Automated real-time calibration
Provided are systems and methods for detecting a vehicle with sensors that are not calibrated properly and calibrating such sensor in real-time. In one example, a method may include iteratively capturing sensor data of a road while the vehicle is travelling on the road; monitoring a calibration of the sensors of the vehicle based on the sensor data, determining that the sensors of the vehicle are not calibrated properly based on the monitoring, generating a calibration target of an object on the road based on the sensor data, and adjusting a calibration parameter of the one or more sensors of the vehicle based on the generated calibration target.
1. A vehicle comprising:
a processor configured to
iteratively receive sensor data associated with a surrounding environment while the vehicle is travelling through the surrounding environment
monitor a calibration of the one or more sensors of the vehicle based on the iteratively received sensor data;
determine that the one or more sensors of the vehicle are not calibrated properly based on the monitoring;
generate a calibration target of an object in the surrounding environment based on the iteratively received sensor data; and
adjust a calibration parameter of the one or more sensors of the vehicle based on the generated calibration target.
2. The vehicle of claim 1 , wherein the processor is configured to receive multiple measurements of a ground plane around the vehicle at multiple points in time, respectively, based on lidar points captured by the one or more sensors and a machine learning model.
3. The vehicle of claim 2 , wherein the processor is configured to determine that a lidar sensor is not properly calibrated based on a comparison of the multiple measurements of the ground plane and a ground plane cost minimization function.
4. The vehicle of claim 1 , wherein the vehicle is travelling on a road within the surrounding environment.
5. The vehicle of claim 4 , wherein the processor is configured to generate multiple measurements of a lane line on the road at multiple points in time, respectively, based on lidar points captured by the one or more sensors and a machine learning model.
6. The vehicle of claim 5 , wherein the processor is configured to determine that a lidar sensor is not properly calibrated based on a comparison of the multiple measurements of the lane line and a line-to-line cost minimization function.
7. The vehicle of claim 1 , wherein the processor is configured to project lidar points labeled as lane lines points or road points, captured by a lidar sensor, into one or more images captured by a camera of the vehicle.
8. The vehicle of claim 7 , wherein the processor is configured to calibrate the camera of the vehicle based on a difference in alignment between the labeled lidar points and corresponding image points of the one or more images captured by the camera and based on one or more cost functions.
9. The vehicle of claim 1 , wherein the processor is configured to dynamically generate the calibration target from lane line data captured of a road and adjust one or more of a lidar calibration parameter and a camera calibration parameter based on the dynamically generated calibration target.
10. The vehicle of claim 1 , wherein the processor is further configured to continue to monitor the one or more sensors, determine that the one or more sensors of the vehicle are still not calibrated properly after the adjusting based on the continued monitoring, and in response, disable the vehicle.
11. A method comprising:
iteratively receiving sensor data associated with a surrounding environment while the vehicle is travelling through the surrounding environment
monitoring a calibration of the one or more sensors of the vehicle based on the iteratively received sensor data;
determining that the one or more sensors of the vehicle are not calibrated properly based on the monitoring;
generating a calibration target of an object in the surrounding environment based on the iteratively received sensor data; and
adjusting a calibration parameter of the one or more sensors of the vehicle based on the generated calibration target.
12. The method of claim 11 , wherein the iteratively receiving comprises receiving multiple measurements of a ground plane around the vehicle at multiple points in time, respectively, based on lidar points captured by the one or more sensors and a machine learning model.
13. The method of claim 12 , wherein the determining comprises determining that a lidar sensor is not properly calibrated based on a comparison of the multiple measurements of the ground plane and a ground plane cost minimization function.
14. The method of claim 11 , wherein the vehicle is travelling on a road within the surrounding environment.
15. The method of claim 14 , wherein the receiving comprises receiving multiple measurements of a lane line on the road at multiple points in time, respectively, based on lidar points captured by the one or more sensors and a machine learning model, and the determining comprises determining that a lidar sensor is not properly calibrated based on a comparison of the multiple measurements of the lane line and a line-to-line cost minimization function.
16. The method of claim 11 , wherein the generating comprises projecting lidar points labeled as lane lines points or road points, captured by a lidar sensor, into one or more images captured by a camera of the vehicle.
17. The method of claim 16 , wherein the adjusting comprises calibrating the camera of the vehicle based on a difference in alignment between the labeled lidar points and corresponding image points of the one or more images captured by the camera and based on one or more cost functions.
18. The method of claim 11 , wherein the generating comprises dynamically generating the calibration target from lane line data captured of a road and adjust one or more of a lidar calibration parameter and a camera calibration parameter based on the dynamically generated calibration target.
19. A computing system comprising:
a memory configured to store sensor data associated with a road captured by one or more sensors of a vehicle in motion on the road; and
a processor configured to
monitor a calibration of the one or more sensors of the vehicle based on the received sensor data;
determine that the one or more sensors are not calibrated properly based on the monitoring;
generate a calibration target of an object on the road based on the received sensor data; and
adjust a calibration parameter of the one or more sensors of the vehicle based on the generated calibration target.
20. The computing system of claim 19 , wherein the processor is configured to dynamically generate the calibration target from lane line data captured of the road and adjust one or more of a lidar calibration parameter and a camera calibration parameter based on the dynamically generated calibration target.