Method for temporally calibrating sensor streams in an autonomous vehicle
View Patent ↗One variation of a method for temporally calibrating sensor streams in an autonomous vehicle includes: deriving a first set of longitudinal velocities of a reference point on the autonomous vehicle from a sequence of inertial data recorded over a period of time; deriving a second set of longitudinal velocities of the reference point based on features detected in a set of LIDAR frames recorded during the period of time; calculating a LIDAR sensor offset time that approximately minimizes a difference between the first set of longitudinal velocities and the second set of longitudinal velocities; and, in response to the LIDAR sensor offset time approximating an previous LIDAR sensor offset time, verifying operation of the LIDAR sensor during the first period of time.
1. A method for temporally calibrating sensor streams, the method comprising:
determining a first set of longitudinal accelerations at a wheel of a vehicle using inertial data output by an inertial sensor over a period of time;
determining a second set of longitudinal accelerations at the wheel using a sequence of angular position changes output by a wheel sensor corresponding to the wheel over the period of time;
calculating a wheel sensor offset time by minimizing a difference between the first set of longitudinal accelerations and the second set of longitudinal accelerations; and
determining a temporal alignment of a first data stream output by the inertial sensor and a second data stream output by the wheel sensor based on a comparison of the wheel sensor offset time to an initial wheel sensor offset time.
2. The method of claim 1 , wherein at least one of the wheel sensor or the inertial sensor is calibrated based on the temporal alignment.
3. The method of claim 1 , wherein the wheel sensor offset time is calculated by shifting a first plot relative to a second plot by a set of test times and identifying a particular test time in the set of test times that minimizes an area between the first plot and the second plot, the first plot corresponding to the first set of longitudinal accelerations and the second plot corresponding to the second set of longitudinal accelerations.
4. The method of claim 1 , further comprising:
triggering a vehicle mode transition for the vehicle based on the temporal alignment.
5. The method of claim 4 , wherein the vehicle mode transition includes a transition from an autonomous mode to a manual mode.
6. The method of claim 1 , further comprising:
triggering a change in operation of the vehicle based on the temporal alignment.
7. The method of claim 6 , wherein the change in operation includes at least one of reducing speed or stopping.
8. The method of claim 1 , further comprising:
detecting an operation issue of at least one of the wheel sensor or the inertial sensor based on the temporal alignment.
9. The method of claim 8 , wherein a maintenance vehicle is dispatched in response to detection of the operation issue.
10. One or more tangible non-transitory computer-readable storage media storing computer-executable instructions for performing a computer process on a computing system, the computer process comprising:
determining a first set of longitudinal velocities of a reference point corresponding to a vehicle using inertial data output by an inertial sensor over a period of time between an initial time and a first time;
determining a second set of longitudinal velocities of the reference point based on features detected in an initial LIDAR frame output by a LIDAR sensor at the initial time and features detected in a first LIDAR frame output by the LIDAR sensor at the first time;
calculating a LIDAR sensor offset time by minimizing a difference between the first set of longitudinal velocities and the second set of longitudinal velocities; and
determining a temporal alignment of a first data stream output by the inertial sensor and a second data stream output by the LIDAR sensor based on a comparison of the LIDAR sensor offset time to an initial LIDAR sensor offset time.
11. The one or more tangible non-transitory computer-readable storage media of claim 10 , the computer process further comprising:
estimating a velocity of one or more objects in a scene around the vehicle based on the temporal alignment.
12. The one or more tangible non-transitory computer-readable storage media of claim 10 , the computer process further comprising:
determining a current position of the vehicle based on the temporal alignment.
13. The one or more tangible non-transitory computer-readable storage media of claim 10 , the computer process further comprising:
verifying an operation of at least one of at least one of the LIDAR sensor or the inertial sensor based on the temporal alignment.
14. The one or more tangible non-transitory computer-readable storage media of claim 10 , the computer process further comprising:
verifying an operation of the LIDAR sensor by comparing the LIDAR sensor offset time to an initial LIDAR sensor offset time.
15. The one or more tangible non-transitory computer-readable storage media of claim 10 , the computer process further comprising:
detecting one or more objects in a scene around the vehicle based on the temporal alignment.
16. One or more tangible non-transitory computer-readable storage media storing computer-executable instructions for performing a computer process on a computing system, the computer process comprising:
determining a first set of longitudinal accelerations at a wheel of a vehicle using a first sequence of data received output by a first sensor of the vehicle over a first period of time;
determining a second set of longitudinal accelerations at the wheel using a second sequence of data output by a second sensor over the first period of time;
calculating a first sensor offset time by minimizing a difference between the first set of longitudinal accelerations and the second set of longitudinal accelerations;
determining a first set of longitudinal velocities of a reference point corresponding to the vehicle using a third sequence of data output by the first sensor over a second period of time between an initial time and the first time;
determining a second set of longitudinal velocities of the reference point using a fourth sequence of data output by a third sensor over the second period of time;
calculating a second sensor offset time by minimizing a difference between the first set of longitudinal velocities and the second set of longitudinal velocities;
verifying an operation of the first sensor during the first period of time based on a comparison of the first sensor offset time and a first initial sensor offset time; and
verifying an operation of the third sensor during the first period of time based on a comparison of the second sensor offset time and a second initial sensor offset time.
17. The one or more tangible non-transitory computer-readable storage media of claim 16 , wherein the first sensor is an inertial sensor, the second sensor is a wheel sensor, and the third sensor is a LIDAR sensor.
18. The one or more tangible non-transitory computer-readable storage media of claim 16 , the computer process further comprising:
triggering a change in operation of the vehicle based on verification of the operation of the first sensor and the operation of the third sensor.
19. The one or more tangible non-transitory computer-readable storage media of claim 16 , wherein a maintenance vehicle is dispatched to the vehicle based on verification of the operation of the first sensor and the operation of the third sensor.
20. The one or more tangible non-transitory computer-readable storage media of claim 16 , the computer process further comprising:
triggering a vehicle mode transition for the vehicle based on the temporal alignment.