Kurtosis based pruning for sensor-fusion systems
This document describes Kurtosis based pruning for sensor-fusion systems. Kurtosis based pruning minimizes a total quantity of comparisons performed when fusing together large sets of data. Multiple candidate radar tracks may possibly align with one of multiple candidate visual tracks. For each candidate vision track, a weight or other evidence of matching is assigned to each candidate radar track. An inverse of matching errors between each candidate vision and each candidate radar track contributes to this evidence, which may be normalized to produce, for each candidate vision track, a distribution associated with all candidate radar tracks. A Kurtosis or shape of this distribution is calculated. Based on the Kurtosis values, some candidate radar tracks are selected for matching and other remaining candidate radar tracks are pruned. The Kurtosis aids in determining how many candidates to retain and how many to prune. In this way, Kurtosis based pruning can prevent combinatorial explosions due to large-scale matching.
1. A method comprising:
determining, by a sensor-fusion system, a plurality of first candidate object-tracks according to first sensor data obtained from a first set of sensors;
determining, by the sensor-fusion system, a set of second object-tracks according to second sensor data obtained from a second set of sensors;
for each of the plurality of first candidate object-tracks:
determining a respective matching error between said each first candidate object-track and each of the set of second object-tracks;
computing a respective weight for consideration of matching said each first candidate object-track to said each second object-tracks to be an inverse of the respective matching error between said each first candidate object-track and said each second object-track;
applying the respective weight for said each second object-track to that first candidate object-track;
determining a respective distribution of all respective weights applied to the set of second object-tracks for said each first candidate object-track; and
determining a Kurtosis value of the respective distribution for said each first candidate object-track;
pruning at least one candidate object-track from the plurality of first candidate object tracks by:
dividing the plurality of first candidate object-tracks into two subsets including a first subset of the plurality of first candidate object-tracks having higher Kurtosis values than a second subset of the plurality of first candidate object-tracks, the second subset of the plurality of first candidate object-tracks including the at least one first candidate object-track; and
discarding the second subset from the plurality of first candidate object-tracks; and
matching, by the sensor-fusion system, said each second object-track with one or more remaining first candidate object-tracks of the plurality of first candidate object-tracks after the second set is discarded and the at least one first candidate object-track is pruned.
2. The method of claim 1 , wherein pruning the at least one first candidate object-track from the plurality of first candidate object-tracks comprises:
discarding the second subset from the plurality of first candidate object-tracks in response to determining that the first subset has the higher Kurtosis values than the second subset.
3. The method of claim 2 , wherein discarding the second subset from the plurality of first candidate object-tracks comprises retaining the at least one first candidate object-track for matching against a different subset of the plurality of first candidate object-tracks.
4. The method of claim 1 , wherein the first set of sensors comprise one or more radar sensors, lidar sensors, and ultrasound sensors, and wherein the second set of sensors comprises one or more visual sensors.
5. The method of claim 1 , wherein each of the plurality of first candidate object-tracks is associated with at least a portion of a stationary object.
6. The method of claim 1 , wherein each of the plurality of first candidate object-tracks is associated with at least a portion of a moving object.
7. The method of claim 1 , wherein the one or more remaining first candidate object-tracks comprises fewer than a total quantity of first candidate object-tracks.
8. The method of claim 1 , wherein the first set of sensors comprises one or more radar sensors, lidar sensors, and ultrasound sensors that are different than other sensors used for the second set of sensors.
9. The method of claim 1 , wherein the second set of sensors comprises one or more visual sensors that are different than other sensors used for the first set of sensors.
10. The method of claim 1 , further comprising:
outputting, to a controller for use in controlling a vehicle, an indication of the one or more remaining first candidate object-tracks that match to the set of second object-tracks.
11. The method of claim 10 , further comprising:
refraining, from outputting to the controller for use in controlling the vehicle, an indication of the at least one first candidate object-track that is pruned.
12. A system, the system comprising:
a processor configured to:
determine a plurality of first candidate object-tracks according to first sensor data obtained from a first set of sensors;
determine a set of second object-tracks according to second sensor data obtained from a second set of sensors;
for each of the plurality of first candidate object-tracks:
determining a respective matching error between said each first candidate object-track and each of the set of second object-tracks;
computing a respective weight for consideration of matching said each first candidate object-track to said each second object-tracks to be an inverse of the respective matching error between said each first candidate object-track and said each second object-track;
applying the respective weight for said each second object-track to that first candidate object-track;
determining a respective distribution of all respective weights applied to the set of second object-tracks for said each first candidate object-track; and
determining a Kurtosis value of the respective distribution for said each first candidate object-track;
pruning at least one candidate object-track from the plurality of first candidate object tracks by:
dividing the plurality of first candidate object-tracks into two subsets including a first subset of the plurality of first candidate object-tracks having higher Kurtosis values than a second subset of the plurality of first candidate object-tracks, the second subset of the plurality of first candidate object-tracks including the at least one first candidate object-track; and
discarding the second subset from the plurality of first candidate object-tracks; and
matching, by the sensor-fusion system, said each second object-track with one or more remaining first candidate object-tracks of the plurality of first candidate object-tracks after the second set is discarded and the at least one first candidate object-track is pruned.
13. The system of claim 12 , wherein the processor is configured to prune the at least one first object-track from the plurality of first object-tracks by:
discarding the second subset from the plurality of first candidate object-tracks in response to determining that the first subset has the higher Kurtosis values than the second subset.
14. The system of claim 13 , wherein the processor is further configured to:
refrain from discarding the at least one first candidate object-track for matching against a different candidate object track from the set of second object-tracks.
15. The system of claim 12 , wherein the first set of sensors comprises one or more radar sensors, lidar sensors, and ultrasound sensors and the second set of sensors comprises one or more visual sensors.
16. The system of claim 12 , wherein the system comprises a sensor-fusion system communicatively coupled to a controller of a vehicle, the processor being further configured to:
output, to the controller, an indication of the one or more remaining first candidate object-tracks for use in controlling the vehicle.
17. The system of claim 12 , wherein said each candidate object-track from the plurality of first candidate object-tracks is associated with at least a portion of a stationary or moving object.
18. The system of claim 12 , wherein the second set of sensors comprises one or more visual sensors that are different than other sensors used for the first set of sensors.
19. The system of claim 12 , wherein the processor is communicatively coupled to a controller of a vehicle, and at least one of:
the processor is further configured to output, to the controller for use in controlling the vehicle, an indication of the one or more remaining first candidate object-tracks that match to the set of second object-tracks; or
the processor is further configured to refrain from outputting, to the controller for use in controlling the vehicle, an indication of the at least one first candidate object-track that is pruned.
20. A system comprising:
means for determining a plurality of first candidate object-tracks according to first sensor data obtained from a first set of sensors;
means for determining a set of second object-tracks according to second sensor data obtained from a second set of sensors;
for each of the plurality of first candidate object-tracks, means for:
determining a respective matching error between said each first candidate object-track and each of the set of second object-tracks;
computing a respective weight for consideration of matching said each first candidate object-track to each of the set of second object-tracks to be an inverse of the respective matching error between said each first candidate object-track and said each second object-track;
applying the respective weight for said each second object-tracks to said each first candidate object-track;
determining a respective distribution of all respective weights applied to the set of second object-tracks for said each first candidate object-track; and
determining a Kurtosis value of the respective distribution for said each first candidate object-track;
means for pruning at least one first candidate object-track from the plurality of first candidate object-tracks, including means for:
dividing the plurality of first candidate object-tracks into two sub-sets including a first subset of the plurality of first candidate object-tracks having higher Kurtosis values than a second subset of the plurality of first candidate object-tracks, the second subset including the at least one first candidate object-track; and
discarding the second subset from the plurality of first candidate object-tracks; and
means for matching said each second object-track with one or more remaining first candidate object-tracks after the second subset is discarded and the at least one first candidate object-track is pruned.