Systems and methods for temporal decorrelation of object detections for probabilistic filtering
Systems and methods for tracking an object. The method comprising: receiving, by a processor, a series of observations made over time for the object; selecting, by the processor, a plurality of sets of observations using the series of observations; causing, by the processor, the plurality of sets of observations to be used by at least one filter to generate a track for the object (wherein the at least one filter uses sensor data associated with each of a plurality of frames of sensor data only once during generation of the track); and causing, by the processor, operations of an autonomous robot to be controlled based on the track for the object.
1. A method for tracking an object, comprising:
receiving, by a processor, a series of observations made over time for the object;
selecting, by the processor, a plurality of sets of observations using the series of observations for a plurality of filters, wherein at least two sets of observations among the plurality of sets of observations comprise at least one observation in common;
causing, by the processor, the plurality of sets of observations to be used by the plurality of filters to generate a track for the object, wherein the plurality of filters uses sensor data associated with each frame of a plurality of frames of sensor data only once during generation of the track; and
causing, by the processor, operations of an autonomous robot to be controlled based on the track for the object.
2. The method according to claim 1 , wherein each observation of said series of observations is defined by a cuboid comprising LiDAR data points.
3. The method according to claim 1 , wherein each observation of said series of observations is associated with a single frame of the plurality of frames of sensor data.
4. The method according to claim 1 , wherein at least one filter of the plurality of filters comprises a Kalman filter.
5. The method according to claim 1 , further comprising causing the plurality of sets of observations to be used by a first filter and a second filter to generate the track for the object, wherein the plurality of filters comprises the first and second filters and each of the first and second filters uses sensor data associated with each of the plurality of frames of sensor data only once during generation of the track.
6. The method according to claim 5 , wherein the track comprises a sequence of object states sequentially generated by the first and second filters using respective ones of the plurality of sets of observations.
7. The method according to claim 1 , wherein said plurality of sets of observations are selected from the series of observations to ensure that the plurality of filters uses sensor data associated with each of a plurality of frames of sensor data only once during generation of the track.
8. The method according to claim 1 , wherein the autonomous robot comprises an autonomous vehicle.
9. A system, comprising:
a processor; and
a non-transitory computer-readable storage medium comprising programming instructions that are configured to cause the processor to implement a method for controlling an autonomous robot, wherein the programming instructions comprise instructions to:
receive a series of observations made over time for an object;
select a plurality of sets of observations using the series of observations for a plurality of filters, wherein at least two sets of observations among the plurality of sets of observations comprise at least one observation in common;
cause the plurality of sets of observations to be used by the plurality of filters to generate a track for the object, wherein the plurality of filters uses sensor data associated with each frame of a plurality of frames of sensor data only once during generation of the track; and
cause operations of the autonomous robot to be controlled based on the track for the object.
10. The system according to claim 9 , wherein each observation of said series of observations is defined by a cuboid comprising LiDAR data points.
11. The system according to claim 9 , wherein each observation of said series of observations is associated with a single frame of the plurality of frames of sensor data.
12. The system according to claim 9 , wherein at least filter of the plurality of filters comprises a Kalman filter.
13. The system according to claim 9 , wherein the programming instructions further comprise instructions to cause the plurality of sets of observations to be used by a first filter and a second filter to generate the track for the object, wherein said plurality of filters comprises said first and second filters and each of the first and second filters uses sensor data associated with each of the plurality of frames of sensor data only once during generation of the track.
14. The system according to claim 13 , wherein the track comprises a sequence of object states sequentially generated by the first and second filters using respective ones of the plurality of sets of observations.
15. The system according to claim 9 , wherein said plurality of sets of observations are selected from the series of observations to ensure that the plurality of filters uses sensor data associated with each of a plurality of frames of sensor data only once during generation of the track.
16. A computer program product comprising a memory and programming instructions that are configured to cause a processor to:
receive a series of observations made over time for an object;
select a plurality of sets of observations using the series of observations for a plurality of filters, wherein at least two sets of observations among the plurality of sets of observations comprise at least one observation in common;
cause the plurality of sets of observations to be used by the plurality of filters to generate a track for the object, wherein the plurality of filters uses sensor data associated with each of a plurality of frames of sensor data only once during generation of the track; and
cause operations of an autonomous robot to be controlled based on the track for the object.