Signal processing system

Abstract

To provide a signal processing system capable of obtaining a stable detection result of speed using a plurality of laser sensors.SOLUTION: The signal processing system comprises: a point group data acquisition portion that acquires first point group data obtained by one laser sensor irradiating a moving body in a first section on a route with a laser beam, and second point group data obtained by the other laser sensor irradiating the moving body in a second section including a section continuous to the first section on the route with the laser beam; a synthesis portion that synthesizes the first and second point group data on common coordinates; a moving body position calculation portion that calculates representative coordinates of the moving body moving on the route at a plurality of time points on the basis of synthesized point group data at the plurality of time points; a speed calculation portion that calculates moving speed of the moving body in the first and second section on the basis of the representative coordinates of the moving body at the plurality of time points; and a smoothing processing portion that performs smoothing processing on time series data on the calculated moving speed of the moving body.SELECTED DRAWING: Figure 1

Description

Technical Field

[0001] The present invention relates to a signal processing system for processing an output signal of a laser sensor.

Background Art

[0002] In the main-line merging of an autonomous vehicle on an expressway, on a connecting road with poor visibility, it is difficult for the autonomous vehicle to grasp traffic information on the main-line side, so there is a risk of disturbing the traffic flow during merging. Therefore, roadside sensors such as 3D-LiDAR are installed to extract traffic information on the main line and perform merging support by distributing it to the connecting road at ITS spots and the like (for example, Non-Patent Document 1, Non-Patent Document 2). As a result, it is possible to expect safe and smooth merging of autonomous vehicles.

Prior Art Documents

Non-Patent Documents

[0003]

Non-Patent Document 1

Non-Patent Document 2

Summary of the Invention

Problems to be Solved by the Invention

[0004] The length of the detection section on the main road by roadside sensors should ideally be several hundred meters, considering the time it takes to distribute traffic information and the speed of vehicles. Therefore, in order to extend the detection range of 3D-LiDAR, multiple roadside sensors are installed in the same orientation along the main road, and the vehicle point clouds (point clouds indicating vehicles) obtained from each roadside sensor are combined.

[0005] However, when point clouds are combined in this way, a significant change occurs in the distribution of the vehicle point cloud at the overlapping section between the upstream and downstream detection sections. Therefore, when estimating vehicle speed from the changes in the center point of the vehicle point cloud, there was a problem that the estimated vehicle speed in this overlapping section could be abnormal and unstable.

[0006] This invention has been made in view of the above-mentioned problems, and aims to provide a signal processing system that can obtain stable detection results when detecting the speed of a moving object using multiple laser sensors. [Means for solving the problem]

[0007] The signal processing system according to the present invention is characterized by comprising: a point cloud data acquisition unit that acquires first point cloud data obtained by irradiating a moving object in a first section on a path along which a moving object is moving with one laser sensor and receiving the reflected light, and second point cloud data obtained by irradiating a moving object in a second section on the path including a section continuous with the first section with another laser sensor and receiving the reflected light; a synthesis unit that synthesizes the first point cloud data and the second point cloud data on a common coordinate system and generates synthesized point cloud data; a moving object position calculation unit that calculates representative coordinates of the moving object at each of the multiple time points based on the synthesized point cloud data at each of the multiple time points; a speed calculation unit that calculates the moving speed of the moving object in the first section and the second section based on the representative coordinates of the moving object at the multiple time points; and a smoothing processing unit that performs smoothing processing on the time-series data of the calculated moving speed of the moving object.

[0008] Furthermore, the signal processing method according to the present invention is characterized by including the steps of: acquiring first point cloud data obtained by irradiating a moving object in a first section on a path along which a moving object is traveling with one laser sensor and receiving the reflected light, and second point cloud data obtained by irradiating a moving object in a second section on the path including a section continuous with the first section with another laser sensor and receiving the reflected light; combining the first point cloud data and the second point cloud data on a common coordinate system to generate combined point cloud data; calculating representative coordinates of the moving object at each of the multiple time points based on the combined point cloud data at each of the multiple time points; calculating the moving speed of the moving object in the first section and the second section based on the representative coordinates of the moving object at the multiple time points; and performing a smoothing process on the time-series data of the calculated moving speed of the moving object.

[0009] Furthermore, the program according to the present invention is characterized by causing a computer to perform the following steps: acquire first point cloud data obtained by irradiating a moving object in a first section on a path along which a moving object is traveling with one laser sensor and receiving the reflected light, and second point cloud data obtained by irradiating a moving object in a second section on the path, which includes a section continuous with the first section, with laser light and receiving the reflected light; combine the first point cloud data and the second point cloud data on a common coordinate system to generate combined point cloud data; calculate representative coordinates of the moving object at each of the multiple time points based on the combined point cloud data at each of the multiple time points; calculate the moving speed of the moving object in the first section and the second section based on the representative position of the moving object at the multiple time points; and perform a smoothing process on the time-series data of the calculated moving speed of the moving object. [Effects of the Invention]

[0010] According to the present invention, it is possible to obtain stable detection results when detecting the speed of a moving object using multiple laser sensors. [Brief explanation of the drawing]

[0011] [Figure 1] This is a block diagram showing the configuration of the signal processing system according to the present invention. [Figure 2] This diagram schematically shows the arrangement of the laser sensors and their positional relationship to the moving object. [Figure 3] This is a flowchart showing the processing routine for speed detection. [Figure 4] This figure shows an example of vehicle speed error in a joint section. [Figure 5] This is a block diagram showing a modified configuration of a signal processing system. [Figure 6] This is a block diagram showing a modified configuration of a signal processing system. [Modes for carrying out the invention]

[0012] Hereinafter, embodiments of the present invention will be described with reference to the drawings. In the following descriptions and accompanying drawings, substantially identical or equivalent parts are denoted by the same reference numerals.

[0013] Figure 1 is a block diagram showing the configuration of a signal processing system 100 according to an embodiment of the present invention. The signal processing system 100 consists of a first laser sensor 11A, a second laser sensor 11B, and a signal processing device 12.

[0014] The first laser sensor 11A and the second laser sensor 11B are measuring devices that irradiate an object with laser light, receive the reflected light, and measure the distance to the object based on the received light. The first laser sensor 11A and the second laser sensor 11B are configured, for example, by a LiDAR (Laser Imaging Detection and Ranging) device.

[0015] The first laser sensor 11A and the second laser sensor 11B are arranged on the roadside of a route such as a highway, and irradiate a moving object such as a vehicle moving on the route with laser light. The first laser sensor 11A and the second laser sensor 11B receive the reflected light generated when the irradiated laser light is reflected by the moving object, and generate data of three-dimensional coordinates indicating the reflection position (hereinafter referred to as point cloud data).

[0016] The signal processing device 12 acquires point cloud data from the first laser sensor 11A and the second laser sensor 11B, and generates object movement information indicating the movement information of the moving object based on the acquired point cloud data.

[0017] In the following description, a case where the moving object that is the irradiation target of the laser light is a vehicle and the vehicle travels on a highway as a route will be described as an example.

[0018] FIG. 2 is a diagram showing the arrangement positions of the first laser sensor 11A and the second laser sensor 11B. The first laser sensor 11A and the second laser sensor 11B are arranged along the main line ML which is the main line of the highway. Here, a case where vehicles M1, M2, M3, and M4 are traveling on the main line ML is shown as an example. The signal processing device 12 is provided in the vicinity of the first laser sensor 11A and the second laser sensor 11B as a roadside processing device.

[0019] In addition, ITS (Intelligent Transport Systems) spots IS1 and IS2 are arranged on the roadside of the acceleration lane AL that merges into the main line ML. For example, the ITS spots IS1 and IS2 are arranged along the acceleration lane AL near the merging point with the main line ML. The ITS spots IS1 and IS2 provide driving support for the vehicles traveling on the acceleration lane AL. Here, a case where an autonomous vehicle SV is traveling near the merging point with the main line ML on the acceleration lane AL is shown as an example.

[0020] The first laser sensor 11A and the second laser sensor 11B continuously irradiate laser light and receive reflected light with respect to a predetermined range on the main line ML. For example, the first laser 11A irradiates laser light and receives reflected light with respect to the detection section DS1 on the main line ML. Also, the second laser 12A irradiates laser light and receives reflected light with respect to the detection section DS2 on the main line ML. The detection section DS1 and the detection section DS2 are continuous sections including an overlapping part. In other words, the detection section DS1 and the detection section DS2 partially overlap in the joint section SS, and the other parts are continuous sections. Therefore, with respect to a vehicle located on the joint section SS, laser light is irradiated by both the first laser sensor 11A and the second laser sensor 11B.

[0021] The first laser sensor 11A and the second laser sensor 11B receive the reflected light generated when the irradiated laser light is reflected by a vehicle traveling on the main line ML. Thereby, data of the three-dimensional coordinates of the reflection position where the laser light is reflected (hereinafter referred to as point cloud data) is obtained.

[0022] In the example shown in FIG. 2, the first laser sensor 11A irradiates laser light to the vehicles M2, M3 and M4 traveling on the detection section DS1 and receives the reflected light. The second laser sensor 11B irradiates laser light to the vehicle M1 traveling on the detection section DS2 and receives the reflected light.

[0023] The first laser sensor 11A and the second laser sensor 11B generate point cloud data indicating the position of each vehicle based on the received reflected light. The first laser sensor 11A and the second laser sensor 11B supply (transmit) the generated point cloud data to the signal processing device 12.

[0024] The signal processing device 12 generates object movement information MD, which is information indicating the position and speed of each vehicle traveling on the main line ML, based on the point cloud data acquired from the first laser sensor 11A and the second laser sensor 11B. The signal processing device 12 transmits the object movement information MD to ITS spots IS1 and IS2.

[0025] ITS spots IS1 and IS2 generate driving support information based on object movement information MD received from the signal processing device 12 and supply (transmit) it to a vehicle merging from the acceleration lane AL to the main lane ML (for example, an autonomous vehicle SV in Figure 2). For example, when an autonomous vehicle SV merges from the acceleration lane AL to the main lane ML, information representing the distance between vehicles traveling on the main lane ML in terms of time (hereinafter referred to as inter-vehicle time information) is required. ITS spots IS1 and IS2 generate inter-vehicle time information based on object movement information MD and supply driving support information including the generated inter-vehicle time information to a vehicle (such as an autonomous vehicle SV) traveling near the merging point of the acceleration lane AL with the main lane ML.

[0026] Referring again to Figure 1, the signal processing device 12 includes a first point cloud acquisition device 21A, a second point cloud acquisition device 21B, a point cloud synthesis unit 22, an object detection unit 23, a representative position calculation unit 24, a past frame detected object information storage unit 25, an object tracking unit 26, a velocity calculation unit 27, a velocity smoothing unit 28, and an object movement information generation unit 29.

[0027] The first point cloud acquisition device 21A is a point cloud acquisition unit that acquires first point cloud data P1 from the first laser sensor 11A. As described above, the section targeted by the first laser sensor 11A is the detection section DS1, so the first point cloud data P1 indicating a vehicle traveling in the detection section DS1 is acquired by the first point cloud acquisition device 21A.

[0028] The second point cloud acquisition device 21B is a point cloud acquisition unit that acquires second point cloud data P2 from the second laser sensor 11B. As described above, the section targeted by the second laser sensor 11B is the detection section DS2, so the second point cloud data P2 indicating a vehicle traveling in the detection section DS2 is acquired by the second point cloud acquisition device 21B.

[0029] The first point cloud acquisition device 21A and the second point cloud acquisition device 21B each acquire point cloud data at the same time at predetermined intervals. In the following explanation, the point cloud data acquired at a certain time (in other words, at point 1) will be referred to as point cloud data for one frame.

[0030] The point cloud merging unit 22 is a merging unit that merges the first point cloud data P1 and the second point cloud data P2 onto a common coordinate system. The point cloud merging unit 22 merges point cloud data by, for example, merging point clouds that are close in time from among the point clouds included in the first point cloud data P1 and the second point cloud data P2. The point cloud merging unit 22 outputs the merging result as merged point cloud data CP.

[0031] The object detection unit 23 detects objects based on the composite point cloud data CP. For example, the object detection unit 23 detects objects by obtaining three-dimensional point cloud information (hereinafter referred to as object point cloud) that represents the object to be detected using background subtraction technology or clustering technology. In this embodiment, since the objects to be detected are vehicles traveling in the detection sections DS1 and DS2 (including the joint section SS) of the main line ML, an object point cloud representing those vehicles is detected.

[0032] The representative position calculation unit 24 calculates the representative position (hereinafter referred to as the representative coordinate) of the object detected by the object detection unit 23 on the coordinate system. For example, the representative position calculation unit 24 calculates the representative coordinate of an object by calculating the centroid of the object point cloud acquired by the object detection unit 23. In this embodiment, since the object to be detected is a vehicle, the position of the centroid of the object point cloud representing the vehicle is calculated as the representative coordinate of the vehicle.

[0033] The past frame detected object information storage unit 25 stores the object point cloud and the representative coordinates of the object detected for each frame of point cloud data as past frame detected object information. The past frame detected object information storage unit 25 is composed of, for example, a semiconductor storage device such as flash memory or an HDD.

[0034] The object tracking unit 26 associates the object to be tracked, i.e., the object whose velocity is to be calculated, with each frame based on the representative coordinates of the object in the current frame calculated by the representative position calculation unit 24 and the past object point cloud and representative coordinates stored in the past frame detected object information storage unit 25. This identifies the transition of the object's representative position across multiple frames. In this embodiment, since the object to be tracked is a vehicle, tracking is performed by identifying the transition of the vehicle's representative coordinates based on the frame-by-frame association.

[0035] For example, let (X_0, Y_0, Z_0) be the representative coordinate position of an object OB calculated by the representative position calculation unit 24 based on the point cloud data of the current frame F_k. The object tracking unit 26 identifies the object with the closest distance to the representative coordinate position from the object point cloud in past frame F_(k-1) read from the past frame detected object information storage unit 25 as the same object and sets it as the object to be tracked. Note that the distance here is the three-dimensional Euclidean distance.

[0036] The velocity calculation unit 27 calculates the velocity based on the transition of the representative coordinates of the object being tracked by the object tracking unit 26. For example, the velocity calculation unit 27 calculates the velocity of the object as a velocity vector. The velocity calculation unit 27 performs velocity calculations for each of the x, y, and z components.

[0037] For example, if the representative position of an object at time t1 is C_t1(x,y,z) and the representative position of the object at time t2 is C_t2(x,y,z), then the velocity vector V_t2(x,y,z) at time t2 is expressed by the following equation (1).

[0038]

number

[0039] The speed calculation unit 27 sequentially calculates the speed of the object being tracked while it is moving through the detection section. For example, in this embodiment, the object being tracked is a vehicle traveling on the main line ML, and the speed of the vehicle is sequentially calculated by the speed calculation unit 27 over the period from when the vehicle enters the detection section DS1 until when it exits the detection section DS2. The speed calculation unit 27 sequentially supplies speed information indicating the calculated speed to the speed smoothing unit 28.

[0040] The speed smoothing unit 28 generates time-series data of speed based on the speed information supplied from the speed calculation unit 27. The speed smoothing unit 28 then performs a smoothing process on the time-series data of speed. In the following description, the process of smoothing the time-series data of speed will be referred to as "speed smoothing."

[0041] In this embodiment, the speed smoothing unit 28 performs speed smoothing by executing calculations using the median filter shown in the following equation (2) as the smoothing filter.

[0042]

number

[0043] Here, Vt is the estimated vehicle speed at time Vt, and the value of Vt with an overline is the smoothed vehicle speed. Note that Vt may be a velocity vector consisting of x, y, and z components, or it may be a scalar value. Smoothing using a median filter removes outliers (i.e., abnormal values) that appear in the vehicle speed of the joint section SS.

[0044] The object movement information generation unit 29 generates object movement information MD, which indicates the position and speed of an object, based on smoothed time-series velocity data. In this embodiment, object movement information MD is generated that shows the changes in the position and speed of the vehicle from the time it enters the detection section DS1 until it exits the detection section DS2. The object movement information MD is transmitted to ITS spots IS1 and IS2 by a transmission unit (not shown) provided in the signal processing device 12 and used to assist the driving of the autonomous vehicle SV.

[0045] Next, the processing operation of the speed detection process performed by the signal processing device 12 in this embodiment will be described. Figure 3 is a flowchart showing the processing routine for the speed detection process.

[0046] The first point cloud acquisition device 21A acquires the first point cloud data P1 from the first laser sensor 11A. The second point cloud acquisition device 21B acquires the second point cloud data P2 from the second laser sensor 11B (STEP 101).

[0047] The point cloud merging unit 22 merges the first point cloud data P1 and the second point cloud data P2 onto a common coordinate system to generate the merged point cloud data CP (STEP 102).

[0048] The object detection unit 23 detects an object point cloud representing the object to be detected (in this embodiment, a vehicle) based on the composite point cloud data CP (STEP 103).

[0049] The representative position calculation unit 24 calculates the representative position (representative coordinates) of the object on the coordinate system by calculating the centroid of the object's point cloud (STEP 104).

[0050] The object tracking unit 26 reads the object point cloud and representative position in past frames from the past frame detection object information storage unit 25. The object tracking unit 26 identifies the object (vehicle) to be tracked by matching the object point cloud and representative position in the current frame calculated in STEP 104 with the object point cloud and representative position in past frames (STEP 105).

[0051] The velocity calculation unit 27 calculates the velocity of the identified object to be tracked based on the transition of the representative position of the object (STEP 106).

[0052] The speed smoothing unit 28 generates time-series data of speed based on the speed calculated sequentially by the speed calculation unit 27 and performs speed smoothing processing (STEP 107).

[0053] The speed detection process by the signal processing device 12 of this embodiment is executed by the processing routine described above.

[0054] The signal processing device 12 in this embodiment calculates the vehicle speed based on point cloud data acquired by the first laser sensor 11A and the second laser sensor 11B, and performs speed smoothing processing using a median filter. This makes it possible to remove outliers, i.e., abnormal values ​​of vehicle speed, that occur in the joint section SS where the detection section DS1 and the detection section DS2 overlap.

[0055] Figure 4 shows the vehicle speed error compared to the actual vehicle speed (true value) with and without smoothing processing. The horizontal axis represents the detection interval; in relation to Figure 2, 90-99 corresponds to detection interval DS1, 96-110 to detection interval DS2, and 96-99 to joint interval SS. The vertical axis shows the deviation (error) from the true value, with the true value set to 0.

[0056] Here, the true value of the vehicle speed is taken from RTK (Real Time Kinematic)-GPS positioning using GNSS (Global Navigation Satellite System). Additionally, for comparison with the median filter, the vehicle speed error when smoothing is performed using the IIR (Infinite Impulse Response) filter shown in equation (3) below is also presented.

[0057]

number

[0058] Without smoothing, the error from the true value is large overall, as shown by the dashed line in Figure 4. In particular, the error in the joint section SS becomes quite large.

[0059] When smoothing is performed using an IIR filter, the overall error can be reduced compared to the case without smoothing, as shown by the dashed line in Figure 4. However, in the joint section SS, error fluctuations still occur even after smoothing.

[0060] In contrast, when smoothing is performed using a median filter, the error from the true value can be significantly reduced overall, including the joint section SS, as shown by the solid line in Figure 4. This is because, by using a median filter, which calculates the median value, outliers in vehicle speed (i.e., abnormal vehicle speed values) are excluded from the smoothing process.

[0061] As described above, the signal processing system 100 of this embodiment makes it possible to perform velocity detection with less error in the detection of the velocity of a moving object using multiple laser sensors. This is achieved by calculating the velocity based on the point cloud data obtained from each laser sensor and then smoothing the velocity. In particular, since outliers in so-called joint sections where the detection sections of each laser sensor overlap can be removed by smoothing, it is possible to obtain stable detection results.

[0062] It should be noted that the embodiments of the present invention are not limited to those described in the above examples. For example, in the above examples, the case in which velocity smoothing is performed using a median filter was described as an example. However, the smoothing calculation is not limited to that shown in the above examples, and smoothing may be performed using filters other than median filters. However, it is preferable that the filter used for smoothing is a filter that can remove outliers and abnormal values, such as a trimmed mean filter or a Hampel filter.

[0063] Furthermore, the configuration of the signal processing system is not limited to that shown in Figure 1. For example, it could be a signal processing system consisting of a slave unit that includes a laser sensor and performs processing up to object detection, and a master unit that performs processing from point cloud synthesis onward.

[0064] Figure 5 is a block diagram showing the configuration of a modified signal processing system 200 having the above configuration. The signal processing system 200 consists of a first slave unit 31A, a second slave unit 31B, and a master unit 32.

[0065] The first slave unit 31A includes a first laser sensor 11A, a first point cloud acquisition device 21A, and an object detection unit 33A. The second slave unit 31B includes a second laser sensor 11B, a second point cloud acquisition device 21B, and an object detection unit 33B. The first slave unit 31A performs object detection based on the point cloud data obtained by the first laser sensor 11A and transmits the detection result to the master unit 32. The second slave unit 31B performs object detection based on the point cloud data obtained by the second laser sensor 11B and transmits the detection result to the master unit 32. Based on the detection results transmitted from the first slave unit 31A and the second slave unit 31B, the master unit 32 performs processing such as point cloud data synthesis, calculation of representative positions, velocity calculation, and velocity smoothing. The signal processing system 200 with such a configuration can also obtain the same effects as the signal processing system 100 of the above embodiment.

[0066] Furthermore, in the above embodiment, the case in which data with smoothed velocity is used for the entire detection section was described as an example. However, velocity smoothing only needs to be performed for at least the joint section SS.

[0067] Figure 6 is a block diagram showing the configuration of a modified signal processing system 300 that switches whether or not to apply smoothing processing (i.e., whether or not to use smoothed data) depending on the position of the object. The signal processing system 300 consists of a first laser sensor 11A, a second laser sensor 11B, and a signal processing device 41.

[0068] The speed smoothing unit 42 provided in the signal processing device 41 differs from the speed smoothing unit 28 in Figure 1 in that it selectively switches between outputting the smoothed speed and outputting the unsmoothed speed, depending on the representative position on the coordinates calculated by the representative position calculation unit 24.

[0069] In other words, the speed smoothing unit 42 replaces the speed data during the travel through the joint section SS with the smoothed speed data from the time-series data of the vehicle's speed when it travels through the detection sections DS1 and DS2. On the other hand, for sections other than the joint section SS, the speed data before smoothing (i.e., the speed calculated by the speed calculation unit 27) is maintained as the speed data.

[0070] In other words, the speed smoothing unit 42 replaces only the data from the time-series speed data calculated by the speed calculation unit 27 that pertains to the time the vehicle is traveling through the joint section SS with the smoothed speed data.

[0071] With this configuration, speed smoothing is effectively performed only in the joint section (SS), making it possible to use speed information in its unsmoothed state for other sections.

[0072] Furthermore, in the above embodiment, the case in which the speed information smoothed by the speed smoothing unit 28 is supplied to the object movement information generation unit 29, and the object movement information generation unit 29 generates object movement information MD and transmits it to the ITS spots IS1 and IS2 was described as an example. However, alternatively, the speed information smoothed by the speed smoothing unit 28 may be displayed on a display unit (not shown) provided in the signal processing device 12.

[0073] Furthermore, in the above embodiment, the case in which the position of the centroid of the object point cloud is calculated as the representative position of the object on the coordinate system was explained as an example. However, the method of calculating the representative position is not limited to this, and for example, an object frame indicating the extent of the object's existence may be set on the coordinate system based on the object point cloud, and the position of the center point of that object frame may be used as the representative position.

[0074] Furthermore, the above embodiment described an example of speed detection for vehicles traveling on the main lane of an expressway. However, the object to be detected is not limited to vehicles such as automobiles, but may also be other moving objects such as ships or airplanes. Also, the detection section is not limited to roads such as expressways, but can be any various path on which a moving object can travel, such as sidewalks, railway tracks, or runways.

[0075] Furthermore, the series of processes described in the above embodiment can be carried out by computer processing according to a program stored in a recording medium such as ROM. In other words, each functional block of the signal processing device 12 in the above embodiment can be formed by a processing control unit provided in a computer executing a predetermined program. [Explanation of symbols]

[0076] 100 Signal Processing Systems 11A First Laser Sensor 11B Second Laser Sensor 12 Signal Processing Devices 21A 1st point cloud acquisition device 21B 2nd point cloud acquisition device 22 Point group synthesis section 23 Object detection unit 24 Representative position calculation section 25 Past Frame Detected Object Information Storage Unit 26 Object Tracking Unit 27 Speed ​​calculation section 28 Speed ​​smoothing section 29 Object movement information generation unit 31A First handset 31B Second Slave Unit 32 Signal Processing Devices 33A Object detection unit 33B Object detection unit 41 Signal Processing Device 42 Speed ​​smoothing section

Claims

1. A point cloud data acquisition unit acquires first point cloud data obtained by irradiating a moving object in a first section on the path along which the moving object is traveling with one laser sensor and receiving the reflected light, and second point cloud data obtained by irradiating a moving object in a second section on the path, which includes a section continuous with the first section, with another laser sensor and receiving the reflected light. A combining unit that combines the first point cloud data and the second point cloud data on a common coordinate system to generate combined point cloud data, A moving body position calculation unit calculates the representative coordinates of a moving body moving along the path at each of the multiple time points based on the composite point cloud data at each of the multiple time points, A speed calculation unit that calculates the moving speed of the moving body in the first section and the second section based on the representative coordinates of the moving body at the aforementioned multiple points in time, A smoothing processing unit performs smoothing on the time-series data of the calculated moving speed of the moving body, It has, The first section and the second section each have a joint section in which a portion of each overlaps with the other. The signal processing system is characterized in that, in the smoothing process, the smoothing processing unit replaces only the data from the time interval in which the moving object is located within the seam interval of the smoothed time series data obtained by smoothing the time series data with the data before processing.

2. The signal processing system according to claim 1, characterized in that the smoothing processing unit performs the smoothing process by executing an arithmetic process using a filter for removing outliers as a smoothing filter.

3. The signal processing system according to claim 2, characterized in that the smoothing processing unit performs calculation processing using a median filter as the smoothing filter.

4. The aforementioned moving body position calculation unit is: An object detection unit that detects an object point cloud representing the moving object based on the composite point cloud data, A representative coordinate calculation unit that calculates the representative coordinates of the moving object based on the object point cloud, An object information storage unit that stores the representative coordinates of the moving object at a past point in time, An object tracking unit identifies the transition of the representative coordinates of the moving object as the object to be tracked, based on the representative coordinates calculated by the representative coordinate calculation unit and the representative coordinates of the moving object at past points in time stored in the object information storage unit. It has, The signal processing system according to any one of 1 to 3, characterized in that the speed calculation unit calculates the moving speed of the moving body based on the transition of the identified representative coordinates.

5. The steps include acquiring first point cloud data obtained by irradiating a moving object in a first section on the path along which the moving object is traveling with one laser sensor and receiving the reflected light, and acquiring second point cloud data obtained by irradiating a moving object in a second section on the path, which includes a section continuous with the first section, with another laser sensor and receiving the reflected light, The steps include: combining the first point cloud data and the second point cloud data on a common coordinate system to generate combined point cloud data; A step of calculating the representative coordinates of a moving object moving along the path at each of the multiple time points based on the composite point cloud data at each of the multiple time points, A step of calculating the moving speed of the moving body in the first section and the second section based on the representative coordinates of the moving body at the aforementioned multiple time points, The steps include: performing a smoothing process on the time-series data of the calculated moving speed of the moving body; Includes, The first section and the second section each have a joint section in which a portion of each overlaps with the other. A signal processing method characterized in that, in the smoothing process, only the data of the time in which the moving object is located within the seam interval of the smoothed time series data obtained by smoothing the time series data is replaced with the data before processing.

6. On the computer, The steps include acquiring first point cloud data obtained by irradiating a moving object in a first section on the path along which the moving object is traveling with one laser sensor and receiving the reflected light, and acquiring second point cloud data obtained by irradiating a moving object in a second section on the path, which includes a section continuous with the first section, with another laser sensor and receiving the reflected light, The steps include: combining the first point cloud data and the second point cloud data on a common coordinate system to generate combined point cloud data; A step of calculating the representative coordinates of a moving object moving along the path at each of the multiple time points based on the composite point cloud data at each of the multiple time points, A step of calculating the moving speed of the moving body in the first section and the second section based on the representative position of the moving body at multiple points in time, The steps include: performing a smoothing process on the time-series data of the calculated moving speed of the moving body; A program that executes, The first section and the second section each have a joint section in which a portion of each overlaps with the other. A program characterized in that, in the smoothing process, only the data from the time interval in which the moving object is located within the seam interval of the smoothed time series data obtained by smoothing the time series data is replaced with the data before processing.

Images