JP5854926B2 - Falling motion judgment device - Google Patents

Description

  The present invention relates to a falling motion determination device that detects that a target is in a falling motion.

  In a sensor system such as a radar, the presence / absence of a target is determined from observations along a time series obtained from a sensor that observes the position of the target, and if the target exists, the position and speed of the target are estimated. In some cases, it is necessary to estimate the motion specifications of the target by the tracking process. Further, it may be necessary to determine a change in the motion state of the target from the estimated motion specifications.

  Techniques for tracking a target using observation values obtained by observing the target with a sensor such as a radar have been taken up in many papers and patent documents, and there are various devices and methods for realizing them. Proposals have been made.

  Here, for example, when a failure occurs in an aircraft that has been tracked, the aircraft may fall, or may turn into a falling motion at a point in time when there is an aircraft that has been moving straight at a constant speed. By quickly and accurately detecting the change from the falling motion state and the straight motion state to the falling motion state from the estimated value of the aircraft motion specifications, it is possible to quickly cope with the falling aircraft.

The determination of the falling motion requires accuracy, and the target may perform various motions such as a traveling direction acceleration motion, a lateral turning motion, and a downward turning motion in addition to the constant-velocity linear motion. It is necessary to correctly detect only the falling motion without misjudging these motions as falling motion.
For example, Non-Patent Document 1 discloses a method for determining from the residual calculated by the tracking process that the movement of the target has changed from a straight movement to another movement such as a turn.

Y. Bar-Shalom, X. et al. Long Li and T.R. Kirubarajan “Estimation with Applications to Tracking and Navigation”, John Wiley & Sons, 2001. pp427-431

A drop determination system that applies the method described in Non-Patent Document 1 to drop determination includes a sensor, a tracking filter unit, and an acceleration estimation unit. The tracking filter section uses a tracking filter based on a motion model of a constant-velocity linear motion state, estimates the motion parameters of the target from the observed values of the target position of the sensor, and performs motion parameter estimation processing at each sampling time To calculate the residual.
Here, the residual is a vector of the difference between the predicted position and the observed position when it is assumed that the target is moving straight at a constant speed, and this residual vector is input to the acceleration estimation unit.

The acceleration estimating unit calculates an acceleration estimated value by a least square method by accumulating residuals in the sliding window. This sliding window is data for n consecutive sampling times including the latest sampling time, which is obtained by shifting the time forward one by one every time one sampling time elapses. The acceleration estimation unit determines that the target is falling when the downward component of the average acceleration is equivalent to the gravitational acceleration (1G, 9.8 m / s ² ) within the estimation error range.

  Here, the observed value of the sensor does not accurately reflect the position of the target, and involves some error. When this observation error is large, it is difficult to distinguish whether the generated residual is caused by a drop motion or an observation error. In such a case, it is necessary to determine the falling motion by extending the sliding window range and removing random components due to observation errors.

However, in the method described in Non-Patent Document 1, for example, the acceleration estimated value calculated by the acceleration estimating unit is an average value of acceleration within the sliding window. Therefore, for example, even when the target trajectory 44 based on the observation value 43 starts to fall in the middle of the sliding window, such as the time t1 sliding window 41 shown in FIGS. The corresponding acceleration estimated value 45 is not calculated. In order to calculate the acceleration estimated value 45 corresponding to the gravitational acceleration by this method, it is necessary that the target is in a falling motion at all sampling times in the sliding window as in the time t2 sliding window 42. .
However, when the sliding window is taken long, it takes time until all the movements of the target object in the sliding window become the falling movement state. That is, there has been a problem that a large delay occurs between the time when the target object actually starts the drop motion and the detection of the drop motion.

  In addition, as described above, the conventional method for detecting the falling motion is to determine the falling motion only by the average value of the acceleration in the sliding window, so that the motion other than the falling motion, for example, the downward turning motion is erroneously dropped. There was a problem that it was easy to make an erroneous determination as an exercise.

  The present invention has been made to solve the above-described problems, minimizes the detection deviation time from when the target object starts the falling motion to the detection of the falling motion state, and exercises other than the falling motion. It is an object of the present invention to provide a falling motion determination device that can prevent erroneous determination of a falling motion.

In order to achieve the above object, a falling motion determination device according to the present invention tracks an object using an observed value of the position of the object obtained from a sensor, and uses information and an observed value obtained from the tracking. And a tracking filter unit for estimating the motion specifications of the target based on the observation value obtained from the sensor in the falling motion determination device for determining the falling motion of the target and the change from the constant-velocity linear motion state to the falling motion state an acceleration estimating unit that calculates an average acceleration of the target within the sliding window observed value information is added from the tracking filter unit is reflected, 2 divides the scan riding window straight period and falling period of the target and a sliding window dividing unit for, sliding window division unit length cubed value of the sliding window before dividing , A downward component of acceleration, based on the gravitational acceleration, the third power value the length of the sliding window of the fall period side from separated division is calculated, obtaining the length of the sliding window of the fall period side from the value Is.

Further, the falling motion determination device according to the present invention tracks the target using the observed value of the position of the target obtained from the sensor, and uses the information obtained from the tracking and the observed value of the sensor to target the target. And a tracking filter unit for estimating a motion specification of a target based on an observation value obtained from a sensor in a falling motion determination apparatus for determining a falling motion and a change from a constant-velocity linear motion state to a falling motion state an acceleration estimating unit that calculates an average acceleration of the target within the sliding window observed value information is added from the parts are reflected, a sliding window divided into two scan riding window straight period and falling period of the target a dividing unit, calculates the likelihood of falling motion by using the observed values, final and falling motion likelihood determination unit which target to determine whether to fall The sliding window dividing unit calculates the length of the sliding window on the falling period side from the division boundary based on the value obtained by cubeing the length of the sliding window before dividing, the downward component of acceleration, and the gravitational acceleration. Is calculated to obtain the length of the sliding window on the drop period side .

Further, the falling motion determination device according to the present invention tracks the target using the observed value of the position of the target obtained from the sensor, and uses the information obtained from the tracking and the observed value of the sensor to target the target. And a tracking filter unit for estimating a motion specification of a target based on an observation value obtained from a sensor in a falling motion determination apparatus for determining a falling motion and a change from a constant-velocity linear motion state to a falling motion state an acceleration estimating unit that calculates an average acceleration of the target within the sliding window observed value information is added from the parts are reflected, a sliding window divided into two scan riding window straight period and falling period of the target a division unit, to calculate the likelihood of falling motion for each of the finite number of falling motion candidates based on observations, most using the sum of the likelihood To a falling motion adaptation judgment unit which target to determine whether to fall, sliding window dividing section, a third power value the length of the previous sliding window split, downward component of acceleration Then, based on the gravitational acceleration, a value obtained by calculating the cube of the length of the sliding window on the falling period side from the division break is calculated, and the length of the sliding window on the falling period side is obtained from the value .

  According to the present invention, it is possible to minimize the detection deviation time from when the target object starts the falling motion to when the falling motion is detected, and erroneously determines that the motion state other than the falling motion is the falling motion. Can be prevented.

It is a block block diagram of the fall movement determination apparatus which concerns on Embodiment 1 of this invention. It is a flowchart which shows the procedure of the process for 1 sampling time of the fall movement determination apparatus which concerns on Embodiment 1 of this invention. It is a figure which shows the example of a division | segmentation of the sliding window by the fall movement determination apparatus which concerns on Embodiment 1 of this invention. It is a figure explaining the division | segmentation of the sliding window of the falling motion determination apparatus which concerns on Embodiment 1 of this invention. It is a block diagram which shows the structure of the fall movement determination apparatus which concerns on Embodiment 2 of this invention. It is a flowchart which shows the procedure of the process for 1 sampling time of the fall movement determination apparatus which concerns on Embodiment 2 of this invention. It is a figure which shows the example of likelihood calculation of the fall motion of the fall motion determination apparatus which concerns on Embodiment 2 of this invention. It is a flowchart which shows the procedure of the other process for 1 sampling time of the fall movement determination apparatus which concerns on Embodiment 2 of this invention. It is a block block diagram of the fall movement determination apparatus which concerns on Embodiment 3 of this invention. It is a flowchart which shows the procedure of the process for 1 sampling time of the fall movement determination apparatus which concerns on Embodiment 3 of this invention. It is a figure which shows the example of the movement model of the target of the falling movement determination apparatus which concerns on Embodiment 3 of this invention. It is a figure which shows the example of likelihood calculation of each exercise | movement of the fall exercise | movement determination apparatus which concerns on Embodiment 3 of this invention. It is the figure shown about the sliding window and the estimated acceleration of the conventional fall motion determination apparatus.

Embodiment 1 FIG.
A falling motion determination device according to Embodiment 1 will be described with reference to the drawings.
FIG. 1 is a block configuration diagram of a falling motion determination device 10 according to Embodiment 1 of the present invention. The falling motion determination device 10 is connected to the sensor 4 as shown in FIG. In addition, a tracking filter unit 3 is provided that estimates the motion specifications of the target based on the observation values obtained from the sensor 4. In addition, an acceleration estimation unit 1 is provided that has a sliding window that reflects the observation value to which the information from the tracking filter unit 3 is added, and calculates the acceleration of the target. In addition, a sliding window dividing unit 2 that divides the sliding window into two is provided.

The tracking filter unit 3 includes a tracking filter based on a motion model of constant-velocity linear motion, and from the observation value of the position of the target input from the sensor 4, the motion of the target is detected at each sampling time using the tracking filter. The residual is calculated while estimating the element.
Here, the residual is a difference vector between the predicted position and the observed position when it is assumed that the target is moving straight at a constant speed.

The acceleration estimation unit 1 has a sliding window in which observation values and residuals for n sampling times are reflected, and the acceleration estimation value in the sliding window of the target is calculated by the least squares method by accumulating the residuals in the sliding window. Is calculated by the following.
The sliding window dividing unit 2 divides the sliding window into a straight running period sliding window and a falling period sliding window based on the acceleration estimated value calculated by the acceleration estimating unit 1 and the length of the sliding window, and the acceleration estimating unit 1 Output.

Next, the process procedure of the fall motion determination by the fall motion determination device 10 will be described. FIG. 2 is a flowchart showing a processing procedure for one sampling time of the falling motion determination device 10 according to the first embodiment of the present invention.
As shown in FIG. 2, first, the tracking filter unit 3 estimates the motion specifications of the target from the position information of the target input from the sensor 4 and calculates a residual (step ST11).

  Next, in the acceleration estimation step ST12, the acceleration estimation unit 1 receives the observed value and the residual from the tracking filter unit 3, adds them to the sliding window, and calculates the estimated acceleration value in the sliding window (step ST12).

If the calculation result of the acceleration estimation step ST12 is not a significant acceleration as shown in step ST13, the acceleration estimation unit 1 determines that “the target is not falling” and ends the process. On the other hand, when the calculation result of the acceleration estimation step ST12 is a significant acceleration estimation value as shown in step ST13, the acceleration estimation unit 1 corresponds to the gravitational acceleration within the range of the estimation error. (Step ST14).
Then, as shown in step ST14, when the acceleration estimated value corresponds to the gravitational acceleration, the acceleration estimating unit 1 determines that “the target is falling” and ends the process. On the other hand, as shown in step ST14, when the estimated acceleration does not correspond to the gravitational acceleration, the process proceeds to the next sliding window division step ST15.

In the sliding window dividing step ST15, the sliding window dividing unit 2 divides the sliding window into a straight running period sliding window and a falling period sliding window based on the acceleration estimated value calculated in step ST12 and the length of the sliding window ( Step ST15).
FIG. 3 is a diagram showing an example of the sliding window division by the falling motion determination device 10 according to the first embodiment of the present invention, and straightly travels through the observed value 23 and the first half of the sliding window 20 shown on the target trajectory 24. The period sliding window 21 is divided, and the latter half is divided as a falling period sliding window 22. The division of the sliding window 20 is determined so that the following expression (1) is established.

は加速度推定ステップＳＴ１２で算出された加速度推定値の下方向の成分である。また右辺のｇは重力加速度であり、ｓ’は算出したい分割の区切りである。 Here, s on the left side of the equation (1) is the length of the original sliding window applied in the acceleration estimation step ST12,

Is a downward component of the acceleration estimated value calculated in the acceleration estimating step ST12. In addition, g on the right side is gravitational acceleration, and s ′ is a division break to be calculated.

FIG. 4 is a diagram for explaining the division of the sliding window of the falling motion determination device 10 according to the first embodiment of the present invention. FIG. 4 shows a characteristic line A indicating the residual due to the estimated acceleration value and a characteristic line B indicating the residual due to the actual drop motion.
The difference in position between the target during acceleration and the target during constant speed linear movement is proportional to the square of the acceleration time at each time. Also, the accumulation of residuals in the sliding window is considered to be proportional to the cube of the value obtained by integrating the difference in position between the target during acceleration and the target during constant speed linear movement.
As a result, the acceleration estimated value calculated in the acceleration estimation step ST12 is generated from the integration of this residual (characteristic line A) when it occurs at all sampling times of the sliding window, and from some sampling time in the sliding window. The integrals of gravitational acceleration (characteristic line B) coincide. This means that equation (1) holds.

  The falling period sliding window is configured by the sampling time group after the calculated s ′, and the straight traveling period sliding window is configured by the sampling time group before the s ′.

  For example, as shown in step ST16, when the division delimiter is not found in equation (1), such as when the division delimiter is calculated outside the sliding window, the sliding window dividing unit 2 indicates that “the target object has fallen. The process is terminated. On the other hand, as shown in step ST16, when a division break is found, the process proceeds to the next acceleration re-estimation step ST17.

  In the acceleration re-estimation step ST17, the acceleration estimation unit 1 re-executes the acceleration estimation in each of the straight traveling period sliding window and the falling period sliding window divided in the sliding window dividing step ST15 (step ST17).

  As for the above-mentioned sliding window segmentation breaks, taking into account acceleration estimation errors, several samples before and after the value determined in Equation (1) are candidates, and gravity falls in the fall period at each segmentation break. It is also possible to calculate the acceleration and select a segment where the acceleration closest to the gravitational acceleration is obtained.

  In this step, when a significant acceleration is not calculated in the straight period sliding window and an acceleration estimated value equivalent to the gravitational acceleration is calculated in the falling period sliding window as shown in step ST18, the acceleration estimating unit 1 sets “target It is determined that the object is falling, and the process is terminated.

  On the other hand, as shown in step ST18, when the acceleration estimated value corresponding to the gravitational acceleration is not calculated in the falling period sliding window, for example, a significant acceleration is calculated in the straight traveling period sliding window, or equivalent to falling in the falling period. When the acceleration estimation value not to be calculated is calculated, the acceleration estimation unit 1 determines that “the target is not falling” and ends the process.

  As described above, in the first embodiment, the tracking filter unit 3 that estimates the motion specification of the target based on the observation value obtained from the sensor 4 and the observation value to which information from the tracking filter unit 3 is added. The acceleration estimation unit 1 that calculates the average acceleration of the target in the sliding window in which the reflection is reflected, and the sliding window that divides the sliding window into the straight traveling period and the falling period of the target based on the acceleration and the length of the sliding window Since it has a dividing unit 2 and divides the sliding window and divides the target into a straight running period and a falling period to estimate acceleration, it goes without saying that a motion other than a falling motion is not erroneously determined as a falling motion. Even if the sliding window is long, the detection deviation time until the falling motion state is detected is minimized and early It is under judgment. Further, since the partitioning interval of the sliding window is determined based on the acceleration calculated by the original sliding window, the processing load is reduced compared with the case of searching for the optimal partition while shifting the partitioning by one.

Embodiment 2. FIG.
A falling motion determination device according to Embodiment 2 will be described with reference to the drawings.
FIG. 5 is a block diagram showing a configuration of the falling motion determination device 11 according to the second embodiment. The falling motion determination device 11 has a falling motion likelihood determination unit 5 that calculates the likelihood of the falling motion using the observation value of the sensor 4 and finally determines whether or not the target is falling. The configuration is different from that of the falling motion determination device 10 shown in FIG. In addition, about the structure similar to Embodiment 1, the same code | symbol is attached | subjected and description is abbreviate | omitted.

  The fall motion likelihood determination unit 5 is an observation value in a sliding window or a fall period sliding window (hereinafter referred to as a “falling sliding window”) in which the acceleration estimation unit 1 determines that the acceleration estimated value corresponds to gravitational acceleration. Thus, the fall motion likelihood determination is performed.

  Next, the process procedure of the fall motion determination by the drop motion determination device 11 will be described. FIG. 6 is a flowchart showing a processing procedure for one sampling time of the falling motion determination device 11 according to Embodiment 2 of the present invention. The processing from step ST21 to step ST28 shown in FIG. 6 is the same as the processing from step ST11 to step ST18 of FIG.

  As shown in FIG. 6, when it is determined in step ST24 that the estimated acceleration value corresponds to the gravitational acceleration, or when it is determined in step ST28 that the estimated acceleration value corresponds to the gravitational acceleration, the fall motion likelihood determining unit 5 performs the fall motion likelihood determination based on the observed value in the sliding window of the fall motion (step ST29). Thereby, the accuracy of determination can be improved.

  When the following inequality (2) regarding the likelihood of the drop motion is satisfied, that is, when the drop motion likelihood is large in step ST30, the drop motion likelihood determination unit 5 determines that “the target is falling”. To finish the process. On the other hand, when the inequality (2) regarding the likelihood of the drop motion does not hold, that is, when the drop motion likelihood is not large in step ST30, the drop motion likelihood determination unit 5 indicates that “the target has not dropped”. To end the process.

また、“ｆａｌｌ”は目標物が落下しているという事象を示す。また、
ｔｈｒｅｓｈｏｌｄは判定における閾値であり、事前に設定されるパラメータである。
不等式（２）の左辺の算出方法を以下の式（４）に示す。 Here, Z in equation (2) is all observed values within the sliding window of the drop motion, as shown in equation (3) below.

“Fall” indicates an event that the target is falling. Also,
“Threshold” is a threshold value in the determination, and is a parameter set in advance.
The calculation method of the left side of inequality (2) is shown in the following expression (4).

ここで、Ｒは観測誤差共分散行列である。また、落下以外の運動のふらつきの要素を観測誤差に加算した誤差共分散行列であってもよい。また、

は以下の式（５）に示すように、落下開始時刻から重力加速度がかかることを仮定して算出した、目標軌道の各サンプリング時刻における目標物の位置である。より具体的には、図７に示すように、落下期間スライディングウィンドウの開始時刻の追尾平滑値２５を初期値とし、下方向に重力加速度がかかり続ける運動として目標軌道Ｃに示すような軌道の計算により特定される目標物の位置である。

Here, R is an observation error covariance matrix. Alternatively, an error covariance matrix obtained by adding an element of motion fluctuation other than falling to the observation error may be used. Also,

Is the position of the target at each sampling time of the target trajectory, calculated by assuming that gravitational acceleration is applied from the drop start time, as shown in the following equation (5). More specifically, as shown in FIG. 7, the tracking smooth value 25 at the start time of the falling period sliding window is set as an initial value, and a trajectory calculation as shown in the target trajectory C is performed as a motion in which gravity acceleration continues to be applied downward. The position of the target specified by

In the processing procedure of the second embodiment, as shown in FIG. 6, when an acceleration estimated value corresponding to the gravitational acceleration is calculated in the acceleration estimating step (step ST22) or the acceleration re-estimating step (step ST27). In order to increase the determination accuracy, the fall motion likelihood determination is performed.
However, as shown in FIG. 8, when an acceleration estimated value corresponding to the gravitational acceleration is calculated in the acceleration estimating step (step ST32) (step ST34), or equivalent to the gravitational acceleration in the acceleration re-estimating step (step ST37). When the estimated acceleration value to be calculated is calculated (step ST38), it is determined that “the target is falling” and the process is terminated, and when no fall is detected in the acceleration estimation step ST32 or the acceleration re-estimation step ST37. The fall motion likelihood determination (step ST39 and step ST40) may be performed.

  As described above, according to the second embodiment, the tracking filter unit 3 that estimates the motion specification of the target based on the observation value obtained from the sensor 4 and the observation to which the information from the tracking filter unit 3 is added. Acceleration estimation unit 1 that calculates the average acceleration of the target in the sliding window that reflects the value, and sliding that divides the sliding window into two parts, the straight traveling period and the falling period of the target, based on the acceleration and the length of the sliding window It is configured to include a window dividing unit 2 and a falling motion likelihood determination unit 5 that calculates the likelihood of the falling motion using the observed value and finally determines whether or not the target is falling. Therefore, the same effect as in the first embodiment can be obtained, and the fall is finally determined while calculating the likelihood from the observed value to the drop motion after the acceleration estimation. Constant of probability is reduced.

Embodiment 3 FIG.
A falling motion determination apparatus according to Embodiment 3 will be described with reference to the drawings.
FIG. 9 is a block configuration diagram of a drop motion determination device 12 according to Embodiment 3 of the present invention. The fall motion determination device 12 calculates the likelihood for a finite number of motion candidates including the fall motion using the observed values, and finally determines the target from the likelihood of the fall motion with respect to the sum of the likelihoods of the motions as candidates. The configuration is different from the drop motion determination device 10 shown in the first embodiment in that it has a drop motion suitability determination unit 6 that determines whether or not the camera is falling. In addition, about the structure similar to Embodiment 1, the same code | symbol is attached | subjected and description is abbreviate | omitted.
The drop motion suitability determination unit 6 performs the drop motion suitability determination based on the observation value in the sliding window of the drop motion in the acceleration estimation unit 1.

  Next, a processing procedure for determining a falling motion by the falling motion determination device 12 will be described. FIG. 10 is a flowchart showing a processing procedure for one sampling time of the falling motion determination device 12 according to the third embodiment of the present invention. The processing from step ST41 to step ST48 shown in FIG. 10 is the same as the processing from step ST11 to step ST18 of FIG.

  As shown in FIG. 10, when it is determined in step ST44 that the estimated acceleration value corresponds to the gravitational acceleration, or in step ST48, it is determined that the estimated acceleration value corresponds to the gravitational acceleration, the falling motion suitability determining unit 6 performs fall motion compatibility determination based on the sensor measurement value in the sliding window of the fall motion (step ST49). This increases the accuracy of the determination and prevents erroneous determination of movement similar to a fall such as turning.

  When the following inequality (6) relating to the posterior probability holds, that is, as a result of the fall motion suitability determination in step ST49, it is determined in step ST50 that the fall motion fit is determined, the fall motion suitability judgment unit 6 sets the “target It is determined that “has fallen” and the process is terminated. On the other hand, when the inequality (6) regarding the likelihood of the drop motion does not hold, that is, when it is determined in step ST50 that the drop motion is not suitable, the drop motion suitability determination unit 6 indicates that “the target has not dropped. ”And the process ends.

  Here, Z is all observed values in the sliding window of the falling motion as shown in the following formula (7).

また、“ｆａｌｌ”は目標物が落下しているという事象を示す。また
ｔｈｒｅｓｈｏｌｄは判定における閾値であり、事前に設定するパラメータである。

“Fall” indicates an event that the target is falling. Threshold is a threshold value for determination, and is a parameter set in advance.

The following equation (8) shows how to calculate the left side of inequality (6). From Bayes' theorem, it can be expanded as follows.

  Here, “motion” indicates an event that the target is performing a specific motion. Although this movement can be innumerable, in the third embodiment, the drop movement 26 shown in FIG. 11 and a finite number of downwards such as the first lower S-shaped swivel 27 and the second lower S-shaped swirl 28, for example. Limited to swivel motion. That is, it is assumed that the other movements are excluded by the process before this step.

  The first term of the numerator and denominator on the right side of Equation (8) is the likelihood of each motion based on the observed value, as shown in Equation (9) below.

は、以下の式（１０）に示すように、想定した運動モデルにより算出した目標軌道の各サンプリング時刻における目標物の位置である。 Here, R is an observation error covariance matrix. Further, an error covariance matrix obtained by adding modeling errors to a limited number of models may be used. Also,

Is the position of the target at each sampling time of the target trajectory calculated by the assumed motion model, as shown in the following equation (10).

  From the tracking smooth value at the start time of the sliding window of the falling motion, the trajectory of each motion is calculated using the above formula to obtain the target trajectory shown in FIG. Here, FIG. 12A shows the target trajectory D calculated as a motion in which the gravitational acceleration continues to be applied in the downward direction with respect to the falling motion. In addition, FIG. 12B shows the target activation E calculated as a motion of turning downward for a certain period 29, reversely turning for a certain period 30 thereafter, and going straight for the subsequent period 31. Yes.

  The prior probabilities P (fall) and P (motion) on the right side of Expression (10) are equal probabilities for all motion models. Further, if observation information other than the position of the target is obtained from the sensor 4, the prior probability may be weighted accordingly.

  As described above, in the third embodiment, the tracking filter unit 3 that estimates the motion specifications of the target based on the observation value obtained from the sensor 4, and the observation value to which the information from the tracking filter unit 3 is added. The acceleration estimation unit 1 that calculates the average acceleration of the target in the sliding window in which the reflection is reflected, and the sliding window that divides the sliding window into the straight traveling period and the falling period of the target based on the acceleration and the length of the sliding window The fall motion adaptation which calculates the likelihood with respect to each of the dividing unit 2 and the finite number of drop motion candidates based on the observation values and finally determines whether or not the target is falling using the sum of the likelihoods Since the determination unit 6 is provided, the same effect as in the first embodiment can be obtained, and a specific value can be obtained from the observed value after the acceleration estimation. Since determining the final fall while calculating the goodness of fit to the dynamic it can be determined more accurately.

  In the present invention, within the scope of the invention, any combination of the embodiments, or any modification of any component in each embodiment, or omission of any component in each embodiment is possible. .

  DESCRIPTION OF SYMBOLS 1 Acceleration estimation part, 2 Sliding window division | segmentation part, 3 Tracking filter part, 4 Sensor, 5 Falling motion likelihood determination part, 6 Falling motion suitability determination part, 10, 11, 12 Falling motion determination apparatus, 20 Sliding window, 21 Straight ahead Periodic sliding window, 22 Falling period sliding window, 23 Observed value, 24 Target trajectory, 25 Tracking period smoothing value of start time of falling period sliding window, 26 Falling motion, 27 First lower S-turn, 28 Second lower S Character turn, 29, 30 Fixed period, 31 period, 41 Sliding window at time t1, 42 Sliding window at time t2, 43 Observation value, 44 Target trajectory, 45 Acceleration estimated value.

Claims (4)

The target is tracked using the observed value of the target position obtained from the sensor, and the target is dropped from the falling motion state and the constant-velocity linear motion state using the information obtained from the tracking and the observed value. In the fall motion determination device that determines the change to the motion state,
A tracking filter unit that estimates the motion specifications of the target based on the observation value obtained from the sensor;
An acceleration estimation unit that calculates an average acceleration of the target in a sliding window in which the observation value to which the information from the tracking filter unit is added is reflected;
And a sliding window dividing section bisecting the previous SL sliding window straight period and falling period of the target,
The sliding window dividing unit is configured to calculate the sliding window of the falling period side from the division boundary based on a value obtained by cubeing the length of the sliding window before the division, the downward component of the acceleration, and the gravitational acceleration. A falling motion determination device characterized in that a value obtained by squaring the length is calculated, and the length of the sliding window on the falling period side is obtained from the value . The acceleration estimating unit falling motion determination apparatus according to claim 1, wherein the sliding window division unit performs a acceleration estimated using the split sliding window divided according to the estimated error range of the acceleration . The target is tracked using the observed value of the position of the target obtained from the sensor, and the falling motion and the constant-velocity linear motion state of the target are obtained using the information obtained from the tracking and the observed value of the sensor. In the falling motion determination device for determining the change from the falling motion state to the
A tracking filter unit that estimates the motion specifications of the target based on the observation value obtained from the sensor;
An acceleration estimation unit that calculates an average acceleration of the target in a sliding window in which the observation value to which the information from the tracking filter unit is added is reflected;
A sliding window dividing section bisecting the previous SL sliding window falling period and the straight period of the target,
A drop motion likelihood determination unit that calculates the likelihood of the drop motion using the observation value, and finally determines whether or not the target is falling ;
The sliding window dividing unit is configured to calculate the sliding window of the falling period side from the division boundary based on a value obtained by cubeing the length of the sliding window before the division, the downward component of the acceleration, and the gravitational acceleration. A falling motion determination device characterized in that a value obtained by squaring the length is calculated, and the length of the sliding window on the falling period side is obtained from the value . The target is tracked using the observed value of the position of the target obtained from the sensor, and the falling motion and the constant-velocity linear motion state of the target are obtained using the information obtained from the tracking and the observed value of the sensor. In the falling motion determination device for determining the change from the falling motion state to the
A tracking filter unit that estimates the motion specifications of the target based on the observation value obtained from the sensor;
An acceleration estimation unit that calculates an average acceleration of the target in a sliding window in which the observation value to which the information from the tracking filter unit is added is reflected;
A sliding window dividing section bisecting the previous SL sliding window falling period and the straight period of the target,
A fall motion that calculates the likelihood of a fall motion for each of a finite number of fall motion candidates based on the observed value and finally determines whether the target is falling using the sum of the likelihoods A conformity determination unit ,
The sliding window dividing unit is configured to calculate the sliding window of the falling period side from the division boundary based on a value obtained by cubeing the length of the sliding window before the division, the downward component of the acceleration, and the gravitational acceleration. A falling motion determination device characterized in that a value obtained by squaring the length is calculated, and the length of the sliding window on the falling period side is obtained from the value .

Images