Method and apparatus for predicting sensor error

Patent metadata

Field	Value
Publication number	US-11651244-B2
Application number	US-201816152092-A
Country	US
Kind code	B2
Filing date	Oct 4, 2018
Priority date	Oct 4, 2018
Publication date	May 16, 2023
Grant date	May 16, 2023

Abstract

An approach is provided for providing predictive classification of sensor error. The approach involves, for example, receiving sensor data from at least one sensor, the sensor data collected at a geographic location. The approach also involves extracting a set of input features from the sensor data, map data representing the geographic location, or combination thereof. The approach further involves processing the set of input features using a machine learning model to calculate a predicted sensor error of a target sensor operating at the geographic location. The machine learning model, for instance, has been trained on ground truth sensor error data to use the set of input features to calculate the predicted sensor error.

First claim

What is claimed is:

1. A computer-implemented method for predicting sensor error comprising:
receiving sensor data from at least one sensor, the sensor data collected at a geographic location;
extracting a set of input features from the sensor data, map data representing the geographic location, or combination thereof; and
processing the set of input features using a machine learning model to calculate a predicted sensor error of a target sensor operating at the geographic location,
wherein the machine learning model has been trained on ground truth sensor error data to use the set of input features to calculate the predicted sensor error,
wherein the ground truth sensor data includes a sensed vehicle pose of a collecting vehicle, the sensed vehicle pose being used to determine a corrected vehicle pose,
wherein the ground truth sensor error values are based on an offset value between the sensed vehicle pose and the corrected vehicle pose.

2. The method of claim 1 , wherein the target sensor is a location sensor, and wherein the predicted sensor error is used as an error prior for localization.

3. The method of claim 2 , further comprising:
reducing a search space for the localization based on the error prior.

4. The method of claim 1 , wherein the machine learning model is deployed in a vehicle to provide for localizing the vehicle.

5. The method of claim 1 , wherein the at least one sensor is different from the target sensor, and wherein the at least one sensor includes a LiDAR sensor, a camera sensor, or a combination thereof.

6. The method of claim 1 , wherein the extracting of the set of input features comprises processing the sensor data to determine one or more structures or a combination thereof at the geographic location.

7. The method of claim 6 , wherein the set of input features includes one or more attributes of the one or more structures, one or more other attributes of the sensor data indicative of the one or more structures, or a combination thereof.

8. The method of claim 1 , wherein the sensor data is location data collected from a single location sensor of a vehicle.

9. An apparatus for predicting sensor error, comprising:
at least one processor; and
at least one memory including computer program code for one or more programs,
the at least one memory and the computer program code configured to, with the at least one processor, cause the apparatus to perform at least the following,
collect ground truth sensor error data for a geographic region, wherein the ground truth sensor data comprises a set of training features extracted from sensor data, map data, or combination thereof labeled with ground truth sensor error values; and
train a machine learning model using the ground truth sensor data to calculate a predicted sensor error from a set of input features,
wherein the ground truth sensor data includes a sensed vehicle pose of a collecting vehicle,
wherein the set of input features is extracted from sensor data subsequently collected from a geographic location for which the predicted sensor error for a target sensor is to be calculated;
run a compute-intensive localizer on the sensed vehicle pose in a grid pattern to identify a corrected vehicle pose,
wherein the ground truth sensor error values are based on an offset value between the sensed vehicle pose and the corrected vehicle pose.

10. The apparatus of claim 9 , wherein the target sensor is a location sensor, and wherein the predicted sensor error is used as an error prior for localization.

11. The apparatus of claim 10 , wherein the apparatus is further caused to:
reduce a search space for the localization based on the error prior.

12. The apparatus of claim 9 , wherein the machine learning model is deployed in a vehicle to provide for localizing the vehicle.

13. The apparatus of claim 9 , wherein the at least one sensor is different from the target sensor, and wherein the at least one sensor includes a LiDAR sensor, a camera sensor, a RADAR sensor or a combination thereof.

14. A non-transitory computer-readable storage medium for predicting sensor error, carrying one or more sequences of one or more instructions which, when executed by one or more processors, cause an apparatus to at least perform the following steps:
receiving sensor data from at least one sensor, the sensor data collected at a geographic location;
extracting a set of input features from the sensor data, map data representing the geographic location, or combination thereof; and
processing the set of input features using a machine learning model to calculate a predicted sensor error of a target sensor operating at the geographic location,
wherein the machine learning model has been trained on ground truth sensor error data to use the set of input features to calculate the predicted sensor error,
wherein the ground truth sensor data includes a sensed vehicle pose of a collecting vehicle, the sensed vehicle pose being used to determine a corrected vehicle pose,
wherein the ground truth sensor error values are based on an offset value between the sensed vehicle pose and the corrected vehicle pose.

15. The non-transitory computer-readable storage medium of claim 14 , wherein the target sensor is a location sensor, and wherein the predicted sensor error is used as an error prior for localization.

16. The non-transitory computer-readable storage medium of claim 14 , wherein the extracting of the set of input features comprises processing the sensor data to determine one or more structures or a combination thereof at the geographic location.

17. The non-transitory computer-readable storage medium of claim 16 , wherein the set of input features includes one or more attributes of the one or more structures, one or more other attributes of the sensor data indicative of the one or more structures, or a combination thereof.

18. The non-transitory computer-readable storage medium of claim 14 , wherein the sensor data is location data collected from a single location sensor of a vehicle.

Assignees

• Here Global Bv

Inventors

• Viswanathan Anirudh

Classifications

• G05D1/0088

  characterized by the autonomous decision making process, e.g. artificial intelligence, predefined behaviours (using knowledge based models G06N5/00) · CPC title

• G01S7/497

  Means for monitoring or calibrating · CPC title

• G06T2207/10028

  Range image; Depth image; 3D point clouds · CPC title

• G06N20/00

  Machine learning · CPC title

• H04N17/002

  for television cameras · CPC title