Methods and apparatus for validating sensor data

1. A computing device comprising:

a memory; and

at least one processor coupled to the memory, the at least one processor configured to:

receive lidar data generated by a lidar device coupled to a vehicle, wherein the lidar data includes a plurality of lidar data points indicative of locations of reflections from an environment of the vehicle;

receive a first image of a first portion of the environment captured by a camera at a first time, wherein the camera is coupled to the vehicle, and wherein the camera has a first pose at the first time according to pose data from a pose sensor coupled to the vehicle;

receive a second image of a second portion of the environment captured by the camera at a second time, wherein the camera has a second pose at the second time according to the pose data from the pose sensor;

determine a first location in the first image that corresponds to a lidar data point of the plurality of lidar data points;

determine a second location in the second image that corresponds to the lidar data point of the plurality of lidar data points;

perform a comparison of image contents at the first location in the first image and image contents at the second location in the second image; and

determine that sensor data associated with at least one of the lidar device, the camera, or the pose sensor is accurate or inaccurate based on the comparison.

2. The computing device of claim 1 , wherein performing the comparison comprises:

determining a first feature descriptor that characterizes, within a feature descriptor space, the image contents at the first location in the first image;

determining a second feature descriptor that characterizes, within the feature descriptor space, the image contents at the second location in the second image; and

determining a distance between the first feature descriptor and the second feature descriptor within the feature descriptor space.

3. The computing device of claim 2 , wherein the distance comprises a Hamming distance.

4. The computing device of claim 2 , wherein the first and second feature descriptors are n-bit binary strings, and wherein the feature descriptor space is an n-dimensional space.

5. The computing device of claim 1 , wherein the at least one processor is further configured to:

receive the pose data from the pose sensor; and

determine the first pose and the second pose based on the pose data.

6. The computing device of claim 1 , wherein the at least one processor is further configured to select the first image and the second image for the comparison based on at least (i) a time difference between the first time and the second time and (ii) a speed of the vehicle.

7. The computing device of claim 2 , wherein the first location corresponds to a first pixel in the first image, and wherein the second location corresponds to a second pixel in the second image.

8. The computing device of claim 7 , wherein the first feature descriptor is determined based on the first pixel and one or more neighboring pixels in the first image, and wherein the second feature descriptor is determined based on the second pixel and one or more neighboring pixels in the second image.

9. The computing device of claim 1 , wherein determining the first location comprises projecting the lidar data point into the first image, wherein determining the second location comprises projecting the lidar data point into the second image, and wherein the first portion of the environment overlaps with the second portion of the environment.

10. The computing device of claim 1 , wherein the lidar data point is associated with a static object in the environment.

11. The computing device of claim 1 , wherein the at least one processor is further configured to send a signal to cause the vehicle to stop, slow down, or pull over in response to determining that the sensor data is inaccurate.

12. The computing device of claim 1 , wherein the plurality of lidar data points comprises a three-dimensional point cloud.

13. A method comprising:

receiving, by a computing device, lidar data obtained by a lidar device coupled to a vehicle, wherein the lidar data includes a plurality of lidar data points indicative of locations of reflections from an environment of the vehicle;

receiving, by the computing device, a first image of a first portion of the environment captured by a camera at a first time, wherein the camera is coupled to the vehicle, and wherein the camera has a first pose at the first time according to pose data from a pose sensor coupled to the vehicle;

receiving, by the computing device, a second image of a second portion of the environment captured by the camera at a second time, wherein the camera has a second pose at the second time according to the pose data from the pose sensor;

determining, by the computing device, a first location in the first image that corresponds to a lidar data point of the plurality of lidar data points;

determining, by the computing device, a second location in the second image that corresponds to the lidar data point of the plurality of lidar data points;

performing, by the computing device, a comparison of image contents at the first location in the first image and image contents at the second location in the second image; and

determining, by the computing device, that sensor data associated with at least one of the lidar device, the camera, or the pose sensor is accurate or inaccurate based on the comparison.

14. The method of claim 13 , wherein performing the comparison comprises:

determining a first feature descriptor that characterizes, within a feature descriptor space, the image contents at the first location in the first image;

determining a second feature descriptor that characterizes, within the feature descriptor space, the image contents at the second location in the second image; and

determining a distance between the first feature descriptor and the second feature descriptor within the feature descriptor space.

15. The method of claim 14 , wherein the distance comprises a Hamming distance.

16. The method of claim 14 , wherein the first and second feature descriptors are n-bit binary strings, and wherein the feature descriptor space is an n-dimensional space.

17. A non-transitory computer-readable medium storing instructions, the instructions being executable by one or more processors to perform functions comprising:

receiving a first image of a first portion of an environment captured by a camera at a first time, wherein the camera is coupled to a vehicle, and wherein the camera has a first pose at the first time according to pose data from a pose sensor coupled to the vehicle;

receiving a second image of a second portion of the environment captured by the camera at a second time, wherein the camera has a second pose at the second time according to the pose data from the pose sensor;

determining a first location in the first image that corresponds to a lidar data point of a plurality of data points;

determining a second location in the second image that corresponds to the lidar data point of the plurality of lidar data points;

performing a comparison of image contents at the first location in the first image and image contents at the second location in the second image; and

determining that sensor data associated with at least one of a lidar device, the camera, or the pose sensor is accurate or inaccurate based on the comparison.

18. The non-transitory computer-readable medium of claim 17 , wherein performing the comparison comprises:

determining a first feature descriptor that characterizes, within a feature descriptor space, the image contents at the first location in the first image;

determining a second feature descriptor that characterizes, within the feature descriptor space, the image contents at the second location in the second image; and

determining a distance between the first feature descriptor and the second feature descriptor within the feature descriptor space.

19. The non-transitory computer-readable medium of claim 18 , wherein the distance comprises a Hamming distance, and wherein the first and second feature descriptors are n-bit binary strings, and wherein the feature descriptor space is an n-dimensional space.

20. The non-transitory computer-readable medium of claim 17 , further comprising instructions for selecting the first image and the second image for the comparison, wherein the selecting is based on at least (i) a time difference between the first time and the second time and (ii) a speed of the vehicle.