Vehicle pose system
Various examples are directed to systems and methods for determining a pose of a vehicle. A first localizer may generate a first pose estimate for the vehicle based at least in part on a comparison of first remote sensor data to a first reference data. A second localizer may generate a second pose estimate for the vehicle based at least in part on a comparison of second remote sensor data to a second reference data. A pose state estimator may generate a vehicle pose for the vehicle based at least in part on a first previous pose of the vehicle, the first pose estimate, and the second pose estimate.
1. A system for determining a pose of a vehicle, comprising:
at least one processor unit;
a machine-readable medium comprising instructions thereon that, when executed by the at least one processor unit, cause the at least one processor unit to perform operations comprising:
generating, by a first localizer executed by the at least one processor unit, a first pose estimate for the vehicle based at least in part on a comparison of first remote sensor data to first reference data;
generating, by a second localizer executed by the at least one processor unit, a second pose estimate for the vehicle based at least in part on a comparison of second remote sensor data to second reference data; and
generating, by a pose state estimator executed by the at least one processor unit, a vehicle pose for the vehicle based at least in part on a first previous pose of the vehicle, the first pose estimate, and the second pose estimate.
2. The system of claim 1 , wherein the first remote sensor data describes ground reflectivity data, and wherein the first reference data describes ground reflectivity data associated with a map.
3. The system of claim 2 , wherein the first remote sensor data comprises ground reflectivity data, and wherein the machine-readable medium further comprises instructions thereon that, when executed by the at least one processor unit, causes the at least one processor unit to perform operations comprising:
selecting, based at least in part on the ground reflectivity data, a first horizontal axis position on a first horizontal axis, a second horizontal axis position on a second horizontal axis perpendicular to the first horizontal axis, and a yaw about a vertical axis perpendicular to the first horizontal axis and the second horizontal axis; and
selecting a vertical axis position, a pitch about the first horizontal axis, and a roll about the second horizontal axis, wherein the selecting is based at least in part on the first horizontal axis position, the second horizontal axis position, the yaw, and the ground reflectivity data.
4. The system of claim 1 , wherein the first remote sensor data describes sensed height and sensed reflectivity for a plurality of ground points.
5. The system of claim 1 , wherein the second remote sensor data describes a position of a first object, and wherein the second reference data describes the position of the first object.
6. The system of claim 1 , wherein the first pose estimate corresponds to a first time stamp and the second pose estimate corresponds to a second time stamp after the first time stamp, wherein the first previous pose is for the first time stamp, and wherein the first previous pose is based at least in part on the first pose estimate.
7. The system of claim 6 , wherein the machine-readable medium further comprises instructions thereon that, when executed by the at least one processor unit, causes the at least one processor unit to perform operations comprising generating, by the pose state estimator, an initial pose for the second time stamp based at least in part on the first pose estimate and first sensor data describing the vehicle.
8. The system of claim 1 , wherein the machine-readable medium further comprises instructions thereon that, when executed by the at least one processor unit, causes the at least one processor unit to perform operations comprising determining, by the pose state estimator, that a first covariance indicator for the first pose estimate indicates a lower confidence in the first pose estimate than a second covariance indicator indicates for the second pose estimate, wherein the vehicle pose is closer to the first pose estimate than to the second pose estimate.
9. The system of claim 1 , wherein the machine-readable medium further comprises instructions thereon that, when executed by the at least one processor unit, causes the at least one processor unit to perform operations comprising:
determining, by the pose state estimator, that a first covariance indicator for the first pose estimate indicates a higher confidence level in a first dimension than a second covariance indicator for the second pose estimate; and
determining, by the pose state estimator, that the second covariance indicator indicates a higher confidence level in a second dimension than the first covariance indicator, wherein the vehicle pose is closer to the first pose estimate than to the second pose estimate in the first dimension, and wherein the vehicle pose is closer to the second pose estimate than to the first pose estimate in the second dimension.
10. A computer-implemented method for determining a pose of a vehicle, comprising:
generating, by at least one processor unit, a first pose estimate for the vehicle based at least in part on a comparison of first remote sensor data to first reference data;
generating, by the at least one processor unit, a second pose estimate for the vehicle based at least in part on a comparison of second remote sensor data to second reference data; and
generating, by the at least one processor unit, a vehicle pose for the vehicle based at least in part on a first previous pose of the vehicle, the first pose estimate, and the second pose estimate.
11. The method of claim 10 , wherein the first remote sensor data includes ground reflectivity data, and wherein the first reference data includes ground reflectivity data associated with a map.
12. The method of claim 11 , wherein the first remote sensor data comprises ground reflectivity data, further comprising:
selecting, based at least in part on the ground reflectivity data, a first horizontal axis position on a first horizontal axis, a second horizontal axis position on a second horizontal axis perpendicular to the first horizontal axis, and a yaw about a vertical axis perpendicular to the first horizontal axis and the second horizontal axis; and
selecting a vertical axis position, a pitch about the first horizontal axis, and a roll about the second horizontal axis, wherein the selecting is based at least in part on the first horizontal axis position, the second horizontal axis position, the yaw, and the ground reflectivity data.
13. The method of claim 10 , wherein the first remote sensor data describes sensed height and sensed reflectivity for a plurality of ground points.
14. The method of claim 10 , wherein the second remote sensor data describes a position of a first object, and wherein the second reference data describes the position of the first object.
15. The method of claim 10 , wherein the first pose estimate corresponds to a first time stamp and the second pose estimate corresponds to a second time stamp after the first time stamp, wherein the first previous pose is for the first time stamp, and wherein the first previous pose is based at least in part on the first pose estimate.
16. The method of claim 15 , further comprising generating, by the at least one processor unit, an initial pose for the second time stamp based at least in part on the first pose estimate and first sensor data describing the vehicle.
17. The method of claim 10 , further comprising determining, by the at least one processor unit, that a first covariance indicator for the first pose estimate indicates that level than a second covariance indicator for the second pose estimate, wherein the vehicle pose is closer to the first pose estimate than to the second pose estimate.
18. The method of claim 10 , further comprising:
determining, by the at least one processor unit, that a first covariance indicator for the first pose estimate indicates a higher confidence level in a first dimension than a second covariance indicator for the second pose estimate; and
determining, by the at least one processor unit, that the second covariance indicator indicates a higher confidence level in a second dimension than the first covariance indicator, wherein the vehicle pose is closer to the first pose estimate than to the second pose estimate in the first dimension, and wherein the vehicle pose is closer to the second pose estimate than to the first pose estimate in the second dimension.
19. A tangible machine-readable medium comprising instructions thereon that, when executed by at least one processor unit, causes the at least one processor unit to perform operations comprising:
generating, by a first localizer executed by the at least one processor unit, a first pose estimate for a vehicle based at least in part on a comparison of first remote sensor data to first reference data;
generating, by a second localizer executed by the at least one processor unit, a second pose estimate for the vehicle based at least in part on a comparison of second remote sensor data to second reference data; and
generating, by a pose state estimator executed by the at least one processor unit, a vehicle pose for the vehicle based at least in part on a first previous pose of the vehicle, the first pose estimate, and the second pose estimate.
20. The machine-readable medium of claim 19 , wherein the machine-readable medium further comprises instructions thereon that, when executed by the at least one processor unit, causes the at least one processor unit to perform operations comprising determining, by the pose state estimator, that a first covariance indicator for the first pose estimate indicates a lower confidence in the first pose estimate than a second covariance indicator indicates for the second pose estimate, wherein the vehicle pose is closer to the first pose estimate than to the second pose estimate.