Vehicle sensor calibration system

• US 9,952,317 B2
• Filed: 05/27/2016
• Issued: 04/24/2018
• Est. Priority Date: 05/27/2016
• Status: Active Grant

• Current Assignee: Uatc, LLC (Uber Technologies, Inc.)
• Original Assignee: Uber Technologies, Inc.
• Inventors: Valois, Jean-Sebastien, Bradley, David McAllister, Watson, Adam Charles, Melick, Peter Anthony, Miller, Andrew Gilbert
• Primary Examiner(s): Sample, Jonathan L
• Application Number:
  US15/166,538
• Publication Number:
  US 20170343654A1
• Time in Patent Office: 697 Days
• Field of Search: 701 23- 28, 701 291- 344
• CPC Class Codes: G01S 17/08 for measuring distance only indirect measurement G01S17/46; active triangulation systems G01S17/48
• CPC Class Codes: G01S 17/931 of land vehicles
• CPC Class Codes: G01S 7/4026 Antenna boresight
• CPC Class Codes: G01S 7/497 Means for monitoring or calibrating
• CPC Class Codes: G01S 7/4972 Alignment of sensor

Abstract

A vehicle sensor calibration system can detect an SDV on a turntable surrounded by a plurality of fiducial targets, and rotate the turntable using a control mechanism to provide the sensor system of the SDV with a sensor view of the plurality of fiducial targets. The vehicle sensor calibration system can receive, over a communication link with the SDV, a data log corresponding to the sensor view from the sensor system of the SDV recorded as the SDV rotates on the turntable. Thereafter, the vehicle sensor calibration system can analyze the sensor data to determine a set of calibration parameters to calibrate the sensor system of the SDV.

Claims

1. A vehicle sensor calibration system for self-driving vehicles (SDVs) comprising:

a turntable which an SDV can be driven onto;

a plurality of fiducial targets positioned around the turntable to enable calibration of a sensor system of the SDV;

a control mechanism to automatically rotate the turntable when the SDV is positioned on the turntable; and

one or more computing systems that include one or more processors and one or more memory resources storing instructions that, when executed by the one or more processors, cause the one or more processors to;

receive, over a communication link with the SDV, a data log corresponding to sensor data from the sensor system of the SDV recorded as the SDV rotates on the turntable; and

analyze the sensor data to determine a set of calibration parameters to calibrate the sensor system.

2. The vehicle sensor calibration system of claim 1, wherein the vehicle sensor calibration system is provided in an indoor space, the vehicle sensor calibration system further comprising:

an environment control system to maximize signal-to-noise ratio for the sensor system during calibration, the environment control system to optimize at least lighting conditions and temperature conditions within the indoor space.

3. The vehicle sensor calibration system of claim 1, wherein the executed instructions further cause the one or more processors to:

transmit the set of calibration parameters to the SDV for automatic calibration of the sensor system.

4. The vehicle sensor calibration system of claim 1, wherein the executed instructions cause the one or more processors to analyze the sensor data by running one or more mathematical models on the sensor data, the one or more mathematical models representing a calibrated sensor configuration for the sensor system.

5. The vehicle sensor calibration system of claim 4, wherein the sensor system of the SDV comprises a plurality of LIDAR sensors and a plurality of camera sensors, and wherein the one or more mathematical models include a dedicated mathematical model for each of the plurality of LIDAR sensors and each of the plurality of camera sensors.

6. The vehicle sensor calibration system of claim 4, wherein the one or more mathematical models implement gradient descent on the sensor data to determine the set of calibration parameters for each respective sensor of the sensor system.

7. The vehicle sensor calibration system of claim 1, further comprising:

one or more detectors to detect a position of the SDV on the turntable;

wherein the control mechanism automatically rotates the turntable based on the one or more detectors detecting the position of the SDV on the turntable.

8. A method of calibrating a sensor system of a self-driving vehicle (SDV), the method being performed by one or more processors of a vehicle sensor calibration system and comprising:

detecting an SDV on a turntable;

in response to detecting the SDV on the turntable, rotating the turntable using a control mechanism to provide the sensor system of the SDV with a sensor view of a plurality of fiducial targets, the plurality of fiducial targets being positioned around the turntable at different locations;

receiving, over a communication link with the SDV, a data log corresponding to sensor data from the sensor system of the SDV recorded as the SDV rotates on the turntable; and

analyzing the sensor data to determine a set of calibration parameters to calibrate the sensor system of the SDV.

9. The method of claim 8, wherein the vehicle sensor calibration system is provided in an indoor space, and wherein the vehicle sensor calibration system comprises an environment control system to maximize signal-to-noise ratio for the sensor system during calibration, the environment control system to optimize at least lighting conditions and temperature conditions within the indoor space.

10. The method of claim 8, further comprising:

transmitting the set of calibration parameters to the SDV for automatic calibration of the sensor system.

11. The method of claim 8, the one or more processors analyze the sensor data by running one or more mathematical models on the sensor data, the one or more mathematical models representing a calibrated sensor configuration for the sensor system.

12. The method of claim 11, wherein the sensor system of the SDV comprises a plurality of LIDAR sensors and a plurality of camera sensors, and wherein the one or more mathematical models include a dedicated mathematical model for each of the plurality of LIDAR sensors and each of the plurality of camera sensors.

13. The method of claim 11, wherein the one or more mathematical models implement gradient descent on the sensor data to determine the set of calibration parameters for each respective sensor of the sensor system.

14. The method of claim 10, wherein the vehicle sensor calibration system comprises one or more detectors to detect a position of the SDV on the turntable, and wherein the one or more processors utilize the control mechanism to automatically rotate the turntable based on the one or more detectors detecting the position of the SDV on the turntable.

15. A non-transitory computer readable medium storing instructions that, when executed by one or more processors of a vehicle sensor calibration system, cause the one or more processors to:

detect an SDV on a turntable;

in response to detecting the SDV on the turntable, rotate the turntable using a control mechanism to provide the sensor system of the SDV with a sensor view of a plurality of fiducial targets, the plurality of fiducial targets being positioned around the turntable at different locations;

receive, over a communication link with the SDV, a data log corresponding to sensor data from the sensor system of the SDV recorded as the SDV rotates on the turntable; and

analyze the sensor data to determine a set of calibration parameters to calibrate the sensor system of the SDV.

16. The non-transitory computer readable medium of claim 15, wherein the executed instructions further cause the one or more processors to:

transmit the set of calibration parameters to the SDV for automatic calibration of the sensor system.

17. The non-transitory computer readable medium of claim 15, wherein the executed instructions cause the one or more processors to analyze the sensor data by running one or more mathematical models on the sensor data, the one or more mathematical models representing a calibrated sensor configuration for the sensor system.

18. The non-transitory computer readable medium of claim 17, wherein the sensor system of the SDV comprises a plurality of LIDAR sensors and a plurality of camera sensors, and wherein the one or more mathematical models include a dedicated mathematical model for each of the plurality of LIDAR sensors and each of the plurality of camera sensors.

19. The non-transitory computer readable medium of claim 18, wherein the one or more mathematical models implement gradient descent on the sensor data to determine the set of calibration parameters for each respective sensor of the sensor system.