Tracking device with deferred activation and propagation of passive tracks

Abstract

A tracking device is configured to estimate a track for at least one possible target and is configured to receive incoming measurements and to process measurements and tracks. The tracking device includes a storage and a computational device. The tracking device is also configured to divide all measurements into a set of considered measurements and a set of unconsidered measurements for each passive track.

Claims

1. A tracking device configured to estimate a track for at least one possible target and configured to receive incoming measurements and to process measurements and tracks, said tracking device comprising: a storage and a computational device having a computational capacity and configured with: an association module configured to calculate an association between a measurement and a track; an output module configured to output a sequence of track updates from an assignment module; an assignment module configured to maintain a set of active tracks and, using the association module to calculate associations between active tracks and the incoming measurements, the calculated associations containing possible track updates, deciding which track updates to keep in the set of active tracks and which track updates to add to a set of passive tracks, the set of passive tracks comprises tracks which are not updated when new measurements are received at the tracker; the computational device being configured to: defer computations on the set of passive tracks until at least one passive track handling criterion is fulfilled, activate at least one passive track from the set of passive tracks and transfer the at least one passive track from the set of passive tracks to the set of active tracks by at least one of the computations, the tracking device further configured to divide all measurements into a set of considered measurements and a set of unconsidered set of measurements for each passive track; and comprising: a propagator module configured to propagate by the computations on the set of passive tracks, a selected passive track using the association module to calculate an association as a function of an unconsidered measurement and the selected passive track; adding the possible track update of the selected passive track and the unconsidered measurement to the set of passive tracks, and marking the measurement considered for both the selected passive track and the possible track update.

2. The tracking device according to claim 1, wherein the passive track handling criterion is a function of available computational capacity and/or a function of track probability or track likelihood.

3. The tracking device according to claim 1, wherein computations on the set of passive tracks are deferred or postponed for up to 25 frames.

4. The tracking device according to claim 1, wherein a division of considered measurements and unconsidered measurements is performed by using an ordered identifier of incoming measurements and storing the ordered identifier of the latest considered measurement on each passive track, and wherein the track propagator is-configured to propagate a selected passive track according to the ordered identifier, advancing the ordered identifier of the latest considered measurement of the selected passive track.

5. The tracking device according to claim 1, configured to propagate passive tracks as a function of available computational capacity.

6. The tracking device according to claim 1, configured to activate passive tracks as a function of available computational capacity.

7. The tracking device according to claim 1, configured to passivate an active track and transfer the active track from the set of active tracks to the set of passive tracks.

8. The tracking device according to claim 1, configured to prioritize the computations on the set of passive tracks based on a function of track likelihood or track probability.

9. The tracking device according to claim 1, configured to prioritize propagation of passive tracks based on a function of track likelihood, L(τ).

10. The tracking device according to claim 1, configured to prioritize propagation of passive tracks based on a function of an estimate of track probability, P(τ).

11. The tracking device according to claim 1, wherein the assignment module is based on one or more of the following assignments algorithms: A. a Nearest Neighbor algorithm; B. a 2-dimensional assignment algorithm; C. an N-dimensional assignment algorithm; D. a K-best multi hypothesis assignment; or E. a K-best N-dimensional assignment.

12. The tracking device according to claim 11, selections A or B, wherein activating tracks is performed by estimating the best hypothesis considering the set of active tracks and a subset of the set of passive tracks, and activating tracks belonging to the hypothesis and passivating tracks not belonging to the hypothesis.

13. The tracking device according claim 1, wherein the best hypothesis among a set of tracks is estimated by repeatedly: selecting the track with the highest estimate of track probability, P(τ) disregarding all conflicting tracks.

14. The tracking device according to claim 10, wherein the estimate of track probability P(τ) is estimated as: $P\left(\tau \right)\simeq P_0\left(\tau \right)\equiv \frac{L\left(\tau \right)}{L_0+\underset{{\tau }^{\prime }\in t}{.Math.}L\left({\tau }^{\prime }\right)}$ where t is the target for which τ is an possible track.

15. The tracking device according to claim 10, wherein the estimate of track probability, P(τ) is estimated by an approximation: $P\left(\tau \right)\approx P_{\mathrm{eff}}\left(\tau \right)\equiv \frac{L\left(\tau \right)X\left(\tau \right)}{L_0+\underset{{\tau }^{\prime }\in t}{.Math.}L\left({\tau }^{\prime }\right)X\left({\tau }^{\prime }\right)}$ $\mathrm{where}\text{:}$ $X\left(\tau \right)\equiv \underset{\mathrm{other}\mathrm{targets},t}{\min }{X}_t\left(\tau \right)$ $\mathrm{where}\text{:}$ $X_t\left(\tau \right)\equiv 1-\underset{{\tau }^{\prime }\in t,\tau .Math.\tau \prime \ne \varnothing }{.Math.}{P}_0\left({\tau }^{\prime }\right)$ $\mathrm{where}\text{:}$ $P_0\left(\tau \right)\equiv \frac{L\left(\tau \right)}{L_0+\underset{{\tau }^{\prime }\in t}{.Math.}L\left({\tau }^{\prime }\right)}$ where t is the target for which τ is a possible track.

16. A method of tracking comprising processing a measurement and a track, said method of tracking comprising: receiving a measurement; associating a measurement with a track by calculating an association between a measurement and a track; assigning tracks by maintaining a set of active tracks by associating and extracting possible track updates and deciding which track updates to keep in the set of active tracks and which track updates to add to a set of passive tracks, the set of passive tracks comprises tracks which are not updated when new measurements are received at the tracker; defer computations on the set of passive tracks until at least one passive track handling criterion is fulfilled, wherein at least one of the computations on the set of passive tracks activate a passive track from the set of passive tracks and transfer the passive track from the set of passive tracks to the set of active tracks; dividing all measurements into a set of considered measurements and a set of unconsidered set of measurements for each passive track; propagating by the computations on the set of passive tracks, a selected passive track by calculating an association as a function of an unconsidered measurement and the selected passive track, adding the possible track update of the selected passive track and the unconsidered measurement to the set of passive tracks, and marking the measurement considered for both the selected passive track and the possible track update; and outputting a track update.

17. The method of tracking according to claim 16, further comprising: dividing all measurements into a set of considered measurements and a set of unconsidered measurements for each passive track; and propagating a selected passive track by associating by calculating an association as a function of an unconsidered measurement and the selected passive track and adding the possible track update of the selected passive track and the unconsidered measurement to the set of passive tracks, marking the measurement considered for both the selected passive track and the possible track update.

18. The method of tracking according to claim 16, wherein propagating of passive tracks is based on a function of track likelihood, L(τ), or on a function of an estimate of track probability, P(τ).

19. The method of tracking according to claim 16, wherein assigning is based on one or more of the following assignments algorithms: A. a Nearest Neighbor algorithm; B. a 2-dimensional assignment algorithm; C. an N-dimensional assignment algorithm; D. a K-best multi hypothesis assignment; or E. a K-best N-dimensional assignment.

20. The method of tracking according to claim 18, wherein propagating is based on an estimate of track probability P(τ) estimated as: $P\left(\tau \right)\simeq P_0\left(\tau \right)\equiv \frac{L\left(\tau \right)}{L_0+\underset{{\tau }^{\prime }\in t}{.Math.}L\left({\tau }^{\prime }\right)}$ where t is the target for which τ is an possible track; or estimated by an approximation: $P\left(\tau \right)\approx P_{\mathrm{eff}}\left(\tau \right)\equiv \frac{L\left(\tau \right)X\left(\tau \right)}{L_0+\underset{{\tau }^{\prime }\in t}{.Math.}L\left({\tau }^{\prime }\right)X\left({\tau }^{\prime }\right)}$ $\mathrm{where}\text{:}$ $X\left(\tau \right)\equiv \underset{\mathrm{other}\mathrm{targets},t}{\min }{X}_t\left(\tau \right)$ $\mathrm{where}\text{:}$ $X_t\left(\tau \right)\equiv 1-\underset{{\tau }^{\prime }\in t,\tau .Math.\tau \prime \ne \varnothing }{.Math.}{P}_0\left({\tau }^{\prime }\right).\mathrm{where}\text{:}$ $P_0\left(\tau \right)\equiv \frac{L\left(\tau \right)}{L_0+\underset{{\tau }^{\prime }\in t}{.Math.}L\left({\tau }^{\prime }\right)}$ where t is the target for which τ is a possible track.

Description

BRIEF DESCRIPTION OF DRAWINGS

(1) The embodiments will now be described more fully hereinafter with reference to the accompanying drawings, in which exemplary embodiments are shown. The claimed Invention may, however, be embodied in different forms and should not be construed as limited to the embodiments set forth herein. Embodiments of the invention will be described in the figures, wherein:

(2) FIG. 1 illustrates an embodiment of a tracking device with a set of active tracks and a set of passive tracks;

(3) FIG. 2 illustrates an embodiment of a tracking device with a propagator module;

(4) FIG. 3 illustrates division of considered measurements and unconsidered measurements;

(5) FIG. 4 illustrates an embodiment of a tracking device with a propagator and receiving module;

(6) FIG. 5 illustrates a method of tracking;

(7) FIG. 6 illustrates a method of tracking including propagating;

(8) FIG. 7 illustrates definitions of targets, measurements, and tracks and track updates;

(9) FIG. 8 illustrates definitions of targets, measurements, and tracks of active and passive sets of measurements; and

(10) FIG. 9 to 15 illustrates a temporal evolution of tracking.

DETAILED DESCRIPTION OF THE DRAWINGS

(11) TABLE-US-00001 Item No Tracking device 1 Measurement 2 Track 4 Track update 5 Target 6 Sensor 8 Process 10 Computational device 12 Computational capacity 14 Receiving module 20 Receiving time 22 Storage 23 Track likelihood 25 Estimate of track probability 26 Conflicting tracks 27 Non-conflicting tracks 28 Association module 30 Associability 31 Association 32 Assignment module 40 Maintaining 41 Keep 42 Add 43 Output Module 50 Measurement container 70 Pruning Module 80 Hypothesis 90 Set of active tracks 100 Active track 102 Set of passive tracks 110 Division of measurements 111 Passive track 112 Considered measurement 113 Unconsidered measurement 114 Ordered identifier 115 Latest considered measurement 116 Selected passive track 117 Advance 118 Activate 120 Passivate 130 Propagator Module 150 Propagate 152 Receiving 200 Method of tracking 222 Associating 300 Assigning 400 Outputting 500 Propagating 600 Dividing 610 Pruning 700

(12) FIG. 1 illustrates a tracking device 1 configured to estimate a track 4 for at least one possible target 6 and configured to receive or obtain incoming measurements 2 and to process 10 measurements 2 and tracks 4. The tracking device 1 provides a track update 5 (not shown).

(13) The measurement 2 may be an observation and may be obtained from a sensor 8 (not shown).

(14) The tracking device 1 is configured with a storage 23 (not shown) and a computational device 12 (not shown) having a computational capacity 14 (not shown). The computational device 12 may be a computer with one or more processing units in in the form of central processing unit(s) (CPU), graphics processing unit(s) (GPU) or a field-programmable gate array(s) (FPGA). It is understood that the computational device has a limited computational capacity 14.

(15) The tracking device 1 may have an output module 50 configured to output a sequence of track updates 5.

(16) The tracking device 1 may have an association module 30 configured to calculate an association 32 between a measurement 2 and a track 4.

(17) In the illustrated embodiment the output module 50 receives the track update 4 from an assignment module 40. The assignment module 40 interacts with the association module 30. The assignment module 40 provides a measurement 2 and an active track 102, possibly paired. The association module 30 provides an association 32.

(18) In this embodiment the assignment module 40 is configured to maintaining 41 a set of active tracks 100 using the association module 30 as a function of active tracks 102 and the incoming measurements 2 to calculate or estimate associations 32, containing possible track updates 5 and deciding which track updates 5 to keep 42 (not shown) or maintain in the set of active tracks 100 and which track updates 5 to add 43 to or to place in a set of passive tracks 110.

(19) In this embodiment, the tracking device 1 is further configured to activate 120 a passive track 112 from the set of passive tracks 110 and transfer the passive track 112 from the set of passive tracks 110 to the set of active tracks 100.

(20) FIG. 2 illustrates in continuation of FIG. 1 an embodiment where a tracking device 1 configured with a propagator module 150 configured to propagate 152 a selected passive track 117 using the association module 30. As will be illustrated in FIG. 3 the tracking device 1 is further configured to divide 111 (not shown) all measurements 2 into a set of considered measurements 113 (not shown) and a set of unconsidered measurements 114 for each passive track 112.

(21) The propagator module 150 is in this embodiment configured to propagate 152 the selected passive track 117 using the association module 30 to calculate an association 32 as a function of an unconsidered measurement 114 and the selected passive track 117. The propagator module 150 then adds the possible track update 5 of the selected passive track 117 and the unconsidered measurement 114 to the set of passive tracks 110. The propagator module 150 then marks the measurement considered 113 for both the selected passive track 117 and possible track update 5.

(22) FIG. 3 illustrates division of measurement 111 between considered measurements 113 and unconsidered measurements 114. For each passive track 112 associations 32 have been calculated between a passive track 112 and each considered measurements 113. The remaining of the measurements 2 are unconsidered 114 with respect to the passive track 112.

(23) In this embodiment division 111 of considered measurements 113 and unconsidered measurements 114 is performed by using an ordered identifier 115 of incoming measurements 2 and storing the ordered identifier 115 of the latest considered measurement 116 on each passive track 112.

(24) This is understood in relation to FIG. 2 wherein the track propagator 150 is configured to propagate 152 a selected passive track 117 according to the ordered identifier 115 and advancing 118 the ordered identifier 115 of the latest considered measurement 116 of the selected passive track 117.

(25) FIG. 4 illustrates in continuation of the previous figures a tracking device 1 further configured with a receiving module 20.

(26) The receiving module 20 may be configured to prepare measurements 20 for processing. The receiving module 20 may further be configured to assign a receiving time 22 (not shown) or generate an ordered identifier 115 (not shown).

(27) Furthermore the tracking devices 1 illustrated may be configured with a storage 23 (not shown), and a measurement container 70 (not shown). Finally, the tracking device may trivially be configured to eliminate obsolete data records using a pruning module 90 (not shown) or equivalent.

(28) The shown tracking devices 1 may be obtained by implementing the outlined functionalities in a different order or configuration.

(29) In this embodiment, the propagation module 150 commands the association module 30 to calculate an association 32 between the selected passive track 117 τ and an unconsidered measurement 114 such that this measurement becomes considered 113 (not shown).

(30) The association 32 might contain a track update 5, τ′=τ∪{m}, which is inserted into the set of passive tracks 100. For all considered measurements 117 the original τ is also considered for new track τ′.

(31) FIG. 5 illustrates a method of tracking 222. Features from the previous figures are implemented or programmed based on the steps and descriptions disclosed herein.

(32) The method comprises receiving a measurement 2 (not shown), which may be from a sensor 8 (not shown) and to process a measurement 2 and a track 4 (not shown).

(33) The method of tracking 222 encompasses receiving 200 a measurement 2. The method of tracking encompasses associating 300 a measurement 2 with a track 4 by calculating an association 32 (not shown) between a measurement 2 and a track 4.

(34) The method of tracking 222 includes assigning 400 tracks 4 by maintaining 41 (not shown) a set of active tracks 100 (not shown) by associating 300 and extracting possible track updates 5 (not shown) and deciding which track updates 5 to keep 42 (not shown) in the set of active tracks 100 and which track updates 5 to add 43 (not shown) to a set of passive tracks 110 (not shown) and further encompassing functionality to activate 120 (not shown) a passive track 112 (not shown) from the set of passive tracks 110 and activate 120 (not shown) the passive track 112 from the set of passive tracks 110 to the set of active tracks 100.

(35) Assigning 400 and associating 300 are interrelated in the same way as the assignment module 40 and association module 30 illustrated in FIG. 1.

(36) The method of tracking 222 encompasses outputting 500 a track update 5.

(37) FIG. 6 illustrates in continuation of FIG. 5 a method of tracking 222. Again reference is made to disclosures herein or previous figures. In particular reference is made to features described in FIG. 2.

(38) The method of tracking 222 includes dividing 610 all measurements 2 (not shown) into a set of considered measurements 113 (not shown) and a set of unconsidered measurements 114 (not shown) for each passive track 112 (not shown).

(39) The method of tracking 222 includes propagating 600 a selected passive track 117 (not shown) by associating 300 by calculating an association 32 (not shown) as a function of an unconsidered measurement 114 and the selected passive track 117 and adding the possible track update 5 (not shown) of the selected passive track 117 and the unconsidered measurement 114 to the set of passive tracks 110, marking the measurement considered 113 for both the selected passive track 117 and the possible track update 5.

(40) For both embodiments of FIGS. 5 and 6 details about assigning 400 can be found in the description or summary regarding the assignment module 40. Likewise, details about associating 300 can be found in the description or summary regarding the association module 30.

(41) FIG. 7 illustrates track updates 5 of measurements 2 using an assignment module based on a 2-D assignment. In this example two targets 6, target I and target II, are tracked. Because 2-D assignment cannot have multiple hypotheses there can only be one track 4 per target 6.

(42) In the top figure two active tracks 102 and a single new incoming measurement 2 are shown.

(43) In the bottom figure, two associations 32 between the two active tracks 102 and the incoming measurement 2 are shown. The upper has the highest associability 31 (not shown) and that track 4 is kept 42 (not shown). or remains, in the set of active tracks 100 (not shown).

(44) FIG. 8 illustrates in the top part in continuation of FIG. 7 two active tracks 102 as the currently most likely representation of target I and target II, respectively.

(45) In the top part, the set of active tracks 100 consists of two active tracks 102—one for each target 6—after processing the single incoming measurement 2 is shown.

(46) In the bottom part, the set of passive tracks 110 is updated as a track update 5 with the shown passive tracks 112. The update is performed by the assignment module 40 (not shown) in the tracking device 1 (not shown) or by assigning 400 (not shown) in the method of tracking 222 (not shown).

(47) The FIGS. 9 to 15 describe a process of an embodiment in an implementation using a 2-D assignment implementation in the assignment module 40 (not shown) exemplified with measurements originating from a scan-based radar sensor with a fixed scan period.

(48) Each figure illustrates the situation with the logarithm of the track likelihood 25 (Log L) versus time (or radar period) plot (top) and the same situation in a position versus time plot (bottom). All figures illustrate the same features, which features are illustrated in FIG. 9 (active tracks 102) and FIG. 10 (passive track 112) and remain self-explanatory here from.

(49) FIG. 9 illustrates the initial tracking with one target 6 and one track 4, τ.sub.1={ . . . m.sub.0, m.sub.1}. Thus, the set of active tracks 100 is one active track 102, τ.sub.1.

(50) In FIG. 10 there are two incoming measurements 2, m.sub.2 and m.sub.3. The association module 30 is used to calculate associations 32 (not shown) and two track updates 5, τ.sub.2=τ.sub.1∪{m.sub.2}={ . . . m.sub.0, m.sub.1, m.sub.2} and τ.sub.3=τ.sub.1∪{m.sub.3}={ . . . m.sub.0, m.sub.1, m.sub.3}. Since τ.sub.1 has a probability of detection P.sub.D(τ.sub.1), the track likelihood 25 L(τ.sub.1) is multiplied with a factor of (1−P.sub.D(τ.sub.1)) as it has not been updated.

(51) The assignment module 40 (not shown) or the step of assigning 400 (not shown) then considers tracks 4, τ.sub.1, τ.sub.2 and τ.sub.3, and keeps τ.sub.2 since it has the highest associability 31, here calculated as the log L(τ.sub.2)−log L(τ.sub.1), where L(τ.sub.1) is the likelihood 25 of a track 4 τ. Thus the set of active tracks 100 (not shown) contains one active track (102), τ.sub.2. τ.sub.1 and τ.sub.3 are added 43 (not shown) to the set of passive tracks 110 (not shown) as passive tracks 112.

(52) Continuing from FIGS. 9 and 10, in FIG. 11 one incoming measurement 2, m.sub.4, is received or obtained. The association module 30 is used to calculated the association 32 between the single active track 102, τ.sub.2, and m.sub.4. The associability 31 is too small because the distance between the expected position of τ.sub.2 and m.sub.4 is too large, therefore there is no track update 5 between τ.sub.2 and m.sub.4. Therefore the track likelihood 25 L(τ.sub.2) is multiplied with a factor of (1−P.sub.D(τ.sub.2)) leading to a decreased log L(τ.sub.2). At this point the probability estimate of τ.sub.3 is still too low given the computational resources for τ.sub.3 to be selected for propagation 600.

(53) In FIG. 12 one incoming measurement 2, m.sub.5, is received. The association module 30 calculates the association 32 between the single active track 102, τ.sub.2, and m.sub.5. The associability 31 is too small because the distance between the expected position of τ.sub.2 and m.sub.5 is too large, therefore there is no track update 5 between τ.sub.2 and m.sub.4. Therefore, the track likelihood 25 L(τ.sub.2) is multiplied with a factor of (1−P.sub.D(τ.sub.2))

(54) At this point the estimate of track probability 26 of τ.sub.3 has grown due to decrease in L(τ.sub.2), and therefore there is now computational resources available for the passive track 112 τ.sub.3 to be selected 117 for propagation 600.

(55) In FIG. 13 the passive track 112, τ.sub.3, is propagated 152 by the propagation module 150. The association module 30 calculates the association 32 between τ.sub.3 and the unconsidered measurement 114, m.sub.4. The associability 31 is large enough to generate a track update 5, τ.sub.4=.sub.3∪{m.sub.4}={ . . . m.sub.0, m.sub.1, m.sub.3, m.sub.4}. Again the likelihood L(τ.sub.3) is lowered by a factor (1−P.sub.D(τ.sub.3)) because track 4 τ.sub.3 is not updated with a measurement 2.

(56) As this stage both the track likelihood 25 and estimate of track probability 26 of track 4 τ.sub.2 exceed the track likelihood 25 and estimate of track probability 26 of track 4 τ.sub.4.

(57) In FIG. 14 the passive track 112, τ.sub.4, is propagated 152 by the propagation module 150. The association module 30 is used to calculate the association 32 between τ.sub.4 and the unconsidered measurement 114 m.sub.5. The associability 31 is large enough to generate a track update 5, τ.sub.5=τ.sub.4∪{m.sub.5}={ . . . m.sub.0, m.sub.1, m.sub.3, m.sub.4, m.sub.5}. Again the likelihood L(τ.sub.4) is lowered by a factor (1−P.sub.D(τ.sub.4)) because track 4 τ.sub.3 is not updated with a measurement 2.

(58) Now both the track likelihood 25 and the estimate of track probability 26 of track 4 τ.sub.2 are lower than the track likelihood 25 and estimate of track probability 26 of track 4 τ.sub.5.

(59) Therefore and importantly as seen in FIG. 15, the passive track 112 τ.sub.5 is activated 120 and moved from the set of passive tracks 110 to the set of active tracks 100.

(60) Because τ.sub.2 and τ.sub.5 are conflicting tracks 27 (not shown), and 2-D assignment only can handle non-conflicting tracks 28 (not shown) In the set of active tracks 100, track 4 τ.sub.2 is passivated 130 and moved to the set of passive tracks 110.

(61) Although particular embodiments have been shown and described, it will be understood that it is not intended to limit the claimed inventions to the preferred embodiments, and it will be obvious to those skilled in the art that various changes and modifications may be made without departing from the spirit and scope of the claimed inventions. The specification and drawings are, accordingly, to be regarded in an illustrative rather than restrictive sense. The claimed inventions are intended to cover alternatives, modifications, and equivalents.