Mobile object tracking system, mobile object tracking method, and program

Abstract

To improve the real-time tracking of mobile MOs based on signals output from optical fibers. [Solution] A mobile object tracking system is configured comprising: a strain distribution waveform acquisition unit that acquires a strain distribution waveform showing the spatial distribution of strain obtained from backscattered light generated in response to light incident on an optical fiber laid in an environment in which a mobile object is moving, at predetermined intervals; and a mobile object detection unit that performs mobile object detection, including detection of the presence of a mobile object and detection of mobile object attributes including the location of the mobile object, based on the strain distribution waveform acquired at predetermined intervals by the strain distribution waveform acquisition unit.

Description

[Technical Field]

[0001] The present invention relates to a mobile object tracking system, a mobile object tracking method, and a program. [Background technology]

[0002] A known technique predicts the characteristics of vehicle traffic flow based on driving vibration trajectory data (waterfall data) generated from vibration signals detected using optical fibers laid along roads. [Prior art documents] [Patent Documents]

[0003] [Patent Document 1] Patent No. 7351357 [Overview of the project] [Problems that the invention aims to solve]

[0004] Waterfall data is continuous data over time. Therefore, tracking a moving object (MO) based on waterfall data will output results from a time period somewhat past the actual situation, which means it lacks immediacy.

[0005] This invention has been made in view of these circumstances, and aims to improve the real-time tracking of a mobile MO based on signals output from an optical fiber. [Means for solving the problem]

[0006] One aspect of the present invention that solves the above-mentioned problems is a mobile tracking system comprising: a strain distribution waveform acquisition unit that acquires a strain distribution waveform showing the spatial distribution of strain obtained from backscattered light generated in response to light incident on an optical fiber laid in an environment in which a mobile object is moving, at predetermined time intervals; and a mobile object detection unit that performs mobile object detection, including detection of the presence of a mobile object and detection of mobile object attributes including the location of the mobile object, based on the strain distribution waveform acquired at predetermined time intervals by the strain distribution waveform acquisition unit.

[0007] One aspect of the present invention is a mobile object tracking method in a mobile object tracking system, comprising: a strain distribution waveform acquisition step in which a strain distribution waveform acquisition unit acquires a strain distribution waveform showing the spatial distribution of strain obtained from backscattered light generated in response to light incident on an optical fiber laid in the environment in which the mobile object is moving, at predetermined time intervals; and a mobile object detection step in which a mobile object detection unit performs mobile object detection, which includes detecting the presence of a mobile object and detecting mobile object attributes including the location of the mobile object, based on the strain distribution waveform acquired at predetermined time intervals by the strain distribution waveform acquisition step.

[0008] One aspect of the present invention is a program for causing a computer in a mobile object tracking system to function as a strain distribution waveform acquisition unit that acquires a strain distribution waveform showing the spatial distribution of strain obtained from backscattered light generated in response to light incident on an optical fiber laid in the environment in which a mobile object is moving, at predetermined intervals, and a mobile object detection unit that performs mobile object detection, including detection of the presence of a mobile object and detection of mobile object attributes including the location of the mobile object, based on the strain distribution waveform acquired at predetermined intervals by the strain distribution waveform acquisition unit. [Effects of the Invention]

[0009] As described above, the present invention provides the effect of improving the immediacy of tracking a mobile MO based on signals output from an optical fiber. [Brief explanation of the drawing]

[0010] [Figure 1]It is a diagram showing an overall configuration example of a moving object tracking system in the first embodiment. [Figure 2] It is a diagram showing a functional configuration example of a moving object tracking device in the first embodiment. [Figure 3] It is a diagram for explaining a specific example of obtaining a strain distribution waveform by a strain distribution waveform acquisition unit in the first embodiment. [Figure 4] It is a diagram showing a specific example of a detection process of a moving object by a moving object detection unit in the first embodiment. [Figure 5] It is a flowchart showing an example of a processing procedure executed by the moving object tracking device in the first embodiment in relation to tracking a moving object. [Figure 6] It is a diagram showing a functional configuration example of a moving object tracking device in the second embodiment. [Figure 7] It is a diagram schematically showing a machine learning process executed by a learning unit in the second embodiment to create a moving object model. [Figure 8] It is a diagram showing an example of a processing procedure executed by the moving object tracking device in the second embodiment in relation to tracking a moving object. [Figure 9] It is a diagram schematically showing a machine learning process executed by a learning unit in the third embodiment to create a moving object model.

Embodiments for Carrying out the Invention

[0011] <First Embodiment> [Overall Configuration Example of Event Identification System] FIG. 1 shows an overall configuration example of the moving object tracking system of the present embodiment. The moving object tracking system in the figure includes an optical fiber 10 and a moving object tracking device 100.

[0012] The figure shows an example in which the optical fiber 10 is laid on a road RD. The optical fiber 10 is a sensor that detects the dynamic strain of the optical fiber 10 caused by vibration and sound corresponding to the movement of a mobile body MO as a vehicle on a travel surface such as a road RD, as a change in an optical signal. In this case, the optical fiber 10 may be laid along the direction of travel of the mobile body MO moving on the road RD. The end of the optical fiber 10 laid on the road RD is connected to a mobile body tracking device 100.

[0013] The mobile object tracking device 100 tracks the mobile object MO moving on the road RD based on signals input from the optical fiber 10. In the figure, the mobile object tracking device 100 is shown as a single device, but it may also be composed of multiple devices, for example, each with a predetermined function distributed among them. Furthermore, the mobile object tracking device 100 may be located on a network or configured as a cloud server or the like.

[0014] [Example of a mobile object tracking device's functional configuration] Figure 2 shows an example of the functional configuration of the mobile object tracking device 100. The mobile object tracking device 100 may be configured with hardware such as a CPU (Central Processing Unit), ROM (Read Only Memory), RAM (Random Access Memory), and storage devices such as an HDD (Hard Disk Drive) or SSD (Solid State Drive). The mobile object tracking device 100 may also be equipped with a GPU (Graphics Processing Unit) as hardware. Furthermore, the mobile object tracking device 100 may be equipped with an FPGA (Field Programmable Gate Array) as hardware. The functions of the mobile object tracking device 100 shown in the figure are realized by the CPU and GPU in the mobile object tracking device 100 executing programs, or by the operation of an FPGA whose circuit configuration is set by a program.

[0015] The mobile tracking device 100 in the figure comprises a tracking processing unit 101 and a storage unit 102. The tracking processing unit 101 performs processing related to tracking the moving object MO. The tracking processing unit 101 includes an input unit 111, a distortion distribution waveform acquisition unit 112, and a moving object detection unit 113.

[0016] The input unit 111 receives a signal (optical signal) from the optical fiber 10. The input unit 111 causes pulsed light to enter the optical fiber 10. In the optical fiber 10, the backscattered light generated as the incident pulsed light is transmitted returns to the incident side as reflected light via the same optical fiber 10. The input unit 111 receives the optical signal as reflected light.

[0017] On the road RD, vibrations and sounds are generated as the moving object MO moves. When these vibrations and sounds are transmitted to the optical fiber 10 via the ground or air, dynamic strain occurs in the optical fiber 10. The dynamic strain in the optical fiber 10 affects the transmitted backscattered light, and also affects the optical signal as reflected light input to the input unit 111. In other words, the optical signal input to the input unit 111 contains strain information corresponding to the vibrations and sounds generated by the movement of the moving object MO on the road RD.

[0018] The distortion distribution waveform acquisition unit 112 acquires a distortion distribution waveform from the optical signal input to the input unit 111. The distortion distribution waveform acquisition unit 112 acquires an optical signal as reflected light from the input unit 111. The distortion distribution waveform acquisition unit 112 analyzes the acquired optical signal to estimate the phase change along the length of the optical fiber, and obtains a distortion distribution waveform from the phase change along the length of the optical fiber. The distortion distribution waveform acquisition unit 112 may perform predetermined post-processing on the acquired distortion distribution waveform. Post-processing may include, for example, smoothing, short-time Fourier transform, spatial Fourier transform, and calculation of the signal-to-noise ratio based on a moving average or moving standard deviation. The distortion distribution waveform acquisition unit 112 acquires a distortion distribution waveform corresponding to the current time from the post-processed optical signal.

[0019] [Regarding the acquisition of distortion distribution waveforms] Referring to Figure 3, a specific example of acquiring a distortion distribution waveform by the distortion distribution waveform acquisition unit 112 will be explained. Figure 3 shows an example of waterfall data obtained based on the optical signal input to the input unit 111. In the waterfall data shown in the figure, the horizontal axis represents the distance of the optical fiber 10, and the vertical axis represents time. The distance on the horizontal axis may be the distance from the end of the optical fiber 10 connected to the mobile object tracking device 100. The waterfall data shows the dynamic strain of the optical fiber, obtained from the change in backscattered light over time at predetermined lengths (positions) along the optical fiber length. The backscattered light changes in accordance with the dynamic strain of the optical fiber 10 caused by vibrations or sounds corresponding to the movement of the mobile object MO on the road RD.

[0020] The distortion distribution waveform acquisition unit 112 acquires the reflected light at each position along the length of the optical fiber 10 from the optical signal at the current time acquired from the input unit 111, analyzes the reflected light to estimate the phase change along the length of the optical fiber, and forms a distortion distribution waveform from the phase change along the length of the optical fiber. Thus, the collection of phase differences of the reflected light at each position of the optical fiber 10, obtained at time t shown on the vertical axis of Figure 3, becomes the distortion distribution waveform. Figure 3 shows an example in which a dynamic distortion waveform was obtained from the time change of the phase difference of the reflected light at each position of the optical fiber 10. The distortion distribution waveform acquisition unit 112 forms a distortion distribution waveform by a phase change along the length of the optical fiber of the return light acquired corresponding to the current time, for example, every certain period of time (for example, 25 ms). In other words, the distortion distribution waveform acquisition unit 112 acquires the distortion distribution waveform in real time at regular intervals. The distortion distribution waveform obtained in this way shows the phase difference of the returned light, which changes according to the dynamic distortion of the optical fiber 10 caused by vibrations and sounds generated in response to the movement of the moving object MO, for each position in space corresponding to the length of the optical fiber 10 at each time point. For example, the strain (ε) of the optical fiber 10 is expressed by the following equation 1. ε = λθ / 4πnGLξ...(Equation 1) In equation 1 above, "λ" is the wavelength, "θ" is the phase difference, "n" is the refractive index, "GL" is the unit length for measuring the strain / phase difference, and "ξ" is the coefficient.

[0021] For example, the invention described in Patent Document 1 performed vehicle traffic flow analysis based on waterfall data shown in Figure 3. In contrast, in this embodiment, instead of using waterfall data, the distortion distribution waveform generated by the distortion distribution waveform acquisition unit 112 at predetermined time intervals is used for tracking the moving object.

[0022] The moving object detection unit 113 uses the strain distribution waveform obtained at regular intervals by the strain distribution waveform acquisition unit 112 to perform processing related to the detection of the moving object MO.

[0023] [Specific examples of moving object detection] Referring to Figure 4, a specific example of the detection process of a moving object MO by the moving object detection unit 113 will be described. Figure 4(A) shows the strain distribution waveform acquired by the strain distribution waveform acquisition unit 112 at time t0, which corresponds to the start of moving object detection. In this figure, the horizontal axis represents the position corresponding to the length of the optical fiber 10, and the vertical axis represents the amplitude. The moving object detection unit 113 detects the location of the moving object MO based on the shape of the distortion distribution waveform shown in the figure. In the distortion distribution waveform, the waveform portion corresponding to the location of the moving object MO shows distortion corresponding to the moving object MO. The moving object detection unit 113 detects the location of the moving object MO based on the distortion shape in the distortion distribution waveform. In the figure, an example is shown in which the presence of the moving object MO is detected at position p0. In other words, at time t0, the moving object detection unit 113 acquires position p0 as a moving object parameter as a moving object detection result.

[0024] The moving object detection unit 113 sets the range (detection range) of the position on the coordinate axis of the position of the distortion distribution waveform to be detected (position in the longitudinal direction of the optical fiber 10) so that the moving object MO first detected at time t0 can be efficiently detected at the next time t1. In this case, the moving object detection unit 113 may set the detection range at time t1 based on, for example, the speed limit on the target road RD or the normal average travel speed of the moving object MO on the target road RD.

[0025] Figure 4(B) shows the distortion distribution waveform acquired by the distortion distribution waveform acquisition unit 112 at time t1, which is a predetermined time (e.g., 25 ms) after time t0. In this figure, the detection range BD1 that was set in accordance with time t1 when the moving object detection unit 113 detected a vehicle at time t0 is shown. In this case, the moving object detection unit 113 detects the moving object MO by targeting the waveform portion within the detection range BD1 of the distortion distribution waveform acquired at time t1. The figure shows the result of detecting that the moving object MO is located at position p1, which is a certain distance from position p0. Furthermore, since the moving object MO was detected at time t1, the time from time t0 to time t1 and the distance traveled from position p0 to position p1 are obtained. Therefore, the moving object detection unit 113 may also calculate the velocity v1 of the moving object MO based on the time from time t0 to time t1 and the distance traveled from position p0 to position p1. In this way, at time t1, when the second moving object MO is detected, the moving object detection unit 113 may acquire the position p1 and velocity v1 as moving object parameters as the moving object detection result for the moving object MO.

[0026] The moving object detection unit 113 sets the detection range to be used for moving object detection at time t2 so that the moving object MO detected at time t1 can be efficiently detected at the next time t2. In this case, the moving object detection unit 113 may set the detection range corresponding to time t2 using, for example, the velocity v1 of the moving object MO calculated at time t1. The detection range corresponding to time t2 set in this way can be set to be narrower than the detection range BD1 corresponding to time t1. By setting a narrower detection range, the possibility of false detection due to the influence of noise can be reduced.

[0027] Figure 4(C) shows the distortion distribution waveform acquired by the distortion distribution waveform acquisition unit 112 at time t2, which is a predetermined time (e.g., 25 ms) after time t1. In this figure, the detection range BD2 set by the moving object detection unit 113 in accordance with time t2 is shown. In this case, the moving object detection unit 113 detects the moving object MO by targeting the waveform portion within the detection range BD2 of the distortion distribution waveform acquired at time t2. The figure shows the result of detecting that the moving object MO is located at position p2, which is a certain distance away from position p1. Furthermore, since the moving object MO was detected at time t2, the time from time t1 to time t2 and the distance traveled from position p1 to position p2 are obtained. Therefore, the moving object detection unit 113 may also calculate the velocity v2 of the moving object MO based on the time from time t1 to time t2 and the distance traveled from position p1 to position p2. Furthermore, at the stage when the moving object MO is detected at time t2, the velocity v1 corresponding to time t1 and the velocity v2 corresponding to time t2 are obtained. Therefore, the moving object detection unit 113 may calculate the acceleration a2 of the moving object MO using the velocity difference between velocity v1 and velocity v2 and the time from time t1 to time t2. In this way, at time t2, when the third moving object MO is detected, the moving object detection unit 113 may acquire the position p2, velocity v2, and acceleration a2 as moving object parameters as the moving object detection result for the moving object MO.

[0028] The moving object detection unit 113 sets the detection range to be used for moving object detection at time t3 so that the moving object MO detected at time t2 can be efficiently detected at the next time t3. In this case, the moving object detection unit 113 may set the detection range corresponding to time t3 using, for example, the velocity v2 and acceleration a2 calculated at time t2, or using either the velocity v2 or acceleration a2. The detection range corresponding to time t3 set in this way can be set to be narrower than the detection range BD2 corresponding to time t2.

[0029] Figure 4(D) shows the distortion distribution waveform acquired by the distortion distribution waveform acquisition unit 112 at time t3, which is a predetermined time (e.g., 25 ms) after time t2. In this figure, the detection range BD3 set by the moving object detection unit 113 in accordance with time t3 is shown. In this case, the moving object detection unit 113 detects the moving object MO by targeting the waveform portion within the detection range BD3 of the distortion distribution waveform acquired at time t3. The figure shows the result of detecting that the moving object MO is located at position p3, which is a certain distance away from position p2. Furthermore, the moving object detection unit 113 may also calculate the velocity v3 of the moving object MO based on the time from time t2 to time t3 and the distance traveled from position p2 to position p3. Furthermore, the moving object detection unit 113 may calculate the acceleration a3 of the moving object MO based on the speed difference between speed v2 and speed v3 and the time from time t2 to time t3. In this way, at time t3, when the fourth moving object MO is detected, the moving object detection unit 113 may acquire the position p3, velocity v3, and acceleration a3 as moving object parameters for the moving object MO as a moving object detection result.

[0030] Furthermore, the moving object detection unit 113 sets the detection range to be used for detecting a moving object at time t4, which follows time t3. In this case, the moving object detection unit 113 may set the detection range corresponding to time t4 using, for example, the velocity v3 and acceleration a3 calculated at time t3, or using either the velocity v3 or acceleration a3.

[0031] Figure 4(E) shows the distortion distribution waveform acquired by the distortion distribution waveform acquisition unit 112 at time t4, which is a predetermined time (e.g., 25 ms) after time t3. In this figure, the detection range BD4 set by the moving object detection unit 113 in accordance with time t4 is shown. In this case, the moving object detection unit 113 detects the moving object MO by targeting the waveform portion within the detection range BD4 of the distortion distribution waveform acquired at time t4. The figure shows the result of detecting that the moving object MO is located at position p4, which is a certain distance away from position p3. Furthermore, the moving object detection unit 113 may also calculate the velocity v4 of the moving object MO based on the time from time t3 to time t4 and the distance traveled from position p3 to position p4. Furthermore, the moving object detection unit 113 may calculate the acceleration a4 of the moving object MO based on the speed difference between speed v3 and speed v4 and the time from time t3 to time t4. In this way, at time t4, when the fifth moving object MO is detected, the moving object detection unit 113 may acquire the position p4, velocity v4, and acceleration a4 as moving object parameters as a result of the moving object detection targeting the moving object MO.

[0032] Subsequently, the moving object detection unit 113 sets a detection range corresponding to the next time, detects the moving object MO from the distortion distribution waveform obtained when the next time becomes the current time, and acquires the position, velocity, and acceleration as moving object parameters for each time. In this way, the moving object detection unit 113 can track the moving object MO moving along the road RD by acquiring moving object parameters, including its position on the road RD.

[0033] Furthermore, the moving object detection unit 113 may also detect the type and weight of the detected moving object MO, as well as the condition of the road RD at the location where the moving object MO was detected (such as steps or road surface conditions), based on the shape of the strain distribution waveform, and include these in the moving object parameters.

[0034] [Example of processing procedure] Referring to the flowchart in Figure 5, an example of the processing procedure performed by the mobile object tracking device 100 in relation to tracking the mobile object MO will be described.

[0035] Step S100: In the moving object tracking device 100, the moving object detection unit 113 initializes the variable n, which indicates the number to be assigned to the time, to "0".

[0036] Step S102: The distortion distribution waveform acquisition unit 112 acquires the current time t from the optical signal input to the input unit 111. n Obtain the corresponding distortion distribution waveform.

[0037] Step S104: The moving object detection unit 113 uses the time t obtained in step S102. n The system performs a process to detect a moving object MO from the distortion distribution waveform. Based on the results of the process to detect the moving object MO, the moving object detection unit 113 determines whether or not a moving object MO has been detected. Note that the time t obtained in step S102 n Depending on the presence of moving objects MO on the road RD, the distortion distribution waveform may correspond to the presence of multiple moving objects MO. In such cases, the moving object detection unit 113 can detect multiple moving objects MO from the distortion distribution waveform.

[0038] Step S106: When the moving object MO is detected in step S104, the moving object detection unit 113 acquires the moving object parameters as a detection result.

[0039] If step S106 is executed in accordance with the time t0 of the first (1st) detection of a moving object, the moving object detection unit 113 may acquire moving object parameters including position p0. Furthermore, if step S106 is performed in accordance with time t1, which corresponds to the second detection of a moving object, the moving object detection unit 113 may acquire moving object parameters including position p1 and velocity v1. Furthermore, step S106 occurs at time t2 or later, when a moving object is detected for the third time or later. n When executed in accordance with, the moving object detection unit 113 determines position pn and speed v n and acceleration a n and may obtain moving body parameters including the same. Further, the moving body detection unit 113 may include, in addition to the position p n , speed v n , and acceleration a n other vehicle types (moving body MO types), the weight of the moving body MO, and road conditions as described above in the moving body parameters.

[0040] Step S108: The moving body detection unit 113 generates or updates a moving body object having the moving body parameters acquired in step S106. In step S108 when the first detection of the moving body MO is performed corresponding to time t0, the moving body detection unit 113 newly generates a moving body object. The moving body detection unit 113 may store the generated moving body object in the storage unit 102. In step S108 when the second and subsequent detections of the moving body MO corresponding to after time t1 are performed, the moving body detection unit 113 updates the moving body object based on the moving body parameters acquired in the current step S106. Note that the moving body detection unit 113 may generate a moving body object based on the moving body parameters acquired in the current step S106 each time the moving body MO is detected, and store the generated moving body objects in the storage unit 102 corresponding to the respective times.

[0041] Step S110: After the processing of step S108 or when it is determined in step S114 described later that the state where the moving body MO is not detected has not continued for a predetermined time, the moving body detection unit 113 sets the detection range BD for the next time. For the detection range BD1 corresponding to the second moving body detection, the moving body detection unit 113 may set it based on the legal speed defined on the road on which the detection target moving body MO travels or the standard speed of the moving body MO on the road RD where the detection of the detection target moving body MO is performed. Furthermore, the detection range BD2 corresponding to the third detection of a moving object in the moving object detection unit 113 may be set by calculating the distance traveled based on the speed obtained as a moving object parameter in the most recent moving object detection. Furthermore, the moving object detection unit 113 may set the detection range BD3 corresponding to the fourth moving object detection by determining the distance traveled based on the velocity and acceleration obtained as moving object parameters in the most recent moving object detection. For the fifth and subsequent detections of a moving object, the detection range may be set based on the distance traveled, calculated based on velocity, acceleration, and the time difference from the current time to the next time, similar to the detection range BD3 for the fourth detection of a moving object.

[0042] Step S112: After processing in step S110, the moving object detection unit 113 increments the variable n and returns to step S102. This process results in the next time t n A moving object detection is performed, and the moving object object is updated in response to the detection of a moving object (MO).

[0043] Step S114: If it is determined in step S104 that no moving object MO was detected, the moving object detection unit 113 will determine the current time t n At this point, it is determined whether or not the state continued for a predetermined time without the moving object MO being detected. n The state in which no moving object MO is detected for a predetermined time is defined as the state in which the moving object generated or updated in step S108 is in the past time t (n-k) (k is any natural number) from the current time t n At each time point up to t, step S104 has determined that no moving object MO is detected within the detection range corresponding to moving object detection. This state, where no moving object MO is detected for a predetermined period of time, is equivalent to confirming that no moving object MO exists. For example, at one time point t nAt this timing, due to the state of the laying of the optical fiber 10 or external disturbances, the detection of the mobile object MO by the optical fiber 10 may fail even though the mobile object MO is actually present. Therefore, in this embodiment, if the mobile object MO is not detected in step S104, step S114 further determines that the mobile object MO does not exist if the state of not detecting the mobile object MO continues for a predetermined time, thereby increasing the reliability of detecting the presence of the mobile object MO. If it is determined in step S114 that the state of not detecting a moving object MO for a predetermined time has not yet continued, the process proceeds to step S110, where the detection range BD for the next time is set, the variable n is incremented in step S112, and the detection of the moving object MO corresponding to the next time is performed by the processing from step S104 onward.

[0044] Step S116: If it is determined in step S114 that a predetermined time has passed without a moving object MO being detected, the moving object detection unit 113 determines whether or not a moving object has been generated. If a moving object has not yet been created, it means that no moving object (MO) has been detected since the start of the moving object detection process. In this case, the process returns to step S100, and the moving object detection corresponding to time t0 is executed again.

[0045] Step S118: On the other hand, if it is determined in step S116 that a moving object has already been created, it means that the moving object MO had been tracked up to this point, but for some reason, such as the moving object MO having moved out of the detection range at the present time, it has been confirmed that the moving object MO that had been tracked up to this point is now outside the detection range. In this case, the moving object detection unit 113 performs a process to invalidate the moving object MO that had been tracked by the moving object detection process up to this point. The invalidation of a mobile object here may, for example, be the deletion of a mobile object that has been stored in the storage unit 102 while being updated according to the previous mobile object detection process. Alternatively, the invalidation of a mobile object may be managed so that the mobile object that has been updated according to the previous mobile object detection process is stored in the storage unit 102 but is not subject to update. In this case, the mobile objects that have been stored in the storage unit 102 may be managed in the storage unit 102 as history information of previous mobile object detections.

[0046] As can be understood from the above explanation, the detection of the moving object MO in this embodiment is performed, for example, based on a distortion distribution waveform acquired at predetermined time intervals corresponding to the current time. A known technique for detecting moving MOs using optical signals input from optical fibers employs waterfall data generated from the optical signal. However, generating waterfall data requires an optical signal with a certain time duration, resulting in a time lag in detecting moving MOs. In other words, it is difficult to obtain an immediate response when detecting moving objects using waterfall data. In contrast, the distortion distribution waveform used in this embodiment can be obtained from the optical signal at the present time, making it possible to detect moving MOs in real time for each current moment, thus improving immediacy.

[0047] <Second Embodiment> Next, a second embodiment will be described. Figure 6 shows an example of the functional configuration of the mobile object tracking device 100 in the second embodiment. In this figure, the same reference numerals are used for parts that are the same as in Figure 2, and their descriptions are omitted as appropriate. Also, the waveform model 122 corresponds to the third embodiment described later, so its description is omitted here.

[0048] The tracking processing unit 101 of the mobile object tracking device 100 in Figure 6 further comprises a learning unit 114. In this embodiment, the learning unit 114 creates a mobile object model 121 by machine learning. The mobile object model 121 predicts predetermined attributes of the detected mobile object MO in response to input of mobile object parameters generated in accordance with the detected mobile object MO.

[0049] Figure 7 schematically shows the machine learning process that the learning unit 114 performs to create the mobile model 121. Referring to this figure, an example of the machine learning process corresponding to the creation of the mobile model 121 will be explained. The moving object detection unit 113 detects at a certain time t n In response to the detection of a moving object MO, a moving object D1 is created or updated. The learning unit 114 includes a learner 1141. The learning unit 114 inputs a learning dataset to the learner 1141, which consists of a mobile object D1 generated or updated by the mobile object detection unit 113 and reference data D2 indicating the attributes of the mobile object detected in accordance with the mobile object parameters. Reference data D2 may include the type of vehicle (manufacturer, model number, etc.), weight, shape, and road conditions (e.g., earthworks, bridge, tunnel, etc.) of the moving object, which are identified based on the image data obtained by imaging the detected moving object. In addition, with respect to weight, reference data D2 may include the weight detected by a vehicle weight measurement system, for example, called WIM (Weight In Motion). The learning unit 114 receives such a learning dataset as input to the learner 1141 at time t n This can be repeated each time.

[0050] The learner 1141 learns from the input training dataset and creates a mobile object model 121 that predicts the attributes of a mobile object in response to input mobile object parameters. The created mobile object model 121 is stored in the memory unit 102. The mobile object model 121 stored in the memory unit 102 is used by the mobile object detection unit 113 to improve the accuracy of mobile object detection.

[0051] The moving object model 121 is detected by the moving object detection unit 113 at time t n When object detection is being performed for each time t n The mobile object detection unit 113 inputs the mobile object parameters obtained from the detection of each mobile object and predicts the attributes of the detected mobile object. The mobile object detection unit 113 updates the mobile object object using the prediction results of the mobile object attributes by the mobile object model 121. The updated mobile object object is corrected to have accurate attributes for the mobile object object, including mobile object parameters obtained based on, for example, the strain distribution waveform.

[0052] Referring to the flowchart in Figure 8, an example of the processing procedure performed by the mobile object tracking device 100 of this embodiment in relation to tracking the mobile object MO will be described. In Figure 8, the processing in steps S200 to S208 is the same as in steps S100 to S108 in Figure 5.

[0053] Step S210: The moving object detection unit 113 inputs the moving object generated or updated in the preceding step S208 into the moving object model 121, and the moving object model 121 determines the time t as the current time. n The mobile object model 121 is then instructed to predict the attributes of the detected mobile object.

[0054] Step S212: The moving object detection unit 113 further updates the moving object generated or updated in step S208 based on the attributes of the moving object predicted in step S210.

[0055] The processing in steps S214 to S222 is the same as in steps S110 to S118 in Figure 5.

[0056] <Third Embodiment> Next, a third embodiment will be described. An example of the functional configuration of the mobile tracking device 100 corresponding to the third embodiment will be described again with reference to Figure 6. In this embodiment, the learning unit 114 creates a waveform model 122. The created waveform model 122 may be stored in the storage unit 102. The waveform model 122 predicts the presence or absence of a moving object based on the input distortion distribution waveform, and if it predicts the presence of a moving object, it also predicts the attributes of the moving object.

[0057] In this embodiment, the mobile tracking device 100 may or may not have a mobile model 121 corresponding to the second embodiment, but for the sake of simplicity, the case in which the mobile model 121 is not included will be given as an example.

[0058] Figure 9 schematically shows the machine learning process that the learning unit 114 performs to create the waveform model 122. Referring to this figure, an example of the machine learning process corresponding to the creation of the waveform model 122 will be explained. The strain distribution waveform acquisition unit 112 acquires a certain time t. n In this process, a distortion distribution waveform D3 is obtained from the optical signal input from the optical fiber 10.

[0059] In this case, the learning unit 114 includes a learner 1142. The learning unit 114 inputs a learning dataset to the learner 1142, which consists of the strain distribution waveform D3 acquired by the strain distribution waveform acquisition unit 112 and reference data D2 indicating the attributes of the moving object detected in accordance with the moving object parameters. The reference data D2 may include, similar to the second embodiment, the type of vehicle (manufacturer, model number, etc.), weight, shape of the vehicle, road conditions (e.g., earthworks, bridge, tunnel, etc.) of the moving object, which are identified based on the image data obtained by imaging the detected moving object. The learning unit 114 receives such a learning dataset as input to the learner 1142 at time t n This can be repeated each time.

[0060] The learner 1142 learns from the input training dataset and creates a waveform model 122 that predicts the presence or absence of a moving object and the attributes of the moving object in response to the input distortion distribution waveform. The created waveform model 122 is stored in the memory unit 102. The waveform model 122 stored in the memory unit 102 is used by the moving object detection unit 113 to detect moving objects.

[0061] In this embodiment, the moving object detection unit 113 utilizes the waveform model 122 when detecting a moving object, for example, in step S104 of Figure 5. That is, in step S104, the moving object detection unit 113 inputs the distortion distribution waveform acquired in step S102 to the waveform model 122, causing the waveform model 122 to predict whether or not a moving object is present, and if a moving object is present, it also predicts the attributes of the moving object. The attributes of the moving object predicted by the waveform model 122 may be included in the moving object parameters acquired in step S106. By performing such processing, the detection accuracy of whether or not a moving object is present in step S104 is improved, and the accuracy of the moving object object generated or updated including the moving object parameters acquired in step S106 is also improved.

[0062] <Note> [1] One aspect of this embodiment is a mobile tracking system comprising: a strain distribution waveform acquisition unit (112) that acquires a strain distribution waveform showing the spatial distribution of strain obtained from backscattered light generated in response to light incident on an optical fiber (10) laid in an environment in which a mobile body (MO) is moving, at predetermined time intervals; and a mobile body detection unit (113) that performs mobile body detection, including detection of the presence of a mobile body and detection of mobile body attributes (e.g., mobile body parameters) including the location of the mobile body, based on the strain distribution waveform acquired at predetermined time intervals by the strain distribution waveform acquisition unit.

[0063] [2] One aspect of this embodiment is the moving object tracking system described in (1), wherein the moving object detection unit may detect the speed of the moving object as the moving object attribute based on two or more positions of the moving object detected based on each of two or more strain distribution waveforms acquired by the strain distribution waveform acquisition unit at different time intervals.

[0064] [3] One aspect of this embodiment is a moving object tracking system according to (1) or (2), wherein the moving object detection unit may detect the acceleration of the moving object as a moving object attribute based on three or more positions of the moving object detected based on the positions of three or more strain distribution waveforms acquired by the strain distribution waveform acquisition unit at different time intervals.

[0065] [4] One aspect of this embodiment is a moving object tracking system according to any one of (1) to (3), wherein the moving object detection unit may detect the road surface conditions at the location where the moving object is present as the moving object attribute based on the strain distribution waveform acquired by the strain distribution waveform acquisition unit.

[0066] [5] One aspect of this embodiment is a moving object tracking system according to any one of (1) to (4), wherein the moving object detection unit may detect the type of moving object as the moving object attribute based on the strain distribution waveform acquired by the strain distribution waveform acquisition unit.

[0067] [6] One aspect of this embodiment is a moving object tracking system according to any one of (1) to (5), wherein the moving object detection unit may detect the weight of the moving object as the moving object attribute based on the strain distribution waveform acquired by the strain distribution waveform acquisition unit.

[0068] [7] One aspect of this embodiment is a moving object tracking system according to any one of (1) to (6), wherein the moving object detection unit may set a spatial range (e.g., detection range BD) in which the moving object will be detected in the next moving object detection based on the result of the current moving object detection.

[0069] [8] One aspect of this embodiment is a mobile tracking system according to any one of (1) to (7), wherein the mobile detection unit may detect a mobile attribute predicted by a learning model (e.g., mobile model 121) obtained by learning learning data that associates mobile attribute (e.g., mobile object) detected by the mobile detection unit with reference data indicating predetermined attributes of the mobile, in response to input of the mobile attribute detected by the mobile detection unit.

[0070] [9] One aspect of this embodiment is a moving object tracking system according to any one of (1) to (8), wherein the moving object detection unit may detect the presence or absence of a moving object and the attributes of the moving object as the detection result, based on a learning model obtained by learning learning data (waveform model 122) which associates a distortion distribution waveform acquired by the distortion distribution waveform acquisition unit with reference data indicating predetermined attributes of the moving object.

[0071]

[10] One aspect of this embodiment is a method for tracking a moving object in a moving object tracking system, comprising: a strain distribution waveform acquisition step in which a strain distribution waveform acquisition unit acquires a strain distribution waveform at predetermined intervals that shows the spatial distribution of strain obtained from backscattered light generated in response to light incident on an optical fiber laid in the environment in which the moving object is moving; and a moving object detection step in which a moving object detection unit performs a moving object detection that includes detecting the presence of a moving object and detecting the attributes of the moving object, including the location of the moving object, based on the strain distribution waveform acquired at predetermined intervals by the strain distribution waveform acquisition step.

[0072]

[11] One aspect of this embodiment is a program that causes a computer in a mobile object tracking system to function as a strain distribution waveform acquisition unit that acquires a strain distribution waveform showing the spatial distribution of strain obtained from backscattered light generated in response to light incident on an optical fiber laid in the environment in which a mobile object is moving, at predetermined intervals, and a mobile object detection unit that performs mobile object detection, including detection of the presence of a mobile object and detection of mobile object attributes including the location of the mobile object, based on the strain distribution waveform acquired at predetermined intervals by the strain distribution waveform acquisition unit.

[0073] Alternatively, the program for realizing the functions of the aforementioned mobile object tracking device 100 may be recorded on a computer-readable recording medium, and the program recorded on this recording medium may be loaded into a computer system and executed to perform the processing of the mobile object tracking device 100. Here, "loading the program recorded on the recording medium into a computer system and executing it" includes installing the program into the computer system. Here, "computer system" includes hardware such as the OS and peripheral devices. Furthermore, "computer system" may include multiple computer devices connected via a network including communication lines such as the Internet, WAN, LAN, and dedicated lines. Furthermore, "computer-readable recording medium" refers to portable media such as flexible disks, magneto-optical disks, ROMs, CD-ROMs, and storage devices such as hard disks built into the computer system. Thus, the recording medium storing the program may also be a non-transient recording medium such as a CD-ROM. Furthermore, the recording medium also includes internal or external recording media that are accessible from the distribution server for distributing the program. The program code stored on the distribution server's recording medium may be different from the program code in a format executable by the terminal device. In other words, the format in which the program is stored on the distribution server is irrelevant, as long as it can be downloaded from the distribution server and installed in an executable format on the terminal device. Furthermore, the program may be divided into multiple parts, each downloaded at a different time and then combined on the terminal device, and different distribution servers may distribute each of the divided programs. In addition, "computer-readable recording medium" includes volatile memory (RAM) within computer systems that act as servers or clients when a program is transmitted over a network, which retains the program for a certain period of time. Moreover, the program may only be used to implement some of the functions described above. Furthermore, the program may be a so-called differential file (differential program) that can implement the functions described above in combination with a program already recorded in the computer system. [Explanation of Symbols]

[0074] 10 Optical fiber, 100 Mobile object tracking device, 101 Tracking processing unit, 102 Memory unit, 111 Input unit, 112 Distortion distribution waveform acquisition unit, 113 Mobile object detection unit, 114 Learning unit, 121 Mobile object model, 122 Waveform model, 1141 Learner, 1142 Learner

Claims

1. A strain distribution waveform acquisition unit acquires a strain distribution waveform, which shows the spatial distribution of strain obtained from backscattered light generated in response to light incident on an optical fiber laid in an environment in which a moving object is moving, at predetermined time intervals. A mobile object detection unit performs mobile object detection, which includes detecting the presence of a mobile object and detecting the attributes of the mobile object, including the location of the mobile object, based on the strain distribution waveform acquired at predetermined time intervals by the strain distribution waveform acquisition unit. A mobile tracking system equipped with the following features.

2. The moving object detection unit detects the speed of the moving object as a moving object attribute based on two or more positions of the moving object detected based on each of two or more strain distribution waveforms acquired by the strain distribution waveform acquisition unit at different time intervals. The mobile tracking system according to claim 1.

3. The moving object detection unit detects the acceleration of the moving object as the moving object attribute based on the three or more positions of the moving object detected based on the positions of three or more strain distribution waveforms acquired by the strain distribution waveform acquisition unit at different time intervals. A mobile tracking system according to claim 1 or 2.

4. The moving object detection unit detects the road surface conditions at the location where the moving object is present as the moving object attribute, based on the strain distribution waveform acquired by the strain distribution waveform acquisition unit. A mobile tracking system according to claim 1 or 2.

5. The moving object detection unit detects the type of moving object as the moving object attribute based on the strain distribution waveform acquired by the strain distribution waveform acquisition unit. A mobile tracking system according to claim 1 or 2.

6. The moving object detection unit detects the weight of the moving object as an attribute of the moving object based on the strain distribution waveform acquired by the strain distribution waveform acquisition unit. A mobile tracking system according to claim 1 or 2.

7. Based on the results of the current mobile object detection, the mobile object detection unit sets the spatial range in which mobile objects will be detected in the next mobile object detection. A mobile tracking system according to claim 1 or 2.

8. The aforementioned mobile object detection unit uses a learning model, which has been trained by associating mobile object attributes detected by the mobile object detection unit with reference data indicating predetermined attributes of the mobile object, as the detection result, based on the mobile object attributes predicted by the learning model in response to the input of mobile object attributes detected by the mobile object detection unit. A mobile tracking system according to claim 1 or 2.

9. The moving object detection unit uses a learning model, which is obtained by learning learning data that associates the strain distribution waveform acquired by the strain distribution waveform acquisition unit with reference data indicating predetermined attributes of the moving object, as the detection result for the presence or absence of a moving object and the attributes of the moving object, predicted in response to the input of the strain distribution waveform acquired by the strain distribution waveform acquisition unit. A mobile tracking system according to claim 1 or 2.

10. A method for tracking a moving object in a moving object tracking system, The strain distribution waveform acquisition unit acquires a strain distribution waveform at predetermined intervals, which shows the spatial distribution of strain obtained from backscattered light generated in response to light incident on an optical fiber laid in an environment in which a moving object is moving. The moving object detection unit performs a moving object detection step which includes detecting the presence of a moving object based on a strain distribution waveform acquired at predetermined time intervals by the strain distribution waveform acquisition step, and detecting the attributes of the moving object, including the location where the moving object is located. A method for tracking moving objects, including the tracking of moving objects.

11. Computers in mobile tracking systems A strain distribution waveform acquisition unit acquires a strain distribution waveform, which shows the spatial distribution of strain obtained from backscattered light generated in response to light incident on an optical fiber laid in an environment in which a moving object is moving, at predetermined time intervals. A mobile object detection unit performs mobile object detection, which includes detecting the presence of a mobile object and detecting the attributes of the mobile object, including the location of the mobile object, based on the strain distribution waveform acquired at predetermined time intervals by the strain distribution waveform acquisition unit. A program designed to function as such.

Images