Detecting sensor occlusion with compressed image data
The technology relates to detecting possible imaging sensor occlusion. In one example, a system including an imaging sensor and one or more processors may be configured to capture first image data using the imaging sensor. The one or more processors may encode the first image data into an uncompressed image file and generate a compressed image file based on the uncompressed image file. The file size of the compressed image file may be determined and based on the file size of the compressed image file, the system may determine that the imaging sensor is possibly occluded.
1. A method of detecting possible imaging sensor occlusion, the method comprising:
capturing first image data using an imaging sensor;
encoding, by one or more processors, the first image data into an uncompressed image file;
generating, by the one or more processors, a compressed image file based on the uncompressed image file;
determining, by the one or more processors, a file size of the compressed image file; and
determining, by the one or more processors, based on the file size of the compressed image file, that the imaging sensor is possibly occluded.
2. The method of claim 1 , further comprising, comparing the file size of the compressed image file to a threshold file size, and
wherein determining that the imaging sensor is occluded further includes determining the file size of the compressed image file meets the threshold file size.
3. The method of claim 2 , further comprising:
determining the threshold file size based on an average file size of compressed image files generated by one or more imaging sensors known to not have an occlusion.
4. The method of claim 2 , further comprising:
determining the threshold file size based on the smallest file size of a compressed image captured during a training period by the imaging sensor, wherein the training period corresponds to a predefined number of frames or a predefined time period.
5. The method of claim 2 , further comprising:
determining the threshold file size based on a running average file size of a set of compressed image files generated by the imaging sensor.
6. The method of claim 5 , wherein the threshold file size is within a predefined range of the running average file size.
7. The method of claim 2 , further comprising:
determining the threshold file size based on compressed image files corresponding to image data captured at a similar time and/or location of the first image data.
8. The method of claim 2 , wherein subsequent to determining that the imaging sensor is occluded, adjusting the threshold file size in response to receiving input that the imaging sensor is not occluded.
9. The method of claim 1 , further comprising:
generating one or more additional compressed image files based on additional image data captured subsequent to the first image data,
determining a file size of the one or more additional image files, and
wherein determining the imaging sensor is occluded is further based on the file size of the one or more additional images.
10. The method of claim 9 , further comprising determining a rate of change between the file size of the compressed image and the file size of the one or more additional image files,
wherein determining the imaging sensor is occluded further includes determining the rate of change is below a threshold value.
11. The method of claim 1 , further comprising, sending a signal to activate a cleaning system in order to clean the imaging sensor based on the determination that the imaging sensor is occluded.
12. The method of claim 1 , wherein the imaging sensor is attached to a vehicle having an autonomous driving mode, and the method further comprises using the determination to control the vehicle in the autonomous driving mode.
13. The method of claim 1 , wherein the imaging sensor is a LIDAR sensor, radar unit, or camera.
14. A system for detecting possible imaging sensor occlusion comprising: an imaging sensor; and
one or more processors, wherein the one or more processors are configured to:
capture first image data using the imaging sensor,
encode the first image data into an uncompressed image file,
generate a compressed image file based on the uncompressed image file,
determine a file size of the compressed image file, and
determine based on the file size of the compressed image file, that the imaging sensor is possibly occluded.
15. The system of claim 14 , wherein the one or more processors are further configured to compare the file size of the compressed image file to a threshold file size, and
wherein determining that the imaging sensor is occluded further includes determining the file size of the compressed image file meets the threshold file size.
16. The system of claim 15 , wherein the one or more processors are further configured to determine the threshold file size based on an average file size of compressed image files generated by one or more imaging sensors known to not have an occlusion.
17. The system of claim 15 , wherein the one or more processors are further configured to determine the threshold file size based on the smallest file size of a compressed image captured during a training period by the imaging sensor, wherein the training period corresponds to a predefined number of frames or a predefined time period.
18. The system of claim 15 , wherein the one or more processors are further configured to determine the threshold file size based on a running average file size of a set of compressed image files generated by the imaging sensor.
19. The system of claim 15 , wherein the one or more processors are further configured to determine the threshold file size based on compressed image files corresponding to image data captured at a similar time and/or location of the first image data.
20. The system of claim 14 , wherein the imaging sensor is a LIDAR sensor, RADAR unit, or camera.