Calibration for autonomous vehicle operation

1. A system, comprising:

an image sensor disposed about an autonomous vehicle and configured to sense an environment and capture image data;

a LIDAR sensor configured to capture LIDAR data; and

one or more processors communicatively coupled to the image sensor and the LIDAR sensor, the one or more processors being configured to perform operations comprising:

receiving a signal from the image sensor, the signal comprising data representing a sensor measurement of the environment;

determining, based at least in part on the data, that the image sensor is miscalibrated, wherein determining that the image sensor is miscalibrated comprises:

determining, based at least in part on the LIDAR data, a set of detected edges associated with an object in the environment; and

determining region data associated with the set of detected edges;

generating an expected sensor measurement, the expected sensor measurement associated with the image sensor;

generating a calibration parameter, associated with the image sensor, based at least in part on the expected sensor measurement; and

modifying a parameter of the image sensor based at least in part on the calibration parameter and a parameter associated with the region data.

2. The system of claim 1 ,

wherein the LIDAR sensor is a first LIDAR sensor,

wherein the system further comprises a plurality of additional LIDAR sensors, and

the operations further comprising determining that the first LIDAR sensor is miscalibrated, and wherein determining that the first LIDAR sensor is miscalibrated comprises:

determining a first sensor measurement of the first LIDAR sensor comprising a first distance or a first reflectivity value;

determining a plurality of additional sensor measurements of the plurality of additional LIDAR sensors comprising a second distance or a second reflectivity value, the first sensor measurement and the plurality of additional sensor measurements being associated with an object in the environment; and

determining a difference between the first sensor measurement and at least one measurement of the plurality of additional sensor measurements, and

the operations further comprising modifying a parameter of the first LIDAR sensor, based at least in part on the difference.

3. The system of claim 1 ,

wherein modifying the parameter of the image sensor comprises adjusting a focus parameter or a lens parameter of the image sensor.

4. The system of claim 1 ,

wherein the system further comprises an IMU configured to capture IMU sensor data and a GPS configured to capture GPS sensor data, and

wherein generating the expected sensor measurement comprises:

fusing at least two of the image data, the LIDAR data, the IMU sensor data, or the GPS sensor data to generate fused data;

receiving, based at least in part on the fused data, map data, the map data indicative of an object in the environment having known properties; and

generating the expected sensor measurement based at least in part on the object.

5. The system of claim 1 , wherein generating the calibration parameter comprises:

receiving additional data from a log file, a map, or an additional sensor disposed about the system; and

generating, based at least in part on the additional data, the calibration parameter, wherein the calibration parameter is configured to alter an intrinsic value associated with the image sensor.

6. The system of claim 5 , wherein the additional data comprises an image captured by a camera, a velocity of the system captured by an odometry sensor, a position captured by a global positioning (GPS) sensor, audio data captured by a microphone, sound navigation and ranging (SONAR) data captured by a SONAR sensor, radio detection and ranging (RADAR) data captured by a RADAR sensor, direction of gravity data captured by an inertial measurement unit (IMU), or a direction of travel data.

7. The system of claim 1 , wherein the operations further comprise:

generating, based at least in part on the calibration parameter, a vehicle trajectory, the vehicle trajectory configured to cause the system to traverse a portion of the environment.

8. The system of claim 1 , wherein the operations further comprise:

generating a first probabilistic model; and

generating, based at least in part on a heuristic rule, a second probabilistic model,

wherein generating the calibration parameter comprises using the first probabilistic model, and

wherein modifying the image sensor comprises using the second probabilistic model.

9. A method, comprising:

receiving a plurality of sensor measurements from a plurality of sensors disposed about a vehicle, wherein the plurality of sensors comprise a LIDAR sensor configured to capture LIDAR sensor data and an image sensor configured to capture image sensor data;

identifying, based at least in part on the plurality of sensor measurements, an anomalous sensor measurement of the plurality of sensor measurements associated with the image sensor of the plurality of sensors, wherein identifying the anomalous sensor measurement comprises:

determining a set of edges in the LIDAR sensor data;

associating, as region data, at least a portion of the image sensor data with the LIDAR sensor data; and

identifying a portion of the region data that is anomalous;

determining, based at least in part on the anomalous sensor measurement, that the image sensor is miscalibrated;

generating an expected sensor measurement, associated with the image sensor; and

generating, based at least in part on the expected sensor measurement, a calibration parameter for calibrating the image sensor.

10. The method of claim 9 , further comprising:

generating fused sensor data using data received from the LIDAR sensor and the image sensor, the fused sensor data indicative of a localization of the vehicle; and

generating a probabilistic map of an environment based at least in part on the fused sensor data, the probabilistic map including a plurality of labeled objects and a probability score assigned to at least one object of the plurality of labeled objects, the plurality of labeled objects being associated with object parameters,

wherein identifying the anomalous sensor measurement comprises comparing data received from the image sensor or the LIDAR sensor with the object parameters.

11. The method of claim 10 , wherein generating the probabilistic map comprises:

accessing historical sensor data corresponding to the plurality of sensors; and

generating the probability score based at least in part on the historical sensor data and the anomalous sensor measurement.

12. The method of claim 9 , wherein identifying the portion of the region data that is anomalous comprises

identifying a portion of the region data that is blurry.

13. The method of claim 9 , wherein the LIDAR sensor is configured to capture a first sensor measurement and an additional LIDAR sensor is configured to capture a second sensor measurement,

the first sensor measurement and the second sensor measurement associated with a distance of an object to the vehicle or a reflectivity of the object, and

wherein identifying the anomalous sensor measurement comprises comparing the first sensor measurement and the second sensor measurement.

14. The method of claim 9 , wherein the vehicle is an autonomous vehicle, the method further comprising:

generating, based on the calibration parameter, a trajectory configured to cause the autonomous vehicle to traverse an environment, and

transmitting the trajectory to the autonomous vehicle.

15. One or more non-transitory computer readable media having instructions stored thereon which, when executed by one or more processors of a system, cause the system to perform operations comprising:

receiving first sensor data associated with a first measurement made by a first sensor disposed about the system, the first sensor comprising an image sensor;

receiving second sensor data associated with a second sensor disposed about the system, the second sensor comprising a LIDAR sensor;

determining, based at least in part on the first sensor data and the second sensor data, that the first sensor data is anomalous, wherein determining that the first sensor data is anomalous comprises:

detecting a set of detected edges in the second data;

determining, as region data, at least a portion of the first sensor data with the set of detected edges; and

determining, based at least in part on the region data, that the first sensor data is anomalous; and

determining an expected sensor measurement based at least in part on the first measurement and the second sensor data;

determining a calibration parameter associated with the first sensor based at least in part on the expected sensor measurement; and

calibrating the first sensor based at least in part on the calibration parameter.

16. The one or more non-transitory computer readable media of claim 15 ,

wherein determining that the first sensor data is anomalous comprises

determining that the first sensor data is blurry.

17. The one or more non-transitory computer readable media of claim 15 , wherein the second data comprises second sensor data acquired from a plurality of additional sensors comprising an additional LIDAR sensor, an additional image sensor, a GPS sensor, an IMU sensor, or a rotary encoder, the operations further comprising:

generating fused sensor data using the first sensor data and the second sensor data, the fused sensor data indicative of a localization of the system; and

accessing, based at least in part on the localization, map data, the map data indicative of an object in an environment, the object associated with one or more object properties,

wherein determining that the first sensor data is anomalous comprises determining that the first sensor data differs from the one or more object properties.

18. The one or more non-transitory computer readable media of claim 17 , wherein determining the calibration parameter comprises:

determining the calibration parameter based at least in part on a generative probabilistic model and using the map data,

the map data further comprising information indicative of past measurements of the object made using the plurality of additional sensors.

19. The one or more non-transitory computer readable media of claim 15 , wherein the second sensor data is acquired from the LIDAR sensor, the first sensor data and the second sensor data corresponding to an object in an environment, wherein

the first sensor data comprises a first distance or a first reflectivity,

the second sensor data comprises a second distance or a second reflectivity, and

determining that the first sensor data is anomalous comprises determining a difference between the first sensor data and the second sensor data.

20. The one or more non-transitory computer readable media of claim 15 , the operations further comprising:

receiving additional sensor data associated with a second measurement made by the calibrated first sensor, the second measurement being associated with an environment;

generating, based at least in part on the additional sensor data, a trajectory, the trajectory configured to cause an autonomous vehicle to traverse a portion of the environment; and

transmitting the trajectory to the autonomous vehicle.