target as sensed by the second sensor;

clustering and segmenting, by the one or more processors, detection points in the 3D point cloud to detect a shape of the 2D calibration target from the partial representation of the 2D calibration target;

determining, by the one or more processors, a center of the 2D calibration target using the detected shape of the 2D calibration target;

determining, by the one or more processors, a center of the 2D calibration target in the second representation based on one or more features of the 2D calibration target in the second representation;

matching, by the one or more processors, the center of the 2D calibration target as determined using the detected shape to the center of the 2D calibration target in the second representation;

computing, by the one or more processors, a six-degree of freedom rigid body transformation of the first sensor and the second sensor based on the matched centers of the 2D calibration target as determined using the detected shape and of the 2D calibration target in the second representation; and

computing, by the one or more processors, a projection of the first sensor to the second sensor based on the computed six-degree of freedom rigid body transformation of the first sensor and the second sensor.

2. The method of claim 1 , wherein the 2D calibration target is placed within a field of view of the first sensor and within a field of view of the second sensor.

3. The method of claim 1 , wherein the centers of the 2D calibration target as determined using the detected shape and of the 2D calibration target in the second representation are matched without using intensity information from the LIDAR sensor and without using GPS information.

4. The method of claim 1 , wherein providing the vehicle within the extrinsic calibration station with one or more calibration targets includes securing the vehicle to be stationary, securing the one or more calibration targets to a rack, and rotating the rack with the one or more calibration targets into one or more predetermined locations around the vehicle such that the one or more calibration targets are provided around an external environment within a threshold distance of the vehicle.

5. A calibration station for multi-sensor extrinsic calibration in a vehicle, comprising:

a two-dimensional (2D) calibration target provided around an external environment within a threshold distance of the vehicle; and

a vehicle placement section configured to accommodate the vehicle on the vehicle placement section,

wherein one or more processors of the vehicle are configured to:

receive sensor data from a first sensor and a second sensor of the vehicle, the first sensor being a LIDAR sensor and the second sensor being an object-detection sensor or an imaging sensor, the sensor data comprising:

a three-dimensional (3D) point cloud including a first representation of the 2D calibration target as sensed by the first sensor, the first representation corresponding to a partial representation of the 2D calibration target sensed by the first sensor; and

other image data including a second representation of the 2D calibration target as sensed by the second sensor;

cluster and segment detection points in the 3D point cloud to detect a shape of the 2D calibration target from the partial representation of the 2D calibration target;

determine a center of the 2D calibration target using the detected shape of the 2D calibration target;

determine a center of the 2D calibration target in the second representation based on one or more features of the 2D calibration target in the second representation;

match the center of the 2D calibration target as determined using the detected shape to the center of the 2D calibration target in the second representation;

compute a six-degree of freedom rigid body transformation of the first sensor and the second sensor based on the matched centers of the 2D calibration target as determined using the detected shape and of the 2D calibration target in the second representation; and

compute a projection of the first sensor to the second sensor based on the computed six-degree of freedom rigid body transformation of the first sensor and the second sensor.

6. The calibration station of claim 5 , wherein the vehicle placement section is configured to secure the vehicle to be stationary within the calibration station and one or more calibration targets to a rack, and is configured to rotate the rack with the one or more calibration targets into one or more predetermined locations around the vehicle.

7. The calibration station of claim 5 , wherein the 2D calibration target is placed within a field of view of the first sensor and within a field of view of the second sensor.

8. The calibration station of claim 5 , wherein the centers of the 2D calibration target as determined using the detected shape and of the 2D calibration target in the second representation are matched without using intensity information from the LIDAR sensor and without using GPS information.

9. A calibration system for multi-sensor extrinsic calibration in a vehicle, comprising:

a two-dimensional (2D) calibration target provided around an external environment within a threshold distance of the vehicle;

a vehicle placement section configured to accommodate the vehicle for detection of the 2D calibration target; and

one or more processors associated with the vehicle, wherein the one or more processors are configured to:

receive sensor data from a first sensor and a second sensor of the vehicle, the first sensor being a LIDAR sensor and the second sensor being an object-detection sensor or an imaging sensor, the sensor data comprising:

a three-dimensional (3D) point cloud including a first representation of the 2D calibration target as sensed by the first sensor, the first representation corresponding to a partial representation of the 2D calibration target sensed by the first sensor; and

other image data including a second representation of the 2D calibration target as sensed by the second sensor;

cluster and segment detection points in the 3D point cloud to detect a shape of the 2D calibration target from the partial representation of the 2D calibration target;

determine a center the 2D calibration target using the detected shape of the 2D calibration target;

determine a center of the 2D calibration target in the second representation based on one or more features of the 2D calibration target in the second representation;

match the center of the 2D calibration target as determined using the detected shape to the center of the 2D calibration target in the second representation;

compute a six-degree of freedom rigid body transformation of the first sensor and the second sensor based on the matched centers of the 2D calibration target as determined using the detected shape and of the 2D calibration target in the second representation; and

compute a projection of the first sensor to the second sensor based on the computed six-degree of freedom rigid body transformation of the first sensor and the second sensor,

wherein the 2D calibration target includes a combination of sensor targets.

10. The calibration system of claim 9 , wherein the 2D calibration target is placed within a field of view of the first sensor and within a field of view of the second sensor.

11. The calibration system of claim 9 , wherein the vehicle placement section is configured to secure the vehicle to be stationary and secure one or more calibration targets to a rack, and is configured to rotate the rack with the one or more calibration targets into one or more predetermined locations around the vehicle.

12. The calibration system of claim 9 , wherein the centers of the 2D calibration target as determined using the detected shape and of the 2D calibration target in the second representation are matched without using intensity information from the LIDAR sensor and without using GPS information.

13. The method of claim 1 , wherein the shape of the 2D calibration target is symmetrical.

14. The calibration station of claim 5 , wherein the shape of the 2D calibration target is symmetrical.

15. The calibration system of claim 9 , wherein the shape of the 2D calibration target is symmetrical.

16. The method of claim 1 , wherein the second sensor is a RADAR sensor or a camera.

17. The calibration station of claim 5 , wherein the second sensor is a RADAR sensor or a camera.

18. The calibration system of claim 9 , wherein the second sensor is a RADAR sensor or a camera.

19. The method of claim 1 , wherein the six-degree of freedom rigid body transformation of the first sensor and the second sensor is computed using singular value decomposition.

20. The calibration station of claim 5 , wherein the six-degree of freedom rigid body transformation of the first sensor and the second sensor is computed using singular value decomposition.