LIDAR-to-camera transformation during sensor calibration for autonomous vehicles

1. A method, comprising:

performing one or more control operations of a first vehicle based at least on a calibration performed with respect to LIDAR data and camera data, the calibration being based at least on calibration operations performed with respect to a second vehicle, the calibration operations including:

fitting a plane through LIDAR coordinates corresponding to a board as detected in a LIDAR scan obtained using a LIDAR sensor of the second vehicle, the fitting of the plane including performing a least squares error determination with respect to a plurality of LIDAR points corresponding to the LIDAR scan in which two or more of the LIDAR points are not coplanar with respect to each other;

projecting the plane from the LIDAR coordinates to a first set of camera coordinates referenced according to a camera of the second vehicle;

detecting the board in a camera image obtained using the camera, the board being detected in the camera image at a second set of camera coordinates referenced according to the camera; and

calibrating the LIDAR sensor of the second vehicle and the camera of the second vehicle by at least determining a transform between the first set of camera coordinates and the second set of camera coordinates.

2. The method of claim 1 , wherein the board has a checkerboard pattern with two white blocks each having a black dot therein or with two black blocks each having a white dot therein.

3. The method of claim 2 , wherein positions of all blocks of the checkerboard pattern are identified based at least on identification of the two white blocks or the two black blocks including in instances in which the checkerboard pattern is partially out of view.

4. A system comprising:

one or more processors to cause the system to perform operations comprising:

performing one or more control operations of a first machine based at least on a calibration performed with respect to LIDAR data and camera data, the calibration being based at least on calibration operations performed with respect to a second machine, the calibration operations including:

fitting a plane through LIDAR coordinates corresponding to a board as detected in a LIDAR scan obtained using a LIDAR sensor of the second machine, the fitting of the plane including using an optimization technique based at least on a loss function corresponding to a distance between the plane and individual points in respective directions that connect the respective points with a center of the LIDAR sensor;

identifying a first set of camera coordinates corresponding to LIDAR coordinates that correspond to a first location of a board in a LIDAR scan obtained using a LIDAR sensor of the second machine, the first set of camera coordinates being referenced according to a camera of the second machine and being identified based at least on a projection of the plane to the first set of camera coordinates;

identifying a second set of camera coordinates referenced according to the camera and corresponding to a second location of the board in a camera image obtained using the camera of the second machine; and

calibrating the LIDAR sensor and the camera by at least determining a transform between the first set of camera coordinates and the second set of camera coordinates.

5. The system of claim 4 , wherein the board has a checkerboard pattern with two white blocks each having a black dot therein or with two black blocks each have a white dot therein.

6. The system of claim 5 , wherein positions of all blocks of the checkerboard pattern are identified based at least on identification of the two white blocks or the two black blocks including in instances in which the checkerboard pattern is partially out of view.

7. A processor comprising:

processing circuitry to cause a system to perform operations comprising:

performing one or more control operations of a first machine based at least on a calibration performed with respect to LIDAR data and camera data, the calibration being based at least on calibration operations performed with respect to a second machine, the calibration operations including:

fitting a plane through LIDAR coordinates corresponding to a board as detected in a LIDAR scan obtained using a LIDAR sensor of the second machine, the fitting of the plane including using an optimization technique based at least on one or more of:

using an optimization technique based at least on a loss function corresponding to a distance between the plane and individual points in respective directions that connect the respective points with a center of the LIDAR sensor; or

performing a least squares error determination with respect to a plurality of LIDAR points corresponding to the LIDAR scan in which two or more of the LIDAR points are not coplanar with respect to each other;

identifying a first set of camera coordinates corresponding to LIDAR coordinates that correspond to a first location of a board in a LIDAR scan obtained using a LIDAR sensor of a machine, the first set of camera coordinates being referenced according to a camera of the second machine;

identifying a second set of camera coordinates referenced according to the camera and corresponding to a second location of the board in a camera image obtained using the camera of the second machine; and

calibrating the LIDAR sensor and the camera by at least determining a transform between the first set of camera coordinates and the second set of camera coordinates; and

directing that the calibration operations be repeated one or more times for different poses of the second machine with respect to a position of the board based at least on a determined accuracy corresponding to the calibrating.

8. The processor of claim 7 , wherein the board has:

a relatively dark and unreflective background.

9. The processor of claim 7 , wherein the board has a checkerboard pattern with two white blocks each having a black dot therein or with two black blocks each have a white dot therein.

10. The processor of claim 9 , wherein positions of all blocks of the checkerboard pattern are identified based at least on identification of the two white blocks or the two black blocks including in instances in which the checkerboard pattern is partially out of view.

11. The method of claim 1 , wherein the calibration operations are performed with respect to one or more different poses of the second vehicle with respect to a position of the board and wherein the one or more different poses are changed based at least on a target position of the board in relation to the second vehicle.

12. The method of claim 11 , wherein at least one of the one or more different poses of the second vehicle with respect to the position of the board is changed according to movement instructions for one or more of the board or the second vehicle, the movement instructions being determined based at least on a target position of the board in relation to the second vehicle.

13. The method of claim 1 , wherein one or more of an audio indication or a visual indication is provided to indicate that a target pose of the second vehicle with respect to a target position of the board has been achieved.

14. The system of claim 4 , wherein the calibration operations are performed with respect to one or more different poses of the second machine with respect to a position of the board and wherein the one or more different poses are changed based at least on movement instructions for one or more of the board or the second machine.

15. The system of claim 14 , wherein at least one of the one or more different poses of the second machine with respect to the position of the board is changed according to movement instructions for one or more of the board or the second machine, the movement instructions being determined based at least on a target position of the board in relation to the second machine.

16. The system of claim 4 , wherein one or more of an audio indication or a visual indication is provided to indicate that a target pose of the second machine with respect to a target position of the board has been achieved.

17. The processor of claim 7 , wherein the calibration operations are performed with respect to one or more different poses of the second machine with respect to a position of the board and wherein the one or more different poses are changed based at least on movement instructions for one or more of the board or the second machine that are determined based at least on a target position of the board in relation to the second machine.

18. The processor of claim 7 , wherein one or more of an audio indication or a visual indication is provided to indicate that a target pose of the second machine with respect to a target position of the board has been achieved.

19. The method of claim 1 , further comprising repeating the calibration operations at one or more different poses of the second vehicle with respect to a position of the board until an estimated error with respect to the calibration is within a threshold tolerance.

20. The method of claim 1 , wherein the first vehicle and the second vehicle are a same vehicle.