Multi-sensor collaborative calibration system

1. A method of autonomous vehicle operation, comprising:

obtaining, from a first sensor and a second sensor located on a vehicle, sensor data items of an area that comprises a plurality of calibration objects;

determining, from the sensor data items for each of the first sensor and the second sensor, attributes of the plurality of calibration objects;

determining, for the first sensor and the second sensor, an initial matrix that describes a first set of extrinsic parameters between the first sensor and the second sensor based at least on the attributes of the plurality of calibration objects;

determining an updated matrix that describes a second set of extrinsic parameters between the first sensor and the second sensor based at least on the initial matrix and a location of at least one calibration object, wherein the updated matrix is determined by using a factor graph that minimizes a summation of at least a first factor associated with the first sensor and a second factor associated with the second sensor, and

wherein each the first factor and the second factor is based at least on:

(1) the initial matrix,

(2) coordinates of at least one calibration object obtained from the sensor data items for the first sensor and the second sensor,

(3) a third set of extrinsic parameters that describe a location or a pose of the second sensor at a first time and another location or another pose of the second sensor at a second time that is later in time than the first time, and

(4) the location of the at least one calibration object; and

performing autonomous operation of the vehicle using the second set of extrinsic parameters and additional sensor data received from the first sensor and the second sensor when the vehicle is operated on a road.

2. The method of claim 1 , wherein the first sensor includes a camera, wherein the second sensor is a light detection and ranging (LiDAR) sensor, and wherein the initial matrix is determined by performing a pairwise calibration operation using:

(1) a first set of coordinates of the plurality of calibration objects determined from a first sensor data item for the camera,

(2) a second set of coordinates of the plurality of calibration objects determined from a second sensor data item for the LiDAR sensor, and

(3) a set of intrinsic parameters that describe optical characteristics of the camera, and

wherein the first set of coordinates and the second set of coordinates are the attributes of the plurality of calibration objects.

3. The method of claim 2 , further comprising:

obtaining, from a second camera located on the vehicle, a third sensor data item of the area that comprises the plurality of calibration objects; and

determining, a second initial matrix by performing another pairwise calibration operation between the LiDAR sensor and the second camera using:

(1) the second set of coordinates of the plurality of calibration objects determined from the second sensor data item for the LiDAR sensor,

(2) a third set of coordinates of the plurality of calibration objects determined from the third sensor data item for the second camera, and

(3) another set of intrinsic parameters that describe optical characteristics of the second camera.

4. The method of claim 3 , further comprising:

determining another updated matrix that describes another set of extrinsic parameters between the LiDAR sensor and the second camera based at least on the second initial matrix and the location of at least one calibration object.

5. The method of claim 3 , wherein the camera faces a front direction to capture a set of images of a region towards which the vehicle is driven, and wherein the second camera faces a side direction to capture another set of images of another region towards a side of the vehicle.

6. The method of claim 1 , wherein the first sensor includes a first light detection and ranging (LiDAR) sensor, wherein the second sensor includes a second LiDAR sensor, and wherein the initial matrix is determined by performing a pairwise calibration operation using:

(1) a first set of coordinates of the plurality of calibration objects determined from a first sensor data item for the first LiDAR, and

(2) a second set of coordinates of the plurality of calibration objects determined from a second sensor data item for the second LiDAR sensor, and

wherein the first set of coordinates and the second set of coordinates are the attributes of the plurality of calibration objects.

7. A system for calibration of multiple sensors on or in a vehicle, the system comprising a processor configured to:

obtain, from a first sensor and a second sensor located on a vehicle, sensor data items of an area that comprises a plurality of calibration objects;

determine, from the sensor data items for each of the first sensor and the second sensor, attributes of the plurality of calibration objects;

determine, for the first sensor and the second sensor, an initial matrix that describes a first set of extrinsic parameters between the first sensor and the second sensor based at least on the attributes of the plurality of calibration objects;

determine an updated matrix that describes a second set of extrinsic parameters between the first sensor and the second sensor based at least on the initial matrix and a location of at least one calibration object, wherein the updated matrix is determined by using a factor graph that minimizes a summation of at least a first factor associated with the first sensor and a second factor associated with the second sensor, and

wherein each the first factor and the second factor is based at least on:

(1) the initial matrix,

(2) coordinates of at least one calibration object obtained from the sensor data items for the first sensor and the second sensor,

(3) a third set of extrinsic parameters that describe a location or a pose of the second sensor at a first time and another location or another pose of the second sensor at a second time that is later in time than the first time, and

(4) the location of the at least one calibration object; and

perform autonomous operation of the vehicle using the second set of extrinsic parameters and additional sensor data received from the first sensor and the second sensor when the vehicle is operated on a road.

8. The system of claim 7 , wherein the location of the at least one calibration object includes a center point of the at least one calibration object.

9. The system of claim 7 , wherein the sensor data items are obtained by each of the first sensor and the second sensor at a same time.

10. The system of claim 7 , wherein the plurality of calibration objects and the vehicle are located in a building or a warehouse.

11. The system of claim 7 , wherein the plurality of calibration objects are located at one or more distances relative to a location of the vehicle.

12. The system of claim 7 , wherein the plurality of calibration objects are located around the vehicle.

13. A non-transitory computer readable storage medium having code stored thereon, the code, when executed by a processor, causing the processor to implement a method comprising:

obtaining, from a first sensor and a second sensor located on a vehicle, sensor data items of an area that comprises a plurality of calibration objects;

determining, from the sensor data items for each of the first sensor and the second sensor, attributes of the plurality of calibration objects;

determining, for the first sensor and the second sensor, an initial matrix that describes a first set of extrinsic parameters between the first sensor and the second sensor based at least on the attributes of the plurality of calibration objects;

determining an updated matrix that describes a second set of extrinsic parameters between the first sensor and the second sensor based at least on the initial matrix and a location of at least one calibration object, wherein the updated matrix is determined by using a factor graph that minimizes a summation of at least a first factor associated with the first sensor and a second factor associated with the second sensor, and

wherein each the first factor and the second factor is based at least on:

(1) the initial matrix,

(2) coordinates of at least one calibration object obtained from the sensor data items for the first sensor and the second sensor,

(3) a third set of extrinsic parameters that describe a location or a pose of the second sensor at a first time and another location or another pose of the second sensor at a second time that is later in time than the first time, and

(4) the location of the at least one calibration object; and

performing autonomous operation of the vehicle using the second set of extrinsic parameters and additional sensor data received from the first sensor and the second sensor when the vehicle is operated on a road.

14. The non-transitory computer readable storage medium of claim 13 , wherein the performing autonomous operation of the vehicle comprises:

determining a first object located on the road from the additional sensor data obtained from the first sensor;

determining a second object located on the road from the additional sensor data obtained from the second sensor; and

determining that the first object is same as the second object using the second set of extrinsic parameters.

15. The non-transitory computer readable storage medium of claim 13 , wherein the first sensor includes a Radar, wherein the second sensor includes a light detection and ranging (LiDAR) sensor, and wherein the initial matrix is determined by performing a pairwise calibration operation using at least:

(1) a radius and an angle, for each calibration object, from the Radar to a point on a plane of a calibration object determined from a first sensor data item for the Radar,

(2) a set of coordinates of the plurality of calibration objects determined from a second sensor data item for the LiDAR sensor, and

wherein the radius and angle for each calibration object and the set of coordinates are the attributes of the plurality of calibration objects.

16. The non-transitory computer readable storage medium of claim 15 , wherein the initial matrix is determined by applying a transformation from polar coordinates to cartesian coordinates using the radius and the angle.

17. The non-transitory computer readable storage medium of claim 15 , wherein the method further comprises:

obtaining, from a second Radar located on the vehicle, a third sensor data item of the area that comprises the plurality of calibration objects;

determining, a second initial matrix by performing another pairwise calibration operation between the LiDAR sensor and the second camera using:

(1) the set of coordinates of the plurality of calibration objects determined from the second sensor data item for the LiDAR sensor, and

(2) a second radius and a second angle, for each calibration object, from the second Radar to the point on the plane of a calibration object determined from a third sensor data item for the second Radar.

18. The non-transitory computer readable storage medium of claim 17 , wherein the method further comprises:

determining another updated matrix that describes another set of extrinsic parameters between the LiDAR sensor and the second Radar based at least on the second initial matrix and the location of at least one calibration object.

19. The non-transitory computer readable storage medium of claim 13 , wherein the updated matrix is determined by using the factor graph that minimizes the summation of at least the first factor, the second factor, and a third matrix that represents a position and an orientation of the vehicle.