Kinematic invariant-space maximum entropy tracker (KISMET)

Patent metadata

Field	Value
Publication number	US-12229216-B2
Application number	US-202217693112-A
Country	US
Kind code	B2
Filing date	Mar 11, 2022
Priority date	May 11, 2021
Publication date	Feb 18, 2025
Grant date	Feb 18, 2025

Abstract

A processor-implemented method for simultaneously tracking one or more objects includes receiving, via a dynamical system with a set of sensors, a first set of unlabeled measurements from one or more objects. Each of the measurements is a function of time. A set of candidate tracks is determined for the one or more objects. Probabilities of each of the first set of unlabeled measurements being assigned to each of the set of candidate tracks are computed. A track from the set of candidate tracks is determined for each of the one or more objects based on a joint probability distribution of track attributes and the probabilistic assignment of each of the first set of unlabeled measurements to each of the set of candidate tracks.

First claim

What is claimed is:

1. A processor-implemented method for tracking objects, the processor-implemented method being performed by at least one processor and comprising:
receiving, from a dynamical system with a first set of sensors, a first set of unlabeled measurements for one or more objects in motion for at least a portion of an observation period, each of the first set of unlabeled measurements being a function of time;
determining a set of candidate tracks for the one or more objects by identifying one or more clusters of unlabeled measurements of the first set of unlabeled measurements;
computing probabilities of each of the first set of unlabeled measurements being assigned to each of the set of candidate tracks;
determining an assignment of a candidate track from the set of candidate tracks for each of the one or more objects based on a joint probability distribution of track attributes and the computed probabilities of each of the first set of unlabeled measurements being assigned to each of the set of candidate tracks; and
tracking a position of the one or more objects, using a second set of sensors, subsequent to the observation period based at least in part on the assignment of the candidate track for each of the one or more objects.

2. The processor-implemented method of claim 1 , further comprising determining a preference of each unlabeled measurement of the first set of unlabeled measurements for each candidate track of the set of candidate tracks.

3. The processor-implemented method of claim 2 , further comprising computing the preference of each unlabeled measurement of the first set of unlabeled measurements for each candidate track of the set of candidate tracks based on a normalized residual between each unlabeled measurement of the first set of unlabeled measurements and each candidate track of the set of candidate tracks at a time of the each unlabeled measurement.

4. The processor-implemented method of claim 1 , further comprising identifying the one or more clusters of measurements based on a similarity threshold.

5. The processor-implemented method of claim 1 , in which each of the first set of unlabeled measurements is a function of dynamic invariants.

6. The processor-implemented method of claim 1 , further comprising determining the assignment of each of the first set of unlabeled measurements to one of the set of candidate tracks based on a stochastic sampling of the computed probabilities.

7. The processor-implemented method of claim 1 , in which one or more of the first set of unlabeled measurements are fused from multiple sensors of the first set of sensors.

8. The processor-implemented method of claim 1 , in which one or more of a position or an orientation of one or more sensors of the first set of sensors is changing relative to the one or more objects.

9. The processor-implemented method of claim 1 , further comprising:
receiving a history of measurement information; and
updating the assignment of the candidate track for each of the one or more objects based on the history of measurement information.

10. The processor-implemented method of claim 1 , further comprising:
receiving a second set of unlabeled measurements from the one or more objects; and
generating an inference for genealogy of one of the one or more objects based on the first set of unlabeled measurements and the second set of unlabeled measurements.

11. An apparatus for tracking objects comprising:
at least one memory; and
at least one processor coupled to the at least one memory, the at least one processor configured:
to receive, from a dynamical system with a first set of sensors, a first set of unlabeled measurements for one or more objects in motion for at least a portion of an observation period, each of the first set of unlabeled measurements being a function of time;
to determine a set of candidate tracks for the one or more objects by identifying one or more clusters of unlabeled measurements of the first set of unlabeled measurements;
to compute probabilities of each of the first set of unlabeled measurements being assigned to each of the set of candidate tracks;
to determine an assignment of a candidate track from the set of candidate tracks for each of the one or more objects based on a joint probability distribution of track attributes and the computed probabilities of each of the first set of unlabeled measurements being assigned to each of the set of candidate tracks; and
to track a position of the one or more objects, using a second set of sensors, subsequent to the observation period based at least in part on the assignment of the candidate track for each of the one or more objects.

12. The apparatus of claim 11 , in which the at least one processor is further configured to determine a preference of each unlabeled measurement of the first set of unlabeled measurements for each candidate track of the set of candidate tracks.

13. The apparatus of claim 12 , in which the at least one processor is further configured to compute the preference of each unlabeled measurement of the first set of unlabeled measurements for each candidate track of the set of candidate tracks based on a normalized residual between each unlabeled measurement of the first set of unlabeled measurements and each candidate track of the set of candidate tracks at a time of the each unlabeled measurement.

14. The apparatus of claim 11 , in which the at least one processor is further configured to identify the one or more clusters of measurements based on a similarity threshold.

15. The apparatus of claim 11 , in which each of the first set of unlabeled measurements is a function of dynamic invariants.

16. The apparatus of claim 11 , in which the at least one processor is further configured to determine the assignment of each of the first set of unlabeled measurements to one of the set of candidate tracks based on a stochastic sampling of the computed probabilities.

17. The apparatus of claim 11 , in which one or more of the first set of unlabeled measurements are fused from multiple sensors of the first set of sensors.

18. The apparatus of claim 11 , in which one or more of a position or an orientation of one or more sensors of the first set of sensors is changing relative to the one or more objects.

19. The apparatus of claim 11 , in which the at least one processor is further configured:
to receive a history of measurement information; and
to update the assignment of the candidate track for each of the one or more objects based on the history of measurement information.

20. The apparatus of claim 11 , in which the at least one processor is further configured:
to receive a second set of unlabeled measurements from the one or more objects; and
to generate an inference for genealogy of one of the one or more objects based on the first set of unlabeled measurements and the second set of unlabeled measurements.

Assignees

• Reservoir Labs Inc
• Qualcomm Inc

Inventors

• Weintraub Lawrence Craig
• Langston Matthew Harper
• Wei Julia
• Lethin Richard
• Nogoy Aimee Kristine
• Harris Mitchell
• Mountcastle Paul

Classifications

• G06F18/2325

  using vector quantisation · CPC title

• G06F18/232

  Non-hierarchical techniques · CPC title

• G06F17/18 Primary

  for evaluating statistical data {, e.g. average values, frequency distributions, probability functions, regression analysis (forecasting specially adapted for a specific administrative, business or logistic context G06Q10/04)} · CPC title

• G06F2218/00

  Aspects of pattern recognition specially adapted for signal processing · CPC title

• G06F18/2155

  characterised by the incorporation of unlabelled data, e.g. multiple instance learning [MIL], semi-supervised techniques using expectation-maximisation [EM] or naïve labelling · CPC title