Sensing data processing device, sensor system, sensing data processing method, and recording medium

Abstract

A sensing data processing device (200) according to the present invention comprises a position estimation means (281) that: acquires a plurality of pieces of coordinate data of candidate points which are samples of points to be candidate positions of an object, on the basis of measurement results of the positions of the object by a sensor (110); performs weighting on each of the plurality of candidate points on the basis of the position of said candidate point in the coordinate space of the sensor (110); and estimates the position of the object on the basis of the plurality of weighted candidate points.　With the present invention, even when the distribution of candidate positions of an object is acquired from a single coordinate value obtained through measurement of the position of the object by a sensor, it is possible to reflect, on position estimation, the differences in sensor sensitivity due to variations in the position of the object in the distribution.

Description

Sensing data processing device, sensor system, sensing data processing method, and recording medium

[0001] The present invention relates to a sensing data processing device, a sensor system, a sensing data processing method, and a recording medium.

[0002] Position estimation may be performed by replacing coordinate values ​​obtained by measurement using a sensor such as a radar with a probability distribution. For example, Patent Document 1 describes a method of performing position estimation by replacing multiple coordinate values ​​with a normal distribution. In the method described in Patent Document 1, a target is measured using multiple sensors, and the coordinate values ​​obtained from each sensor are replaced with a probability distribution based on a normal distribution using a variance value that indicates the accuracy of the coordinate values. Then, in this method, the probability distributions for each sensor are multiplied, and the coordinate values ​​at the peak of the obtained distribution are identified as the most likely coordinate values ​​of the target.

[0003] JP 2012-163495 A

[0004] The sensitivity of a sensor may differ depending on the position of an object to be estimated. Even when a distribution of candidate positions of an object is obtained from a single coordinate value obtained by measuring the position of the object with a sensor, it is preferable to reflect the difference in sensor sensitivity due to the difference in the position of the object in the distribution in the position estimation.

[0005] An example of an object of the present invention is to provide a sensing data processing device, a sensor system, a sensing data processing method, and a recording medium that can solve the above-mentioned problems.

[0006] According to a first aspect of the present invention, a sensing data processing device includes a position estimation means that acquires a plurality of coordinate data of candidate points, which are samples of points that are candidates for the position of the object, based on the measurement results of the position of the object by a sensor, weights each of the plurality of candidate points based on the position of the candidate point in the coordinate space of the sensor, and estimates the position of the object based on the weighted plurality of candidate points.

[0007] According to a second aspect of the present invention, a sensor system includes a sensor that measures the position of an object, and a position estimation means that acquires, based on the measurement results of the sensor, a plurality of coordinate data of candidate points that are samples of points that are candidates for the position of the object, weights each of the plurality of candidate points based on its position in the coordinate space of the sensor, and estimates the position of the object based on the weighted plurality of candidate points.

[0008] According to a third aspect of the present invention, a sensing data processing method includes a computer acquiring, based on a measurement result of the position of an object by a sensor, a plurality of coordinate data of candidate points which are samples of points that are candidates for the position of the object, weighting each of the plurality of candidate points based on the position of the candidate point in the coordinate space of the sensor, and estimating the position of the object based on the weighted plurality of candidate points.

[0009] According to a fourth aspect of the present invention, a recording medium is a recording medium having recorded thereon a program that causes a computer to acquire, based on the measurement results of the position of an object by a sensor, a plurality of coordinate data of candidate points that are samples of points that are candidates for the position of the object, weight each of the plurality of candidate points based on the position of the candidate point in the coordinate space of the sensor, and estimate the position of the object based on the weighted plurality of candidate points.

[0010] According to the present invention, even when a distribution of candidate positions of an object is obtained from a single coordinate value obtained by measuring the position of the object with a sensor, differences in the sensitivity of the sensor due to differences in the position of the object in the distribution can be reflected in the position estimation.

[0011] 1 is a diagram illustrating an example of the configuration of a sensor system according to at least one embodiment; FIG. 2 is a diagram illustrating an example of the configuration of a sensing data processing device according to at least one embodiment; FIG. 3 is a diagram illustrating an example of a tracking trajectory when a sensor system according to at least one embodiment includes multiple sensors; FIG. 4 is a diagram illustrating an example of a hypothesis of a tracking trajectory when a sensor system according to at least one embodiment includes one sensor; FIG. 5 is a diagram illustrating an example of the space of an object observed by a sensor according to at least one embodiment; FIG. 6 is a diagram illustrating an image of a portion of the optimization calculation performed by an optimization calculation unit according to at least one embodiment, the portion being based on an index value that performs weighting taking into account the coordinate space of the sensor; FIG. 7 is a diagram illustrating an image of a portion of the optimization calculation performed by an optimization calculation unit according to at least one embodiment, the portion being based on an index value based on the Kullback-Leibler distance; FIG. 8 is a diagram illustrating an example of a processing procedure performed by a sensing data processing device according to at least one embodiment; FIG. 9 is a diagram illustrating an example of the configuration of a sensing data processing device according to at least one embodiment; FIG. 10 is a diagram illustrating an example of the configuration of a sensor system according to at least one embodiment; FIG. 11 is a diagram illustrating an example of a processing procedure in a sensing data processing method according to at least one embodiment; FIG. 12 is a schematic block diagram illustrating the configuration of a computer according to at least one embodiment;

[0012] The following describes embodiments of the present invention, but the following embodiments do not limit the scope of the invention. Furthermore, not all of the combinations of features described in the embodiments are necessarily essential to the solution of the invention. Below, letters with circumflexes may be indicated with a "^" after the letter. For example, a circumflex f may also be written as f^.

[0013] <First embodiment> Fig. 1 is a diagram showing an example of the configuration of a sensor system according to at least one embodiment. In the configuration shown in Fig. 1, the sensor system 1 includes n sensors 110 and a sensing data processing device 200. Here, n is an integer such that n ≥ 1. When distinguishing between the n sensors 110, they are also referred to as sensor 110-1, sensor 110-2, ..., sensor 110-n.

[0014] The sensor system 1 measures the position of an object and estimates the position of the object based on the measurement results. The object here refers to an object whose position is to be estimated. The object is not limited to a specific object, but can be various movable objects. For example, the object may be an object that can move in the air or space, such as an aircraft, drone, or rocket. Alternatively, the object may be an object that can move on land or sea, such as an automobile or ship. The following describes an example in which the object moves in the air.

[0015] The sensor 110 measures the position of an object. The sensor system 1 may estimate the positions of all objects located within an area where the sensor 110 can measure the positions. In this case, all objects located within an area where the sensor 110 can measure the positions correspond to the target object.

[0016] Alternatively, the sensor system 1 may estimate the position of an object that satisfies a predetermined condition, such as an object that is moving at a speed equal to or greater than a predetermined speed, among objects located within an area whose position can be measured by the sensor 110. In this case, the object that satisfies the predetermined condition corresponds to the target object.

[0017] The sensor 110 is not limited to a specific type of sensor, and may be any of various sensors capable of measuring the position of an object. For example, the sensor 110 may be a radar, an infrared sensor such as a far-infrared sensor, an image sensor, or a light detection and ranging (LiDAR). Furthermore, multiple types of sensors may be used as the sensor 110.

[0018] If the sensor 110 is a type of sensor that emits a signal and receives a reflected signal, such as a radar or ultrasonic sensor, the transmitter (the device that emits the signal) and the receiver (the device that receives the reflected signal) may be in the same location or in different locations.

[0019] Hereinafter, time will be represented by time steps such as time 1, 2, 3, ..., and the sensor 110 will measure the position of an object at each time step. If the sensor system 1 includes multiple sensors 110, each of the sensors 110 will measure the position of an object at each time step.

[0020] The sensing data processing device 200 estimates the position of the object based on the measurement result of the position of the object by the sensor 110. The sensing data processing device 200 may be configured using a computer. Here, the measurement result of the position of the object by the sensor 110 may contain an error. Therefore, in the sensor system 1, the measurement result by the sensor 110 is not used as is, but the sensing data processing device 200 estimates the position of the object.

[0021] 2 is a diagram showing an example of the configuration of the sensing data processing device 200. In the configuration shown in Fig. 2, the sensing data processing device 200 includes a communication unit 210, a display unit 220, an operation input unit 230, a storage unit 270, and a processing unit 280. The processing unit 280 includes a position estimation unit 281 and an output control unit 291. The position estimation unit 281 includes a sensing data acquisition unit 282, a tracking processing unit 283, a tracking correlation calculation unit 284, a distance index calculation unit 285, a distance index integration unit 286, an optimization calculation unit 287, and a trajectory integration unit 288.

[0022] The communication unit 210 communicates with other devices. For example, the communication unit 210 acquires sensing data indicating measurement results of the position of an object from the sensor 110. The display unit 220 has a display screen such as a liquid crystal panel or an LED (Light Emitting Diode) panel, and acquires various images. For example, the display unit 220 may visually display the estimated position or estimated trajectory of the object, such as by displaying it as a graph.

[0023] The operation input unit 230 includes input devices such as a keyboard and a mouse, and receives user operations. For example, the operation input unit 230 may receive user operations for setting values ​​used by the sensing data processing device 200 to estimate the position of an object, such as weighting coefficient values ​​for each type of distance.

[0024] The storage unit 270 stores various data. For example, the storage unit 270 may store history data of the measurement results of the position of an object by the sensor 110. The storage unit 270 is configured using a storage device provided in the sensing data processing device 200. The processing unit 280 controls each unit of the sensing data processing device 200 to perform various processes. The functions of the processing unit 280 are performed, for example, by a CPU (Central Processing Unit) provided in the sensing data processing device 200 reading and executing a program from the storage unit 270.

[0025] The position estimation unit 281 acquires coordinate data of a plurality of candidate points based on the measurement results of the position of the object by the sensor 110. The candidate points here are a sample of points that are candidates for the position of the object. The sample here means a part of the population. The fact that the candidate points are a sample of points that are candidates for the position of the object means that the candidates for the position of the object are not necessarily limited to the candidate points acquired by the position estimation unit 281.

[0026] The position estimation unit 281 acquires coordinate data for multiple candidate points, and weights each of the multiple candidate points based on the position of the candidate point in the coordinate space of the sensor. The position estimation unit 281 then estimates the position of the object based on the weighted multiple candidate points. The position estimation unit 281 is an example of a position estimation means.

[0027] The sensing data acquisition unit 282 acquires sensing data from the sensor 110. The sensing data from the sensor 110 is coordinate data of the measurement results of the position of an object. The sensing data acquisition unit 282 may acquire additional information in addition to the coordinate data of the measurement results of the position of the object as input information from the sensor 110. For example, the sensing data acquisition unit 282 may acquire, as the additional information, a time-series sensor signal such as a reflected wave, a sensor parameter such as the signal-to-noise ratio (SNR) of the signal received by the sensor 110, external information such as information on the sensor installation location, information on the observation time (information for time synchronization), and the observation environment (temperature, topography), or inter-sensor information such as network information between the sensors 110, or a combination of these.

[0028] The tracking processing unit 283 performs tracking processing on the sensing data from the sensor 110. The tracking processing here refers to associating the coordinates indicated by the sensing data with the tracking trajectory of the object. The tracking trajectory here is a trajectory estimated by the sensing data processing device 200 as a history of the trajectory of the object.

[0029] Specifically, the tracking processing unit 283 estimates the position of the object at time t based on the tracking trajectory up to time t-1 and the sensing data at time t, and adds the estimated position at time t to the tracking trajectory up to time t-1. In this way, the tracking processing unit 283 updates the tracking trajectory up to time t-1 to the tracking trajectory up to time t.

[0030] When the sensor system 1 has multiple sensors 110, the tracking processing unit 283 uses the location information of the sensors 110 acquired by the sensing data acquisition unit 282, information such as the observation time, to achieve temporal and spatial synchronization between the sensors 110.

[0031] When the sensor system 1 includes multiple sensors 110 and there is one target object, the storage unit 270 may store a tracking trajectory for each sensor 110. The tracking processing unit 283 may then associate the tracking trajectory up to time t-1 for each sensor 110 with any of the coordinates of the measurement results of the target object's position at time t obtained by each of the multiple sensors 110. The tracking processing unit 283 calculates an estimated position of the target object at time t based on the tracking trajectory up to time t-1 and the coordinates associated with that tracking trajectory, and adds the estimated position at time t to the tracking trajectory up to time t-1. In this way, the tracking processing unit 283 updates the tracking trajectory up to time t-1 for each sensor 110 to the tracking trajectory up to time t for each sensor 110 up to time t.

[0032] If the sensor system 1 includes multiple sensors 110 and multiple objects, the storage unit 270 may store a tracking trajectory for each sensor 110 and for each object. The tracking processing unit 283 may then associate the tracking trajectory for each sensor 110 and for each object up to time t-1 with any of the coordinates of the measurement results of the positions of each of the multiple objects at time t obtained by each of the multiple sensors 110. The tracking processing unit 283 calculates an estimated position of the object at time t based on the tracking trajectory up to time t-1 and the coordinates associated with the tracking trajectory, and adds the estimated position at time t to the tracking trajectory up to time t-1. In this way, the tracking processing unit 283 updates the tracking trajectory for each sensor 110 and for each object up to time t-1 to a tracking trajectory for each sensor 110 and for each object up to time t.

[0033] However, the actual number of objects does not need to be known to the sensing data processing device 200. For example, the tracking processing unit 283 may perform tracking processing assuming that there are the same number of objects as the positions of the objects detected by the sensor 110.

[0034] Fig. 3 is a diagram showing an example of a tracking trajectory when the sensor system 1 includes a plurality of sensors 110. Fig. 3 shows an example when the sensor system 1 includes three sensors 110, namely, sensors 110-1, 110-2, and 110-3.

[0035] Lines L11 to L16 each show an example of a tracking trajectory. Fig. 3 shows an example in which there are two objects, and each sensor 110 detects the positions of two objects. As a result, the tracking processing unit 283 calculates six tracking trajectories.

[0036] For each of these six tracking trajectories, the tracking processing unit 283 associates the tracking trajectory up to time t-1 with any of the positions detected by the sensor 110 at time t. As a result, the tracking processing unit 283 updates each of these six tracking trajectories from the tracking trajectory up to time t-1 to the tracking trajectory up to time t.

[0037] When the sensor system 1 includes one sensor 110 and there is one target object, the storage unit 270 may store multiple hypotheses for the tracking trajectory. For example, when the sensing data from the sensor 110 contains noise, multiple hypotheses for the tracking trajectory may be generated by changing the weighting of the degree of reliability, i.e., the degree of reliability of the sensing data at each time. The tracking processing unit 283 may then estimate the position of the target object at time t for each hypothesis for the tracking trajectory up to time t-1 based on that hypothesis and the measured value of the target object's position at time t, and add the estimated position to the hypotheses for the tracking trajectory. In this way, the tracking processing unit 283 updates each of the hypotheses for the tracking result up to time t-1 to the hypothesis for the tracking result up to time t.

[0038] 4 is a diagram showing examples of tracking trajectory hypotheses when the sensor system 1 includes one sensor 110. Lines L21 to L23 each represent a tracking trajectory hypothesis. For each of these three hypotheses, the tracking processing unit 283 updates the tracking trajectory hypothesis up to time t-1 to a tracking trajectory hypothesis up to time t based on the position detected by the sensor 110 at time t.

[0039] When the sensor system 1 includes one sensor and there are multiple targets, the storage unit 270 may store a tracking trajectory for each target. The tracking processing unit 283 may then associate the tracking trajectory for each target up to time t-1 with any of the coordinates of the measurement results of the positions of each of the multiple targets at time t. The tracking processing unit 283 calculates an estimated position of the target at time t based on the tracking trajectory up to time t-1 and the coordinates associated with that tracking trajectory, and adds the estimated position at time t to the tracking trajectory up to time t-1. In this way, the tracking processing unit 283 updates the tracking trajectory for each target up to time t-1 to the tracking trajectory for each target up to time t.

[0040] The tracking processing unit 283 updates each tracking trajectory using sensing by the same sensor 110. That is, the tracking processing unit 283 calculates an estimated position of the object at time t using sensing data from the same sensor 110 as the sensor 110 used to generate the tracking trajectory up to time t-1. Then, the tracking processing unit 283 adds the calculated estimated position of the object at time t to the tracking trajectory.

[0041] In the following, an example will be described in which the tracking processing unit 283 calculates a probability distribution of the estimated position of the object for each tracking trajectory and for each time step. That is, an example will be described in which the tracking processing unit 283 generates a tracking trajectory in which candidates for the position of the object are represented by a probability distribution.

[0042] For example, the tracking processing unit 283 may predict a probability distribution of the position of the object at time t based on the tracking trajectory up to time t-1. Then, the tracking processing unit 283 may calculate, for the predicted probability distribution, a posterior distribution under the observed value of the position of the object by the sensor 110 at time t, as a probability distribution of the estimated position of the object at time t. The probability distribution of the estimated position of the object corresponds to an example of a distribution of candidates for the position of the object.

[0043] However, the method by which the tracking processing unit 283 obtains the distribution of candidate positions of the target object is not limited to the method of obtaining a probability distribution of the estimated position of the target object. For example, the tracking processing unit 283 may obtain coordinate data of a plurality of candidate points.

[0044] The tracking processing unit 283 may use a known method such as a Joint Probabilistic Data Association Filter (JPDA), a Probability Hypothesis Density Filter (PHD), or a Multiple Hypothesis Tracking (MHT) as a method for determining the distribution of candidate objects. The tracking processing unit 283 may perform tracking processing using additional information in addition to the coordinate data of the position of the object obtained by the sensor 110.

[0045] The tracking correlation calculation unit 284 calculates the correlation between the tracking trajectories. The tracking correlation calculation unit 284 determines that two tracking trajectories with a correlation stronger than a certain level are the same tracking trajectory. For example, the tracking correlation calculation unit 284 may calculate the spatiotemporal distance between the two tracking trajectories, such as the distance between points on the two tracking trajectories at each time, as the correlation between the two tracking trajectories. Then, the tracking correlation calculation unit 284 may determine that tracking trajectories with a distance smaller than a certain level are the same tracking trajectory. However, the method by which the tracking correlation calculation unit 284 calculates the correlation between the tracking trajectories is not limited to a specific method. The tracking processing unit 283 may determine the identity of the tracking trajectories by using additional information in addition to the coordinate data of the position of the target object obtained by the sensor 110.

[0046] The distance index calculation unit 285 calculates an index value of the spatial distance between two probability distributions of candidate object positions. In particular, the distance index calculation unit 285 calculates an index value of the spatial distance, including an index value that performs weighting taking into account the coordinate space of the sensor 110, such as Earth Mover's Distance (EMD). The index of the spatial distance is also referred to as a distance index.

[0047] In calculating the weighted index value taking into account the coordinate space of the sensor 110, the distance index calculation unit 285 acquires coordinate data for multiple candidate points. As described above, the candidate points are samples of points that are candidates for the position of the target object. The distance index calculation unit 285 then weights each of the multiple candidate points based on the position of the candidate point in the coordinate space of the sensor 110.

[0048] For example, the distance index calculation unit 285 may sample candidate points according to a probability distribution. Alternatively, the distance index calculation unit 285 may divide an area of ​​the probability distribution where the probability density is equal to or greater than a predetermined value into a mesh and sample one point for each square.

[0049] The distance index calculation unit 285 may calculate an index value using Earth Mover's Distance (EMD) as an index value for weighting that takes into account the coordinate space of the sensor 110. Here, the first set of two sets of pairs of positions and weights in the Earth Mover's Distance is designated as X, and is expressed as in Equation (1).

[0050]

[0051] x ｉ indicates the position. ｘｉ is the position x ｉ Here, i is an integer in the range of 1≦i≦m. Also, m is the weight of the position x ｉ And weight lol ｘｉ Pair with (x ｉ , w ｘｉ ) and m is an integer of 2 or more. ｉ indicates the coordinate values ​​of a candidate point sampled from the probability distribution of candidate positions of the object included in the tracking trajectory.

[0052] Weight w ｘｉ The value of x is determined depending on the position of the sensor 110 in the coordinate space, so that the value becomes larger as the accuracy (resolution) of the sensor 110 is considered to be higher. For example, at the position x ｉThe farther away from the position of the sensor 110, the greater the position x ｉ Therefore, the measurement accuracy of the position x ｉ The farther away from the position of the sensor 110, the greater the weight w ｘｉ In addition, if the sensor 110 is an infrared sensor, it is considered that the accuracy of the infrared sensor is lower than that of other types of sensors in areas where there are clouds because infrared rays do not penetrate clouds. Therefore, if the sensor 110 is an infrared sensor and the position x ｉ If is located within the cloud region, the weight w ｘｉ The weight w may be set to a relatively small value. ｘｉ The value of may be determined based on additional information such as the signal-to-noise ratio of the signal received by the sensor 110 .

[0053] Fig. 5 is a diagram showing an example of a space to be observed by the sensor 110. Three sensors, 110-1, 110-2, and 110-3, are shown in Fig. 5. Lines L31 to L33 each show an example of a tracking trajectory. Area A11 is a cloud area.

[0054] For example, consider a case where the tracking trajectory L31 is a trajectory based on sensing data from the sensor 110-1. In this case, it is conceivable to assign a smaller weight value to the sensing data from the sensor 110-1 as the position indicated by the tracking trajectory L31 becomes farther from the sensor 110-1. Furthermore, if the sensor 110-1 is an infrared sensor, it is conceivable to assign a relatively smaller weight value to the sensing data from the sensor 110-1 for positions within the cloud region A11.

[0055] Furthermore, the second set of two sets of pairs of positions and weights in the Earthmover distance is denoted as Y and expressed as in equation (2).

[0056]

[0057] y ｊ indicates the position. ｙｊ is the position y ｊHere, j is an integer in the range of 1≦j≦n. Also, here, n is the weight for the position y ｊ And weight lol ｙｊ Pair with (y ｊ , w ｙｊ ) and n is an integer of 2 or more. ｉ indicates the coordinate values ​​of candidate points included in a tracking trajectory obtained by integrating multiple tracking trajectories.

[0058] Weight w ｙｊ The value of is the weight w ｘｉ The position y in the coordinate space of the sensor 110 that is referenced when determining the value of ｊ Based on this, the weight w ｘｉ The value of the position x is determined by the same criteria as that of the position x. ｉ and position y ｉ The distance between ｉｊ The distance d ｉｊ is expressed as in equation (3).

[0059]

[0060] d(x ｉ , y ｊ ) is the position x ｉ and position y ｉ The distance d may be, but is not limited to, the Euclidean distance. ＥＭＤ (X, Y) is expressed as in equation (4).

[0061]

[0062] f ｉｊ is the earthmover distance D ＥＭＤ This is the amount by which the value is adjusted in the optimization calculation (minimization) of (X, Y). ｉｊ It is assumed that satisfies the following four constraints: One of the constraints is shown as equation (5).

[0063]

[0064] The two constraints are expressed as in equation (6).

[0065]

[0066] The three constraints are expressed as in equation (7).

[0067]

[0068] The four constraints are expressed as in equation (8).

[0069]

[0070] The sensing data processing device 200 may use, as an index value of spatial distance, an index value that is weighted in consideration of the coordinate space of the sensor 110, as well as other index values. That is, the distance index calculation unit 285 may calculate, as an index value of spatial distance, an index value that is weighted in consideration of the coordinate space of the sensor 110, as well as other index values. For example, the sensing data processing device 200 may use the index value shown in equation (9).

[0071]

[0072] X ｋ and X ｋ ' are the set of coordinates of the candidate points at time k, ｋ , ..., xN ｋ}. Z １：ｋ ｊ indicates the observation value by the sensor 110 from time 1 to time k. ｉ , f ｊ and f respectively represent probability distributions. ｉ (X ｋ |Z １：ｋ ｉ Equation (9) using ∫f(X)δX is a generalization of covariance intersection (CI) to multiple objects. ω denotes the parameter of the exponential mixture of distributions. The integral ∫f(X)δX represents an integral such as Equation (10).

[0073]

[0074] Index value f ω (x ｋ |Z １：ｋ ｉ, Z １：ｋ ｊ The index value f based on the Kullback-Leibler divergence is also called the index value based on the Kullback-Leibler divergence. ω (x ｋ |Z １：ｋ ｉ , Z １：ｋ ｊ ) can be interpreted as an index value that evaluates the accuracy of the measurement of the position of the object by the sensor 110 as the reliability of the information.

[0075] In the following, an example will be described in which the sensing data processing device 200 uses an index value based on the Earthmover distance and the Kullback-Leibler distance as an index value of the spatial distance. ＥＭＤ (X, Y) is D ＥＭＤ (f || f ｉ The index value f based on the Kullback-Leibler distance is also expressed as ω (x ｋ |Z １：ｋ ｉ , Z １：ｋ ｊ ) to D ＫＬ (f || f ｉ ) is also written as f, f ｉ and f respectively indicate probability distributions. ｉ denotes the probability distribution of candidate points on one of the tracking trajectories that are the targets of integration by the distance indicator integration unit 286. f denotes the probability distribution of candidate points on the tracking trajectory obtained by integration.

[0076] The distance index integrating unit 286 integrates the multiple types of distance index values ​​calculated by the distance index calculating unit 285. For example, the distance index calculating unit 285 weights and averages the multiple types of distance index values ​​calculated by the distance index calculating unit 285 based on the value of a weight parameter provided for each type of distance index.

[0077] The distance index integrating unit 286 may calculate a weighted arithmetic mean of multiple types of distance index values. For example, the distance index integrating unit 286 may calculate an arithmetic mean of an index value based on the Earthmover distance and the Kullback-Leibler distance as an index value of spatial distance based on Equation (11).

[0078]

[0079] η is a weighting coefficient in the range of 0≦η≦1. Alternatively, the distance index integrating unit 286 may calculate a weighted geometric mean of multiple types of distance index values. For example, the distance index integrating unit 286 may calculate a geometric mean of an index value based on the Earthmover distance and the Kullback-Leibler distance as an index value of spatial distance based on Equation (12).

[0080]

[0081] The distance index calculation unit 285 may calculate three or more types of distance index values. Then, the distance index integrating unit 286 may calculate a weighted average of the three or more types of distance index values. Alternatively, the distance index calculation unit 285 may calculate only one type of distance index value. In this case, the distance index integrating unit 286 may output the distance index value calculated by the distance index calculation unit 285 as is.

[0082] The optimization calculation unit 287 searches for values ​​of weight parameters for integrating tracking trajectories, using as an objective function the functions with which the distance index calculation unit 285 and the distance index integrating unit 286 calculate index values. Specifically, the optimization calculation unit 287 searches for values ​​of weight parameters that minimize the index value calculated by the distance index integrating unit 286. The optimization calculation unit 287 may perform optimization calculation based on equation (13).

[0083]

[0084] In this case, (1-ω)D(f||f ｉ ) + ωD(f||f ｊ) corresponds to the objective function in the optimization calculation. ω^ indicates the value of the weight parameter ω obtained by the optimization calculation. f^ indicates the probability distribution of the object position obtained by the optimization calculation.

[0085] Alternatively, the optimization calculation unit 287 may perform the optimization calculation based on the equation (14).

[0086]

[0087] In this case, D(f||f ｉ ) １－ω D(f || f ｊ ) ω corresponds to the objective function in the optimization calculation. The optimization calculation unit 287 may search for the value ω^ of the parameter ω and the probability distribution f^ based on equation (13) or equation (14). Alternatively, the optimization calculation unit 287 may search for only the value ω^ of the parameter ω based on equation (13) or equation (14).

[0088] In equations (13) and (14), the optimization calculation unit 287 calculates the probability distribution f ｉ and f ｊ However, the optimization calculation unit 287 may search for the probability distribution f^ based on three or more probability distributions. If the tracking correlation calculation unit 284 determines that three or more tracking trajectories are the same tracking trajectory, the optimization calculation unit 287 may calculate a weighted average of the three or more probability distributions indicated by the three or more tracking trajectories in equation (13) or equation (14).

[0089] 6 is a diagram showing an image of a part of the optimization calculation performed by the optimization calculation unit 287, which is an optimization calculation based on an index value that performs weighting taking into account the coordinate space of the sensor 110. In the example of FIG. 6, x １ （１） , x ２ （１） , ..., x ｍ１ （１）indicates the coordinates of the candidate points obtained based on the sensing data from the sensor 110-1. Here, m1 indicates the number of candidate points obtained based on the sensing data from the sensor 110-1.

[0090] Coordinate x １ （１） , x ２ （１） , ..., x ｍ１ （１） The candidate point indicated by x １ （１） , x ２ （１） , ..., x ｍ１ （１） Candidate point x １ （１） , x ２ （１） , ..., x ｍ１ （１） The tracking processing unit 283 may calculate a probability distribution of the position of the target object based on the sensing data from the sensor 110-1, and may sample points according to the calculated probability distribution.

[0091] x １ （２） , x ２ （２） , ..., x ｍ２ （２） indicates the coordinates of the candidate points obtained based on the sensing data by the sensor 110-2. Here, m2 indicates the number of candidate points obtained based on the sensing data by the sensor 110-2.

[0092] Coordinate x １ （２） , x ２ （２） , ..., x ｍ２ （２） The candidate point indicated by x １ （２） , x ２ （２） , ..., x ｍ２ （２） Candidate point x １ （２） , x ２ （２） , ..., x ｍ２ （２）The tracking processing unit 283 may calculate a probability distribution of the position of the target object based on the sensing data from the sensor 110-2, and may sample points according to the calculated probability distribution.

[0093] y １ , y ２ , ..., y ｎ is the candidate point x １ （１） , x ２ （１） , ..., x ｍ１ （１） and candidate point x １ （２） , x ２ （２） , ..., x ｍ２ （２） Here, n indicates the coordinates of the candidate point x １ （１） , x ２ （１） , ..., x ｍ１ （１） and candidate point x １ （２） , x ２ （２） , ..., x ｍ２ （２） The number of candidate points obtained by integrating

[0094] Coordinate y １ , y ２ , ..., y ｎ The candidate point indicated by １ , y ２ , ..., y ｎ The position estimation unit 281 (for example, the optimization calculation unit 287) uses a known sensor fusion method to calculate the candidate point y １ , y ２ , ..., y ｎ may be set.

[0095] As described above, the optimization calculation unit 287 obtains the value ω^ of the weight parameter ω, or obtains the value ω^ of the weight parameter ω and the probability distribution f^ of the object's position through optimization calculation. ω^ can be considered to be a value indicating the ratio of the influence of the probability distribution of the object's position based on the sensing data from sensor 110-1 and the probability distribution of the object's position based on the sensing data from sensor 110-2 on the probability distribution f^ of the object's position obtained through optimization calculation.

[0096] 7 is a diagram showing an image of the optimization calculation based on the index value based on the Kullback-Leibler distance, among the optimization calculations performed by the optimization calculation unit 287. ｉ (X ｋ |Z １：ｋ ｉ ) is a generalization of covariance intersection to multiple objects. In the example of FIG. 7, i=1 or i=2. Also, in the example of FIG. 7, k=1, 2, 3, ..., takes integer values. f ｉ (X ｋ |Z １：ｋ ｉ ) can be thought of as indicating a probability distribution.

[0097] The optimization calculation unit 287 obtains ω^ or obtains ω^ and f^, and in relation to the index value based on the Kullback-Leibler distance, the probability distribution f １ The Kullback-Leibler distance between and f^ and the probability distribution f ２ This can be understood as finding a probability distribution f^ such that the weighted average value of the Kullback-Leibler distance between f^ and f^, with weights ω^ and (1-ω^), is as small as possible.

[0098] The trajectory integration unit 288 integrates multiple tracking trajectories using the weight parameter values ​​calculated by the optimization calculation unit 287. The trajectory integration unit 288 integrates tracking trajectories that the tracking correlation calculation unit 284 has determined to be identical. For example, the trajectory integration unit 288 may calculate a weighted average of multiple probability distributions shown for each time step for multiple tracking trajectories that the tracking correlation calculation unit 284 has determined to be identical tracking trajectories using the weight parameter values ​​calculated by the optimization calculation unit 287. The trajectory integration unit 288 may calculate the weighted average of the multiple probability distributions, for example, by weighting the parameter values ​​of the probability distributions, such as the mean values ​​of the probability distributions.

[0099] The integration of tracking trajectories performed by the trajectory integration unit 288 can be considered as a process of estimating the trajectory of the object based on multiple candidates for the trajectory of the object. Therefore, the integration of tracking trajectories performed by the trajectory integration unit 288 can be considered as a process of estimating the position of the object based on multiple candidate points weighted by the distance index calculation unit 285.

[0100] The output control unit 291 controls the output of the estimation result of the position of the object by the sensing data processing device 200. The method by which the sensing data processing device 200 outputs the estimation result of the position of the object is not limited to a specific method. For example, the display unit 220 may display the estimated position or tracking trajectory of the object in a graph under the control of the output control unit 291. Alternatively, the communication unit 210 may transmit coordinate data of the estimated position of the object or time-series data of coordinates indicating the tracking trajectory of the object to another device under the control of the output control unit 291.

[0101] Fig. 8 is a diagram showing an example of the procedure of processing performed by the sensing data processing device 200. The sensing data processing device 200 repeats the processing of Fig. 8 for each time step. In the processing shown in Fig. 8, the sensing data acquisition unit 282 acquires sensing data from the sensor 110 (step S101).

[0102] Next, the tracking processing unit 283 performs tracking processing on the sensing data from the sensor 110 acquired by the sensing data acquisition unit 282 (step S102). As described above, the tracking processing is to associate the coordinates indicated by the sensing data with the tracking trajectory of the object. As described above, the tracking trajectory is a trajectory estimated by the sensing data processing device 200 as a history of the trajectory of the object. The tracking correlation calculation unit 284 generates a tracking trajectory in which candidates for the position of the object are indicated by a probability distribution.

[0103] Next, the tracking correlation calculation unit 284 calculates the correlation between the tracking trajectories, and determines that two tracking trajectories with a correlation stronger than a certain level are the same tracking trajectory (step S103). The tracking trajectories determined to be the same by the tracking correlation calculation unit 284 become targets for integration by the trajectory integration unit 288.

[0104] Next, the distance index calculation unit 285 calculates distance index values ​​of the two probability distributions of the object position candidates (step S104). The distance index used to calculate the index value by the distance index calculation unit 285 is used in the evaluation function used in the optimization calculation by the optimization calculation unit 287. The distance index value calculated by the distance index calculation unit 285 indicates the spatial proximity between the probability distribution of the object position candidates on one of the tracking trajectories to be integrated by the trajectory integration unit 288 and the probability distribution of the object position candidates on the tracking trajectory obtained by the integration.

[0105] Next, the distance index integrating unit 286 integrates the multiple types of distance index values ​​calculated by the distance index calculating unit 285 (step S105). For example, the distance index integrating unit 286 calculates a weighted average of the multiple types of distance index values ​​calculated by the distance index calculating unit 285 using weights set for each type of distance index.

[0106] Next, the optimization calculation unit 287 searches for values ​​of weight parameters for integrating the tracking trajectories using the evaluation function whose values ​​are calculated by the distance index calculation unit 285 and the distance index integrating unit 286 (step S106). The solution search performed by the optimization calculation unit 287 in step S106 corresponds to one solution search in the optimization calculation such that the evaluation function values ​​calculated by the distance index calculation unit 285 and the distance index integrating unit 286 are as small as possible.

[0107] Next, the optimization calculation unit 287 determines whether a termination condition for the optimization calculation is met (step S107). The termination condition here is not limited to a specific condition. For example, the termination condition here may be a condition that the decrease in the evaluation function value during the repetition of the loop from steps S104 to S107 is smaller than a predetermined condition. Alternatively, the termination condition here may be a condition that the loop from steps S104 to S107 has been repeated a predetermined number of times or more.

[0108] If the optimization calculation unit 287 determines that the termination condition is not satisfied (step S107: NO), the process returns to step S104. In this case, the optimization calculation unit 287 continues to perform optimization calculations using the evaluation function values ​​calculated by the distance index calculation unit 285 and the distance index integration unit 286.

[0109] On the other hand, if the optimization calculation unit 287 determines in step S107 that the termination condition is met (step S107: YES), the trajectory integration unit 288 integrates the tracking trajectories that the tracking correlation calculation unit 284 has determined to be identical (step S111). For example, the tracking correlation calculation unit 284 may perform weighted averaging based on the weights calculated by the optimization calculation unit 287 for each time step on the probability distributions indicated in the multiple tracking trajectories to be integrated. Then, the tracking correlation calculation unit 284 may use the tracking trajectory obtained by aggregating the probability distributions obtained for each time step as the integrated tracking trajectory.

[0110] After step S111, the output control unit 291 controls the output of the position estimation result by the position estimation unit 281 (step S112). The display unit 220 may display the position estimation result by the position estimation unit 281 under the control of the output control unit 291. Alternatively, the communication unit 210 may transmit the position estimation result by the position estimation unit 281 to another device under the control of the output control unit 291.

[0111] The display unit 220 or the communication unit 210 may output a tracking trajectory or the latest estimated position of the object. The display unit 220 or the communication unit 210 may output a probability distribution of the estimated position of the object as the estimated position of the object. Alternatively, the display unit 220 or the communication unit 210 may output coordinates of a point indicating the estimated position of the object, such as outputting the point with the highest probability density of the estimated position of the object as the estimated position of the object. After step S112, the sensing data processing device 200 ends the processing of FIG. 8.

[0112] As described above, the position estimation unit 281 acquires multiple coordinate data of candidate points based on the measurement results of the position of the object by the sensor 110, weights each of the multiple candidate points based on the position of the candidate point in the coordinate space of the sensor 110, and estimates the position of the object based on the multiple weighted candidate points.

[0113] According to the sensing data processing device 200, by weighting each of the multiple candidate points based on the position of the candidate point in the coordinate space of the sensor 110, it is possible to reflect in the estimation of the position of the object the difference in sensitivity of the sensor 110 due to the difference in the position of the candidate point. In this respect, it is expected that the sensing data processing device 200 can estimate the position of the object with relatively high accuracy.

[0114] In particular, according to the sensing data processing device 200, even when a distribution of candidate positions of an object is obtained from a single coordinate value obtained by measuring the position of the object using the sensor 110, differences in the sensitivity of the sensor due to differences in the position of the object in the distribution can be reflected in the position estimation.

[0115] Furthermore, with the sensing data processing device 200, it is possible to reflect various distributions, not limited to specific types of distributions such as normal distributions, in the estimation of the position of the object, with respect to the probability distribution of the position of the object. In this respect, it is expected that the sensing data processing device 200 will be able to estimate the position of the object with relatively high accuracy.

[0116] The position estimation unit 281 also weights multiple types of index values ​​related to the proximity of positions, including index values ​​calculated using the weighted candidate points, for each type of index, and estimates the position of the object based on the weighted index values. According to the sensing data processing device 200, the characteristics of multiple types of indexes related to the proximity of positions can be reflected in the position estimation of the object.

[0117] For example, consider a case where the first index provides a higher accuracy in measuring the position of an object in a certain region, and the second index provides a higher accuracy in measuring the position of an object in another region. In this case, by using both the first index and the second index, the sensing data processing device 200 is expected to be able to avoid an extreme decrease in the accuracy in measuring the position of the object in either region.

[0118] Furthermore, the sensing data processing device 200 can adjust the degree of influence of the characteristics of each index on the estimation of the position of the target object by weighting each type of index.

[0119] Furthermore, the position estimation unit 281 weights the candidate point based on the position of the candidate point in the coordinate space of the sensor 110, depending on the distance from the sensor 110 to the candidate point. The sensing data processing device 200 can estimate the position of the object taking into consideration differences in the measurement accuracy of the object's position due to differences in the distance from the sensor 110 to the candidate point. In this respect, the sensing data processing device 200 is expected to be able to estimate the position of the object with relatively high accuracy.

[0120] In particular, it is considered that the greater the distance from the sensor 110 to the candidate point, the lower the accuracy of measuring the position of the object by the sensor 110. When the sensor system 1 includes a plurality of sensors 110, it is expected that the sensing data processing device 200 can estimate the position of the object with higher accuracy by setting a relatively large weight for the sensor 110 that is close to the candidate point.

[0121] Furthermore, the position estimation unit 281 weights the candidate point based on its position in the coordinate space of the sensor 110, based on the position of the candidate point, the position of the cloud, and the degree to which the electromagnetic waves used by the sensor 110 are attenuated by the cloud. According to the sensing data processing device 200, the accuracy of measurement of the object by the sensor 110 at the position of the cloud can be reflected in the estimation of the position of the object, and in this respect, it is expected that the position of the object can be estimated with relatively high accuracy.

[0122] For example, if the sensor 110 is an infrared sensor, infrared rays do not penetrate clouds, so the accuracy of the infrared sensor may be lower than that of other types of sensors in areas with clouds. If the sensor system 1 includes multiple sensors 110 including infrared sensors, in areas with clouds, the weight for the sensor 110 that is an infrared sensor is set to be greater than the weight for the other sensors 110, and it is expected that the sensing data processing device 200 will be able to estimate the position of the target with higher accuracy.

[0123] The position estimation unit 281 also acquires coordinate data of multiple candidate points and searches for an estimated value of the object's position using an evaluation function that uses earthmover distance, which is calculated by weighting each of the multiple candidate points based on the position of the candidate in the coordinate space of the sensor 110. The sensing data processing device 200 can weight the candidate points using the earthmover distance. When the optimization calculation unit 287 performs optimization calculations, it can use a known algorithm for solving optimal transportation problems using earthmover distance.

[0124] Furthermore, the position estimation unit 281 samples multiple candidate points according to the probability distribution of the object's position. The sensing data processing device 200 can more accurately reflect the probability distribution of the object's position in the object's position estimation compared to, for example, a case in which candidate points are uniformly sampled from an area where the probability density of the object's position is equal to or greater than a certain value. In this respect, the sensing data processing device 200 is expected to be able to estimate the object's position with relatively high accuracy with a relatively small number of samples.

[0125] In addition, the position estimation unit 281 acquires data showing multiple probability distributions for the position of the object, acquires coordinate data for multiple candidate points for each probability distribution, and searches for an estimated value for the position of the object using an evaluation function that uses the Earthmover distance and the Kullback-Leibler distance related to the probability distribution.

[0126] According to the sensing data processing device 200, when the optimization calculation unit 287 performs optimization calculations, a known algorithm for solving optimal transportation problems using the earthmover distance can be used for the part of the evaluation function that uses the earthmover distance. Also, a gradient method such as the steepest descent method can be used for the part that uses the Kullback-Leibler distance.

[0127] Second Embodiment Fig. 9 is a diagram showing an example of the configuration of a sensing data processing device according to at least one embodiment. In the configuration shown in Fig. 9, a sensing data processing device 610 includes a position estimation unit 611.

[0128] With this configuration, the position estimation unit 611 acquires coordinate data for multiple candidate points, which are samples of points that are candidates for the position of the object, based on the measurement results of the position of the object by the sensor, weights each of the multiple candidate points based on the position of the candidate point in the coordinate space of the sensor, and estimates the position of the object based on the multiple weighted candidate points. The position estimation unit 611 is an example of a position estimation means.

[0129] The sensing data processing device 610 weights each of the multiple candidate points based on the position of the candidate point in the coordinate space of the sensor, so that the difference in sensor sensitivity due to the difference in the position of the candidate point can be reflected in the estimation of the position of the object. In this respect, the sensing data processing device 610 is expected to be able to estimate the position of the object with relatively high accuracy.

[0130] In particular, with the sensing data processing device 610, even when a distribution of candidate positions of an object is obtained from a single coordinate value obtained by measuring the position of the object with a sensor, differences in the sensitivity of the sensor due to differences in the position of the object in that distribution can be reflected in the position estimation.

[0131] Furthermore, with the sensing data processing device 610, it is possible to reflect various distributions, not limited to specific types of distribution such as a normal distribution, in the probability distribution of the object's position in the object's position estimation. In this respect, it is expected that the sensing data processing device 610 will be able to estimate the object's position with relatively high accuracy.

[0132] The position estimation unit 611 can be realized using the functions of the position estimation unit 281 in FIG. 2, for example.

[0133] Third Embodiment Fig. 10 is a diagram showing an example of the configuration of a sensor system according to at least one embodiment. In the configuration shown in Fig. 10, a sensor system 620 includes a sensor 621 and a position estimation unit 622.

[0134] In this configuration, sensor 621 measures the position of the object. Based on the measurement results by sensor 621, position estimation unit 622 acquires coordinate data for multiple candidate points, which are samples of points that are candidates for the position of the object, weights each of the multiple candidate points based on its position in the coordinate space of the sensor, and estimates the position of the object based on the multiple weighted candidate points. Position estimation unit 622 is an example of a position estimation means.

[0135] According to the sensor system 620, by weighting each of the multiple candidate points based on the position of the candidate point in the coordinate space of the sensor, it is possible to reflect the difference in sensitivity of the sensor due to the difference in the position of the candidate point in the target position estimation. In this respect, the sensor system 620 is expected to be able to estimate the target position with relatively high accuracy.

[0136] In particular, with sensor system 620, even when a distribution of candidate positions of an object is obtained from a single coordinate value obtained by measuring the position of the object with a sensor, differences in sensor sensitivity due to differences in the position of the object in that distribution can be reflected in the position estimation.

[0137] Furthermore, with the sensor system 620, it is possible to reflect various distributions, not limited to specific types of distributions such as normal distributions, in the estimation of the position of the object, with respect to the probability distribution of the position of the object. In this respect, it is expected that the sensor system 620 will be able to estimate the position of the object with relatively high accuracy.

[0138] The sensor 621 can be realized using, for example, the function of the sensor 110 in Fig. 1. The position estimation unit 622 can be realized using, for example, the function of the position estimation unit 281 in Fig. 2.

[0139] 11 is a diagram showing an example of a processing procedure in a sensing data processing method according to at least one embodiment. The sensing data processing method shown in FIG. 11 includes estimating a position (step S611).

[0140] In estimating the position (step S611), the computer acquires coordinate data for multiple candidate points, which are samples of points that are candidates for the position of the object, based on the measurement results of the object's position by the sensor, weights each of the multiple candidate points based on the position of the candidate point in the coordinate space of the sensor, and estimates the position of the object based on the weighted multiple candidate points.

[0141] According to the sensing data processing method shown in Fig. 11, by weighting each of a plurality of candidate points based on the position of the candidate point in the coordinate space of the sensor, it is possible to reflect differences in sensor sensitivity due to differences in the positions of the candidate points in the estimation of the position of the object. In this respect, it is expected that the sensing data processing method shown in Fig. 11 can estimate the position of the object with relatively high accuracy.

[0142] In particular, according to the sensing data processing method shown in FIG. 11, even when a distribution of candidate positions of an object is obtained from a single coordinate value obtained by measuring the position of the object with a sensor, differences in the sensitivity of the sensor due to differences in the position of the object in that distribution can be reflected in the position estimation.

[0143] Furthermore, according to the sensing data processing method shown in Fig. 11, it is possible to reflect various distributions in the estimation of the position of the object, not limited to a specific type of distribution such as a normal distribution, regarding the probability distribution of the position of the object. In this respect, it is expected that the sensing data processing method shown in Fig. 11 can estimate the position of the object with relatively high accuracy.

[0144] 12 is a schematic block diagram illustrating the configuration of a computer according to at least one embodiment. In the configuration shown in FIG. 12, a computer 700 includes a CPU (Central Processing Unit) 710, a main memory device 720, an auxiliary memory device 730, and an interface 740.

[0145] One or more of the sensing data processing device 200 and the sensing data processing device 610, or a part thereof, may be implemented in the computer 700. In this case, the operation of each of the above-described processing units is stored in the auxiliary storage device 730 in the form of a program. The CPU 710 reads the program from the auxiliary storage device 730, loads it into the main storage device 720, and executes the above-described processing in accordance with the program. The CPU 710 also allocates storage areas in the main storage device 720 corresponding to each of the above-described storage units in accordance with the program. Communication between each device and other devices is executed by the interface 740, which has a communication function, and performs communication under the control of the CPU 710.

[0146] When the sensing data processing device 200 is implemented in a computer 700, the operations of the processing unit 280 and each unit thereof are stored in the form of a program in the auxiliary storage device 730. The CPU 710 reads the program from the auxiliary storage device 730, loads it into the main storage device 720, and executes the above-described processing in accordance with the program.

[0147] Furthermore, the CPU 710 allocates a storage area for the storage unit 270 in the main storage device 720 in accordance with the program. Communication with other devices by the communication unit 210 is achieved by the interface 740 having a communication function and operating under the control of the CPU 710. Display of images by the display unit 220 is achieved by the interface 740 having a display device and displaying various images under the control of the CPU 710. Reception of user operations by the operation input unit 230 is achieved by the interface 740 having an input device and receiving user operations under the control of the CPU 710.

[0148] When the sensing data processing device 610 is implemented in the computer 700, the operation of the position estimation unit 611 is stored in the form of a program in the auxiliary storage device 730. The CPU 710 reads the program from the auxiliary storage device 730, loads it into the main storage device 720, and executes the above-described processing in accordance with the program.

[0149] Furthermore, the CPU 710 allocates a storage area in the main storage device 720 for the sensing data processing device 610 to perform processing in accordance with the program. Communication between the sensing data processing device 610 and other devices is achieved by the interface 740 having a communication function and performing communication under the control of the CPU 710. Interaction between the sensing data processing device 610 and a user is achieved by the interface 740 having a display device and an input device, displaying various images under the control of the CPU 710, and accepting user operations.

[0150] Alternatively, a program for executing all or part of the processing performed by the sensing data processing device 200 and the sensing data processing device 610 may be recorded on a computer-readable recording medium, and the program may be loaded into a computer system and executed to perform the processing of each component. Note that the term "computer system" here includes hardware such as an operating system (OS) and peripheral devices. Furthermore, the term "computer-readable recording medium" refers to portable media such as flexible disks, optical magnetic disks, read-only memories (ROMs), and compact disc read-only memories (CD-ROMs), as well as storage devices such as hard disks built into computer systems. The program may be designed to implement part of the aforementioned functions, or may be capable of implementing the aforementioned functions in combination with a program already stored in the computer system.

[0151] Although the embodiments have been described above, the specific configuration is not limited to these embodiments, and the present invention also includes designs that do not deviate from the gist of the present invention. Furthermore, the above-described embodiments can be combined with other embodiments as appropriate.

[0152] A part or all of the above-described embodiments can be described as, but not limited to, the following supplementary notes.

[0153] (Supplementary Note 1) A sensing data processing device comprising: a position estimation means for acquiring, based on a measurement result of a position of an object by a sensor, a plurality of coordinate data of candidate points which are samples of points that are candidates for the position of the object, weighting each of the plurality of candidate points based on the position of the candidate point in a coordinate space of the sensor, and estimating the position of the object based on the plurality of weighted candidate points.

[0154] (Supplementary Note 2) The sensing data processing device according to Supplementary Note 1, wherein the position estimation means weights a plurality of types of index values ​​relating to proximity of positions, including index values ​​calculated using the weighted candidate points, for each type of index, and estimates the position of the object based on the weighted index values.

[0155] (Supplementary Note 3) The sensing data processing device according to Supplementary Note 1 or Supplementary Note 2, wherein the position estimation means performs weighting based on a position of the candidate point in a coordinate space of the sensor, in accordance with a distance from the sensor to the candidate point.

[0156] (Supplementary Note 4) The sensing data processing device according to any one of Supplementary Notes 1 to 3, wherein the position estimation means weights the candidate points based on their positions in the coordinate space of the sensor, based on the positions of the candidate points, the positions of clouds, and the degree to which the electromagnetic waves used by the sensor are weakened by clouds.

[0157] (Supplementary Note 5) The sensing data processing device according to any one of Supplementary Notes 1 to 4, wherein the position estimation means acquires coordinate data of the plurality of candidate points, and searches for an estimated value of the position of the object using an evaluation function that uses an earthmover distance calculated by weighting each of the plurality of candidate points based on the position of the candidate in the coordinate space of the sensor.

[0158] (Supplementary Note 6) The sensing data processing device according to Supplementary Note 5, wherein the position estimation means samples the plurality of candidate points according to a probability distribution of the position of the object.

[0159] (Supplementary Note 7) The sensing data processing device according to Supplementary Note 5 or Supplementary Note 6, wherein the position estimation means acquires data indicating a plurality of probability distributions for the position of the object, acquires coordinate data of a plurality of the candidate points for each of the probability distributions, and searches for an estimated value of the position of the object using an evaluation function that uses the Earthmover distance and the Kullback-Leibler distance related to the probability distribution.

[0160] (Supplementary Note 8) A sensor system comprising: a sensor that measures the position of an object; and a position estimation means that acquires, based on the measurement results of the sensor, a plurality of coordinate data of candidate points that are samples of points that are candidates for the position of the object, weights each of the plurality of candidate points based on the position of the candidate point in a coordinate space of the sensor, and estimates the position of the object based on the weighted plurality of candidate points.

[0161] (Supplementary Note 9) The sensor system according to Supplementary Note 8, wherein the position estimation means weights a plurality of types of index values ​​relating to proximity of positions, including index values ​​calculated using the weighted candidate points, for each type of index, and estimates the position of the object based on the weighted index values.

[0162] (Supplementary Note 10) The sensor system according to Supplementary Note 8 or Supplementary Note 9, wherein the position estimation means weights the candidate point based on a position of the candidate point in a coordinate space of the sensor, depending on a distance from the sensor to the candidate point.

[0163] (Supplementary Note 11) The sensor system according to any one of Supplementary Notes 8 to 10, wherein the position estimation means weights the candidate points based on their positions in a coordinate space of the sensor, based on the positions of the candidate points, the positions of clouds, and the degree to which electromagnetic waves used by the sensor are attenuated by clouds.

[0164] (Supplementary Note 12) The sensor system according to any one of Supplementary Notes 8 to 11, wherein the position estimation means acquires coordinate data of the plurality of candidate points, and searches for an estimated value of the position of the object using an evaluation function that uses an earthmover distance calculated by weighting each of the plurality of candidate points based on the position of the candidate in the coordinate space of the sensor.

[0165] (Supplementary Note 13) The sensor system according to Supplementary Note 12, wherein the position estimation means samples the plurality of candidate points according to a probability distribution of the position of the object.

[0166] (Supplementary Note 14) The sensor system according to Supplementary Note 12 or Supplementary Note 13, wherein the position estimation means acquires data indicating a plurality of probability distributions for the position of the object, acquires coordinate data of a plurality of the candidate points for each of the probability distributions, and searches for an estimated value of the position of the object using an evaluation function that uses the Earthmover distance and the Kullback-Leibler distance related to the probability distribution.

[0167] (Supplementary Note 15) A sensing data processing method including: a computer, based on a measurement result of the position of an object by a sensor, acquiring a plurality of coordinate data of candidate points which are samples of points that are candidates for the position of the object, weighting each of the plurality of candidate points based on the position of the candidate point in the coordinate space of the sensor, and estimating the position of the object based on the plurality of weighted candidate points.

[0168] (Supplementary Note 16) The sensing data processing method described in Supplementary Note 15, wherein estimating the position includes the computer weighting multiple types of index values ​​related to proximity of positions, including index values ​​calculated using the weighted candidate points, for each type of index, and estimating the position of the object based on the weighted index values.

[0169] (Supplementary Note 17) The sensing data processing method according to Supplementary Note 15 or Supplementary Note 16, wherein estimating the position includes weighting, by the computer, the candidate point based on its position in the coordinate space of the sensor, depending on the distance from the sensor to the candidate point.

[0170] (Supplementary Note 18) The sensing data processing method according to any one of Supplementary Notes 15 to 17, wherein estimating the position includes weighting the candidate point based on its position in the coordinate space of the sensor, based on the position of the candidate point, the position of clouds, and the degree to which the electromagnetic waves used by the sensor are attenuated by clouds.

[0171] (Supplementary Note 19) The sensing data processing method according to any one of Supplementary Notes 15 to 18, wherein estimating the position includes the computer acquiring coordinate data of a plurality of the candidate points, and searching for an estimated value of the position of the object using an evaluation function that uses an earthmover distance calculated by weighting each of the plurality of candidate points based on the position of the candidate in the coordinate space of the sensor.

[0172] (Supplementary Note 20) The sensing data processing method according to Supplementary Note 19, wherein estimating the position includes the computer sampling the plurality of candidate points according to a probability distribution of the position of the object.

[0173] (Supplementary Note 21) The sensing data processing method described in Supplementary Note 19 or Supplementary Note 20, wherein estimating the position includes the computer acquiring data indicating a plurality of probability distributions for the position of the object, acquiring coordinate data of a plurality of the candidate points for each of the probability distributions, and searching for an estimated value of the position of the object using an evaluation function that uses the Earthmover distance and the Kullback-Leibler distance related to the probability distribution.

[0174] (Supplementary Note 22) A recording medium having recorded thereon a program that causes a computer to execute the following: based on the measurement results of the position of an object by a sensor, acquire multiple coordinate data of candidate points that are samples of points that are candidates for the position of the object, weight each of the multiple candidate points based on the position of the candidate point in the coordinate space of the sensor, and estimate the position of the object based on the multiple weighted candidate points.

[0175] (Supplementary Note 23) The recording medium described in Supplementary Note 22, wherein the program causes the computer to weight multiple types of index values ​​related to proximity of positions, including index values ​​calculated using the weighted candidate points, for each type of index, and estimate the position of the object based on the weighted index values.

[0176] (Supplementary Note 24) The recording medium according to Supplementary Note 22 or Supplementary Note 23, wherein in estimating the position, the program causes the computer to perform weighting based on the position of the candidate point in a coordinate space of the sensor, depending on the distance from the sensor to the candidate point.

[0177] (Appendix 25) The recording medium according to any one of Appendices 22 to 24, wherein in estimating the position, the program causes the computer to weight the candidate point based on its position in the coordinate space of the sensor, based on the position of the candidate point, the position of clouds, and the degree to which the electromagnetic waves used by the sensor are weakened by clouds.

[0178] (Appendix 26) The recording medium described in any one of Appendices 22 to 25, wherein in estimating the position, the program causes the computer to acquire coordinate data of a plurality of the candidate points, and search for an estimated value of the position of the object using an evaluation function that uses an earthmover distance calculated by weighting each of the plurality of candidate points based on the position of the candidate in the coordinate space of the sensor.

[0179] (Supplementary Note 27) The recording medium according to Supplementary Note 26, wherein in estimating the position, the program causes the computer to sample the plurality of candidate points according to a probability distribution of the position of the object.

[0180] (Appendix 28) The recording medium described in Appendix 26 or Appendix 27, wherein in estimating the position, the program causes the computer to acquire data indicating a plurality of probability distributions for the position of the object, acquire coordinate data for a plurality of the candidate points for each of the probability distributions, and search for an estimated value of the position of the object using an evaluation function that uses the Earthmover distance and the Kullback-Leibler distance related to the probability distribution.

[0181] The present invention may be applied to a sensing data processing device, a sensor system, a sensing data processing method, and a recording medium.

[0182] 1, 620 Sensor system 100 Sensor group 110 Sensor 200, 610 Sensing data processing device 210 Communication unit 220 Display unit 230 Operation input unit 270 Storage unit 280 Processing unit 281, 611, 622 Position estimation unit 282 Sensing data acquisition unit 283 Tracking processing unit 284 Tracking correlation calculation unit 285 Distance index calculation unit 286 Distance index integration unit 287 Optimization calculation unit 288 Trajectory integration unit 291 Output control unit

Claims

1. A sensing data processing device comprising: a position estimation means for acquiring multiple pieces of coordinate data of candidate points which are samples of points that are candidates for the position of an object based on the measurement results of the position of the object by a sensor, weighting each of the multiple candidate points based on the position of the candidate point in the coordinate space of the sensor, and estimating the position of the object based on the multiple weighted candidate points.

2. The sensing data processing device according to claim 1, wherein the position estimation means weights multiple types of index values ​​relating to positional proximity, including index values ​​calculated using the weighted candidate points, for each type of index, and estimates the position of the target object based on the weighted index values.

3. The sensing data processing device according to claim 1 or 2, wherein said position estimation means performs weighting based on the position of said candidate point in the coordinate space of said sensor, depending on the distance from said sensor to said candidate point.

4. A sensing data processing device as claimed in any one of claims 1 to 3, wherein the position estimation means weights the candidate point based on its position in the coordinate space of the sensor, based on the position of the candidate point, the position of the cloud, and the degree to which the electromagnetic waves used by the sensor are weakened by the cloud.

5. A sensing data processing device as claimed in any one of claims 1 to 4, wherein the position estimation means acquires coordinate data of a plurality of the candidate points, and searches for an estimated value of the position of the object using an evaluation function that uses an earthmover distance calculated for each of the plurality of candidate points by weighting the candidate point based on its position in the coordinate space of the sensor.

6. The sensing data processing device according to claim 5, wherein said position estimation means samples a plurality of said candidate points according to a probability distribution of the position of said object.

7. A sensing data processing device as described in claim 5 or claim 6, wherein the position estimation means acquires data indicating a plurality of probability distributions for the position of the object, acquires coordinate data for a plurality of the candidate points for each of the probability distributions, and searches for an estimate of the position of the object using an evaluation function using the Earthmover distance and the Kullback-Leibler distance related to the probability distribution.

8. A sensor system comprising: a sensor for measuring a position of an object; and a position estimation means for acquiring a plurality of coordinate data of candidate points which are samples of points that are candidates for the position of the object based on the measurement results by the sensor, weighting each of the plurality of candidate points based on its position in the coordinate space of the sensor, and estimating the position of the object based on the plurality of weighted candidate points.

9. A sensing data processing method comprising: a computer acquiring, based on the measurement results of the position of an object by a sensor, a plurality of coordinate data of candidate points which are samples of points that are candidates for the position of the object, weighting each of the plurality of candidate points based on the position of the candidate point in the coordinate space of the sensor, and estimating the position of the object based on the plurality of weighted candidate points.

10. A recording medium having recorded thereon a program that causes a computer to execute the following: based on the measurement results of an object's position by a sensor, obtain multiple coordinate data of candidate points that are samples of points that are candidates for the object's position, weight each of the multiple candidate points based on its position in the coordinate space of the sensor, and estimate the object's position based on the multiple weighted candidate points.

Images