Lidar obstruction detection
Techniques for detecting an obstruction associated with a lidar sensor are discussed herein. For example, a computing device can implement an obstruction detection component to detect rain, mud, dirt, dust, snow, ice, animal droppings, etc., on and/or near an outer surface of the lidar sensor. The obstruction detection component can apply one or more heuristics and/or models to the lidar data and/or compare pulse information associated with the lidar data to a threshold to determine a size, a type, or a location of an obstruction blocking a lidar beam.
1. A system comprising:
one or more processors; and
one or more non-transitory computer-readable media storing computer-executable instructions that, when executed, cause the one or more processors to perform operations comprising:
receiving lidar data associated with a lidar sensor;
determining that the lidar data comprises first pulse information indicating a first reflection of a first pulse at a first distance that is less than or equal to a threshold distance from a lens of the lidar sensor and that the first pulse was emitted at a first angle;
determining second pulse information indicating a second reflection of a second pulse at a second distance that is greater than the threshold distance from the lens of the lidar sensor and that the second pulse was emitted at a second angle;
determining a difference between the first distance and the second distance;
determining, based at least in part on the difference meeting or exceeding a threshold difference, a degree of transparency associated with the lens of the lidar sensor;
determining that the degree of transparency is below a transparency threshold; and
performing an action based at least in part on the degree of transparency being below the transparency threshold, the action including at least one of:
initiating a cleaning operation to clean the lens of the lidar sensor; or
causing an autonomous vehicle to traverse an environment.
2. The system of claim 1 , the operations further comprising:
receiving angle information or spatial information associated with the lidar data; and
determining, based at least in part on the angle information or the spatial information, a size of an obstruction associated with the lens of the lidar sensor,
wherein performing the action is further based at least in part on the size of the obstruction.
3. The system of claim 1 , the operations further comprising:
receiving angle information or spatial information associated with the lidar data; and
determining, based at least in part on the angle information or the spatial information, a region of the lens comprising an obstruction,
wherein performing the action is further based at least in part on the region of the lens comprising the obstruction.
4. The system of claim 1 , the operations further comprising:
determining, based at least in part on the degree of transparency, an obstruction on at least a portion of the lens of the lidar sensor; and
determining, based at least in part on the degree of transparency, a classification of the obstruction,
wherein performing the action is further based at least in part on the classification.
5. The system of claim 4 , wherein determining the classification of the obstruction is based at least in part on a comparison between the first pulse information and the second pulse information.
6. One or more non-transitory computer-readable media storing instructions executable by a processor, wherein the instructions, when executed, cause the processor to perform operations comprising:
receiving lidar data associated with a lidar sensor;
detecting a first pulse of the lidar data indicating a first reflection of the first pulse at a first distance that is less than or equal to a threshold distance from a lens of the lidar sensor, the first pulse emitted at a first angle;
detecting a second pulse of the lidar data indicating a second reflection of the second pulse at a second distance greater than the threshold distance from the lens of the lidar sensor, the second pulse emitted at a second angle;
determining a difference between the first distance and the second distance;
determining, based at least in part on the difference meeting or exceeding a threshold difference, a degree of transparency associated with the lens of the lidar sensor;
determining that the degree of transparency is below a transparency threshold; and
performing an action based at least in part on the degree of transparency being below the transparency threshold, the action including at least one of:
initiating a cleaning operation to clean the lens of the lidar sensor; or
causing an autonomous vehicle to traverse an environment.
7. The one or more non-transitory computer-readable media of claim 6 , wherein determining the degree of transparency associated with the lens of the lidar sensor comprises inputting the lidar data into a machine learned model and receiving, from the machine learned model, the degree of transparency.
8. The one or more non-transitory computer-readable media of claim 6 , wherein determining the first pulse or the second pulse comprises determining a portion of the lidar data meets or exceeds a threshold difference from a noise floor.
9. The one or more non-transitory computer-readable media of claim 6 , wherein:
the first pulse is generated in response to an emission from the lidar sensor at the first angle, and
determining the degree of transparency is based at least in part on the first pulse and the second pulse associated with the lidar data generated in response to the emission.
10. The one or more non-transitory computer-readable media of claim 6 , the operations further comprising:
comparing, as a comparison, the first angle or first spatial information associated with the first pulse to the second angle or second spatial information associated with the second pulse; and
determining, based at least in part on the comparison, a size or a location of an obstruction on the lidar sensor,
wherein performing the action is further based at least in part on the size or the location of the obstruction on the lidar sensor.
11. The one or more non-transitory computer-readable media of claim 6 , wherein the action comprises generating a request for lidar point cloud data, the lidar point cloud data associated with the second distance exceeding the threshold distance, and the operations further comprising:
inputting the lidar point cloud data and the lidar data into a machine learned model; and
receiving, from the machine learned model, an output indicating a likelihood that an obstruction on the lens of the lidar sensor,
wherein performing the action is further based at least in part on the output.
12. The one or more non-transitory computer-readable media of claim 6 , wherein:
the autonomous vehicle is bi-directional, and
controlling the autonomous vehicle in the environment comprises determining a direction of travel for the autonomous vehicle based at least in part on a location of the lidar sensor on the autonomous vehicle.
13. The one or more non-transitory computer-readable media of claim 6 , wherein the lidar data includes raw lidar data that represents data prior to be converted to lidar point cloud data.
14. The one or more non-transitory computer-readable media of claim 6 , the operations further comprising:
determining, based at least in part on the degree of transparency, an obstruction on at least a portion of the lens of the lidar sensor; and
determining a classification of the obstruction,
wherein performing the action is further based at least in part on the classification.
15. The one or more non-transitory computer-readable media of claim 6 , the operations further comprising:
adjusting a power output of the lidar sensor based at least in part on the degree of transparency associated with the lens of the lidar sensor.
16. The one or more non-transitory computer-readable media of claim 6 , wherein detecting the first pulse or the second pulse comprises comparing a first region associated with the lidar data to a second region associated with the lidar data.
17. A method comprising:
receiving lidar data associated with a lidar sensor;
detecting a first pulse of the lidar data indicating a first reflection of the first pulse at a first distance less than or equal to a threshold distance from a lens of the lidar sensor, the first pulse emitted at a first angle;
detecting a second pulse of the lidar data indicating a second reflection of the second pulse at a second distance greater than the threshold distance from the lens of the lidar sensor, the second pulse emitted at a second angle;
determining a difference between the first distance and the second distance;
determining, based at least in part on the difference meeting or exceeding a threshold difference, a degree of transparency associated with the lens of the lidar sensor;
determining that the degree of transparency is below a transparency threshold; and
performing an action based at least in part on the degree of transparency being below the transparency threshold, the action including at least one of:
initiating a cleaning operation to clean the lens of the lidar sensor; or
causing an autonomous vehicle to traverse an environment.
18. The method of claim 17 , wherein initiating the cleaning operation to clean the lens of the lidar sensor comprises at least one of: generating a signal to cause an air jet or cleaning system to remove an obstruction on the lens of the lidar sensor, or adjusting a parameter of the lidar sensor to filter additional lidar data.
19. The method of claim 17 , further comprising:
comparing, as a comparison, the first angle or first spatial information associated with the first pulse of the lidar data to the second angle or second spatial information associated with the second pulse of the lidar data; and
determining, based at least in part on the comparison, a size or a location of an obstruction on the lidar sensor,
wherein performing the action is further based at least in part on the size or the location of the obstruction on the lidar sensor.
20. The method of claim 17 , wherein the action comprises generating a request for lidar point cloud data, the lidar point cloud data associated with a distance meeting or exceeding a distance threshold, and further comprising:
inputting the lidar point cloud data and the lidar data into a machine learned model, and
receiving, from the machine learned model, an output indicating a likelihood that an obstruction on the lens of the lidar sensor,
wherein performing the action is further based at least in part on the output.