Detection device, detection method, and program

Abstract

To provide a detection device, a detection method, and a program capable of accurately detecting a track obstacle even in a track having many irregularities by using a sensor device such as a radar having a wide detection range.SOLUTION: The above-mentioned problem can be solved by a detection device (1, 6) or the like comprising: a sensor device (10) for periodically detecting a detection point of a track (3, 4) around a mobile body (2); and a track obstacle determination unit (25, 25') for determining whether or not a track obstacle (52) exists based on a detection point (32) for each distance region obtained by dividing the track (3) into a plurality of distance regions.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] The present invention relates to a detection device, a detection method, and a program, and more particularly to a detection device, a detection method, and a program capable of detecting obstacles on a moving object's path. [Background technology]

[0002] One method for detecting obstacles on a mobile vehicle's path, such as concave obstacles, is to monitor the path around the vehicle using sensor devices such as radar or sonar attached to the vehicle. For example, Patent Document 1 describes a technology for an autonomous mobile device that detects concave areas in the road surface by detecting the distance to the road surface using sonar directed towards the road surface and capturing changes in that distance. [Prior art documents] [Patent Documents]

[0003] [Patent Document 1] Japanese Patent Publication No. 2013-235409 [Overview of the project] [Problems that the invention aims to solve]

[0004] For mobile vehicles such as agricultural and construction vehicles that travel on relatively unpaved roads such as field paths, farm paths, and embankment roads, it is essential to detect obstacles on the road, such as steps and ditches. However, the technology described in Patent Document 1 assumes that the vehicle is traveling on a well-maintained road where the distance between the road and the sonar is nearly constant, using highly directional sonar. Therefore, it is difficult to obtain stable detection results on roads with many irregularities, such as farm roads and construction sites.

[0005] The present invention aims to provide a detection device, detection method, and program that can accurately detect obstacles on uneven roads, even when using a sensor device such as radar, in which variations in distance between the sensor and the road have little effect on the detection result and which has a wide detection range. [Means for solving the problem]

[0006] The above problem can be solved by a detection device comprising a sensor device that periodically detects detection points on the track around a moving object, and a track obstacle determination unit that determines whether or not a track obstacle exists for each distance region into which the track is divided into multiple distance regions, based on the detection points.

[0007] Here, "mobile object" does not have to be a single vehicle, but includes a collection of moving objects such as a tractor and trailer. Also, "sensor device" refers to a device that can be attached to a mobile object to detect the position of the track, such as a radar device, LiDAR, or camera. "Track obstacle" refers to obstacles on and around the track, including terrain with steep drops in height (concave obstacles), such as steps, ditches, and cliffs.

[0008] The track obstacle detection unit includes a point cloud density detection unit, which, for each distance region, compares the point cloud density of detected points with a predetermined point cloud density threshold. It is desirable that the unit determines the presence of a track obstacle when there is a distance region with a point cloud density smaller than the point cloud density threshold, and the distance between that distance region and the moving object is smaller than a predetermined risk threshold.

[0009] Here, "point cloud density" is the number of detected points relative to distance, and includes not only the number of detected points per unit distance (linear density), but also an index shown by the number of detected points present in each distance region when the road is divided into multiple equally spaced distance regions.

[0010] The point cloud density threshold may be set based on the average value of the point cloud density of detected points within a predetermined interval. Alternatively, the point cloud density threshold may be set based on the average value of the number of point clouds of detected points within a predetermined period.

[0011] Furthermore, the track obstacle detection unit is equipped with a track data monitoring unit that detects changes in the track environment based on detection information from the sensor device, and it is desirable that the point cloud density determination unit updates the point cloud density threshold when a change in the environment is detected.

[0012] Furthermore, it is desirable that the track obstacle detection unit includes a height determination unit, which compares the average height of the detection point with a predetermined height threshold for each distance region, and determines that a track obstacle exists when there is a distance region where the average height is smaller than the height threshold.

[0013] Furthermore, it is desirable that the track obstacle detection unit issue a warning when a track obstacle is present.

[0014] Furthermore, the above problem can also be solved by a detection method that includes the steps of periodically detecting detection points on the track around a moving object using a sensor device, and determining whether or not a track obstacle exists for each distance region into which the track is divided into multiple distance regions, based on the detection points.

[0015] Furthermore, the above problems can also be solved by a computer program that provides the following functions: a function to periodically detect detection points on the track around a moving object using a sensor device, and a function to determine whether or not obstacles exist on the track based on the detection points for each distance region into which the track is divided into multiple distance regions. [Effects of the Invention]

[0016] According to the present invention, it is possible to provide a detection device, detection method, and program that can accurately detect obstacles on uneven roads using a sensor device such as a radar with a wide detection range. [Brief explanation of the drawing]

[0017] [Figure 1] This is a schematic diagram of the detection device according to the present invention. [Figure 2] This is an explanatory diagram of detection using radar equipment. [Figure 3] This is a flowchart of the detection method and program according to the present invention. [Figure 4] This is a flowchart of detection using radar equipment. [Figure 5] This is an explanatory diagram of height correction based on IMU information. [Figure 6] This is a flowchart for determining obstacles on the track (step 1). [Figure 7] This is a flowchart for determining point cloud density. [Figure 8] This is an explanatory diagram for point cloud density determination. [Figure 9] This is a flowchart for determining the level of risk. [Figure 10] This is an explanatory diagram for risk assessment. [Figure 11] This is a flowchart illustrating a modified example of point cloud density determination. [Figure 12] This is a flowchart for determining obstacles on the track (2). [Figure 13] This is an explanatory diagram for the track obstacle detection method 2. [Figure 14] This is a schematic diagram of a modified example of the detection device according to the present invention. [Figure 15] This is a flowchart for determining point cloud density in a modified detection device. [Figure 16] This is a flowchart for determining environmental changes. [Figure 17] This is an explanatory diagram for determining environmental changes. [Modes for carrying out the invention]

[0018] Figure 1 shows a schematic configuration diagram of a detection device 1, which is an embodiment of the present invention. The detection device 1 includes a radar device 10, which is a type of sensor device that periodically detects detection points on the track around a moving object 2, and a track obstacle determination unit 25 connected to the radar device 10, which determines whether or not track obstacles such as concave obstacles exist based on the detection points.

[0019] The detection device 1 is connected to a higher-level device 29 that performs processing such as automatic steering to avoid obstacles on the road based on the detection results of road obstacles. Although Figure 1 shows only one radar device 10, the detection device 1 may be equipped with multiple radar devices. Furthermore, the connection between devices may be configured by wire or by wireless network.

[0020] The radar device 10 periodically transmits and receives signals in a predetermined frequency band (for example, every 50 ms) and is mounted around a moving object 2 such as agricultural machinery, construction machinery, or general automobiles (for example, in the front direction (front or front corner), in the rear direction (rear or rear corner), or to the side of the moving object, or a combination thereof) to detect the condition of the road surrounding the moving object. Specifically, as shown in Figure 2, it can detect the position of the road as seen from the moving object 2. Figure 2(a) is a top view of the moving object 2 on which the radar device 10 is mounted, and Figure 2(b) is a side view of the moving object 2. The radar device 10 detects multiple detection points 32, and for each detection point 32, three-dimensional positional information including height is detected. Based on this group of detection points 32, the detection range 31 of the radar device 10 can be determined. The detected information is used to identify obstacles on the road, such as ditches and cliffs, around the mobile vehicle 2, and to prevent the vehicle's wheels from getting stuck in ditches or the vehicle from falling off cliffs. It can also be used to determine the width of the road by detecting ditches on both sides of the road.

[0021] The radar system 10 includes a signal generation unit 14 that generates a periodic transmission signal pattern, a transmission control unit 13 connected to the signal generation unit 14 and controlling the operation of the signal generation unit 14, an oscillator unit 15 connected to the signal generation unit 14 that oscillates a transmission signal based on the signal pattern from the signal generation unit 14, a transmitting antenna 11 connected to the oscillator unit 15 that radiates the transmission wave oscillated by the oscillator unit 15 in the direction toward the track, a receiving antenna 12 that receives the reflected wave reflected by the track 40, an ADC 17 connected to the receiving antenna 12 that converts the signal received by the receiving antenna 12 into a digital signal, a distance calculation unit 18 that calculates the relative distance of the target 40 to the radar system 10 based on the digital signal, a velocity calculation unit 19 that calculates the velocity of the moving object 2 based on the digital signal, a threshold setting unit 20 that sets the detection threshold used by the detection unit 21, a detection unit 21 that detects the peak in the distance direction, an angle calculation unit 22 that determines the angle of arrival (horizontal angle, elevation angle) at the distance detected by the detection unit 21, and an IMU (Internal Measurement Unit) that measures the deviation in the detection direction. The system includes a correction unit 23 that corrects using Unit information, and a filter unit 24 that performs filtering to improve the accuracy of track detection by excluding point cloud information other than the track from the correction results of the correction unit 23. The oscillator unit 15 and the ADC 17 can be directly connected by a switch 16. The radar device 10 of the detection device 1 employs the FCM (Fast Chirp Modulation) method, but is not limited to the FCM method and may also use a pulse method.

[0022] The track obstacle detection unit 25 determines whether or not a track obstacle exists for each distance region into which the track has been divided into multiple distance regions, based on detected points. The track obstacle detection unit 25 comprises two types of detection units (point cloud density detection unit 26 and height detection unit 27) that use different methods for determining track obstacles. The track obstacle detection unit 25 determines whether or not a track obstacle exists using one or both of the point cloud density detection unit 26 and the height detection unit 27. The track obstacle detection unit 25 performs the functions of each unit by describing the functions of each unit in a program and executing it by a computer, but each unit may be configured by individual hardware or software. The program is stored in the computer's storage device.

[0023] Next, the operation of the detection device 1, that is, the detection method and program which are embodiments of the present invention, will be described based on the flowchart in Figure 3.

[0024] First, the radar device 10 periodically detects detection points on the track around the moving object 2 and provides the detection information to the track obstacle determination unit 25 (step 101). The detection information includes the speed of the moving object 2, three-dimensional position information including the height of the track for each detection point, and the received intensity of the reflected wave. The detection information is stored in the memory of the detection device 1 in chronological order for each frame corresponding to each period.

[0025] Next, the track obstacle determination unit 25 divides the track that the moving body 2 is scheduled to travel along into multiple distance regions, and for each divided distance region, it determines whether or not there are track obstacles on or around the track that would hinder travel, based on the detection points by the radar device 10 (step 102). There are two types of track obstacle determination processes: track obstacle determination 1 by the point cloud density determination unit 26 and track obstacle determination 2 by the height determination unit 27. However, it is possible to determine whether or not there are track obstacles by performing only one of the track obstacle determination processes, or to make a determination based on the determination results of both track obstacle determinations. When performing both track obstacle determinations, they may be performed in parallel or sequentially. When performed sequentially, the order of the determination processes does not matter.

[0026] If the track obstacle detection unit 25 determines that no track obstacles exist (step 103), the process is terminated. Conversely, if it determines that track obstacles exist, the detection device 1 issues a warning (step 104) and terminates the process. The warning may be transmitted to the host device 29 that an obstacle has been detected, or it may be issued directly to the user of the mobile body 2 via the display device of the mobile body 2.

[0027] Next, the radar detection (step 101) and road obstacle detection (step 102) processes will be explained in detail.

[0028] Figure 4 is a flowchart that specifically illustrates the operation of the radar detection (step 101) described above. First, the transmission control unit 13, signal generation unit 14, and oscillation unit 15 generate a transmission signal of a predetermined frequency and perform a transmission process in which the transmission wave is radiated from the transmission antenna 11 toward the track in front of the moving object 2 (step 201). Next, the reflected wave, which is reflected by one or more targets (tracks) 40 of the transmission wave, is received as a received signal by the receiving antenna 12 and converted into a digital signal by the ADC 17 (step 202).

[0029] Next, the distance calculation unit 18 performs a distance calculation to determine the relative distance between the radar device 10 and the track 40, which is one of the targets, based on the digital signal (step 203). The velocity calculation unit 19 also performs a velocity calculation based on the digital signal (step 204). Specifically, the velocity calculation unit 19 calculates relative velocity information with one or more objects based on the received signal. The velocity calculation allows for the detection of the object's velocity. Subsequently, the threshold setting unit 20 sets a detection threshold for detecting the peak to be used by the detection unit 21 (step 205). Specifically, the threshold is adjusted based on the mounting position, height, angle, etc., of the receiving antenna 12. Then, the detection unit 21 detects the peak in the distance direction using a predetermined method such as CFAR (Constant False Alarm Rate) (step 206). Next, the angle calculation unit 22 performs angle calculation processing to determine the arrival angle (horizontal angle, elevation angle) of the reflected wave at the distance detected by the detection unit 21, and calculates the direction in which the target (such as a road) 40 exists in a virtual space with the position of the radar device 10 as the origin and the radiation direction as the reference axis (step 207).

[0030] Furthermore, the correction unit 23 corrects for the deviation in the detection position that occurs when the radar device is tilted due to the effects of unevenness in the road surface, etc., by utilizing IMU information (step 208). As shown in Figure 5A, when the moving object 2 is traveling on a flat road surface, the positional relationship between the radar device 10 and the road surface 3 does not change, so the height can be detected without fluctuation. As is clear from the graph in the figure, the detection point 32 is located within a predetermined range 60. In contrast, as shown in Figure 5B, when the moving object 2 travels on a road surface with bumps, the position of the target (road surface) 40 obtained by the radar device 10 is determined by its relative position from the radar, resulting in a deviation. For example, when the radar device 10 is tilted upward, the road surface is detected lower than its original position, and a detection point 33 is detected that is significantly outside the range 60 where it should be located. To resolve this discrepancy, the change in the attitude of the mobile object 2 (amount of movement in three dimensions) is recognized from the IMU information of the mobile object 2, and the coordinate information as seen from the radar device 10 is converted to global coordinates (coordinates with the starting position as the origin).

[0031] Subsequently, the filter unit 24 performs filtering (step 209) to improve the accuracy of track detection by removing point cloud information other than the track from the correction results by the correction unit 23, and then terminates the process. In the filtering process by the filter unit 24, if the standard deviation in the height direction of the track is large compared to other frames or distances, the data of that frame may be excluded.

[0032] Through the above processing, the radar device 10 can detect detection information including the three-dimensional position information of each detection point, the received intensity of the reflected wave, and the velocity of the object. The detection information of the radar device 10 is stored in the memory of the detection device 1 in chronological order for each frame corresponding to each period.

[0033] Next, based on the flowchart shown in Figure 6, a detailed explanation of the process for Track Obstacle Determination 1, one of the two track obstacle determination methods, will be provided. Track Obstacle Determination 1 divides the track into multiple distance regions, and for each distance region, it compares the point cloud density of detected points with a predetermined point cloud density threshold. If there is a distance region with a point cloud density smaller than the point cloud density threshold, and the distance between the distance region with a point cloud density smaller than the point cloud density threshold and the moving object is smaller than a predetermined danger threshold, it is determined that a track obstacle exists. Track Obstacle Determination 1 is performed by the computer (processor) constituting the point cloud density determination unit 26 executing a program stored in memory.

[0034] First, a point cloud density determination is performed to determine whether or not there is a distance region with a low point cloud density of the detected points, based on the location information of the detected points stored in the memory of the detection device 1 (step 301). If there is no distance region with a low point cloud density (step 302), it is determined that there are no obstacles on the track (step 306). On the other hand, if there is a distance region with a low point cloud density, a hazard determination is performed to determine the hazard level of that distance region (step 303). In other words, possible causes of low point cloud density include the point cloud not being detected because the distance region is beyond the detection range 31 of the radar device 10, and the presence of obstacles on the track such as steps, ditches, or cliffs. The hazard determination distinguishes between the two and detects only the presence of obstacles on the track. In other words, if the distance region with a low point cloud density is a distance region beyond the detection range 31, it is determined that the hazard level at this time is low (step 304), and it is determined that there are no obstacles on the track (step 306), and the process is terminated. In contrast, if a distance region with a low point cloud density is within the detection range 31, it is determined that the risk level is high, and it is determined that a road obstacle exists (step 305), and the process is terminated.

[0035] Next, we will explain in detail the point cloud density determination (step 301) and hazard determination (step 303) processes within the track obstacle detection 1.

[0036] Figure 7 is a flowchart for point cloud density determination. Point cloud density determination involves dividing the road into multiple distance regions and comparing the point cloud density of detected points in each distance region with a predetermined point cloud density threshold. Distance regions with a point cloud density smaller than the point cloud density threshold are determined to be distance regions with low point cloud density.

[0037] First, the point cloud information of the detected points stored in the memory of the detection device 1 is retained and stored (step 401). Next, the road is divided into multiple distance regions (step 402), and the retained point cloud is distributed to each divided distance region based on its position information, and the point cloud density den for each distance region d is calculated. d The following is calculated (step 403). At this time, if the width of each distance region is made equal, the number of detection points allocated to each distance region can be treated as the point cloud density, thereby reducing the amount of computation required by the computer. In the detection device 1, the distance region is divided into predetermined widths (e.g., 0.1m) and the number of points per unit length is determined for each divided distance region to determine the point cloud density for each distance region. However, the division method and the size of the division width are not limited to this and may be changed according to the speed of the moving object 2. For example, the width may be made larger when the speed is fast and smaller when the speed is slow. The reference point that serves as the origin of the distance region may be the starting position of the moving object 2 or the current position of the moving object 2.

[0038] Figure 8(a) is a plan view from above the track showing the moving object 2 traveling towards a step, which is one of the track obstacles 52, and Figure 8(b) shows the point cloud density den d This is the calculation result. In Figure 8(b), the horizontal axis is the distance from the starting position of the moving object 2, and the track obstacle 52 is located at a position 3m in the direction of travel. The point cloud density of each region is shown by the intensity of the color, with darker colors indicating higher point cloud density. As mentioned above, the width of the distance region is processed as 0.1m, but to avoid complexity, the intensity of the color is displayed every 0.5m in Figure 8(b). As is clear from the figure, in the upper track 51 in front of the track obstacle 52, the point cloud density den d The point cloud density is high, but near obstacle 52 on the track, the point cloud density is low. d It can be seen that the level has decreased.

[0039] Point group density den d Following the calculation of the point group density, a point group density threshold th den is set to determine the high or low of the point group density (step 404). The point group density threshold th den can be set as appropriate. In the detection device 1, it is set based on the average value of the point group density of the detection points detected in a predetermined section. That is, the average value of the point group density den ref of a predetermined section of the traveled road surface and the number of superimposed frames accum frame are used to obtain it by the following formula. th den = den ref / accum frame ···(1)

[0040] Here, the number of superimposed frames is, for example, the number of frames until the change between frames (time) of the number of point groups in the predetermined section used for threshold creation becomes a predetermined value or less. The change between frames uses, for example, the change rate of the point group density between the past and current frames. Also, not limited to the change rate, the change amount or the like may be used. Note that the past frame is not limited to one frame, and the point group density of a plurality of frames may be used.

[0041] Next, for each distance region, the point group density den d is compared with the point group density threshold th den (step 405). If there is a distance region d where the point group density den d is smaller than the point group density threshold th den , it is determined that there is a distance region where the point group density of the detection points is low (step 406), and the process ends. On the other hand, if the point group density den d of all distance regions is greater than or equal to the point group density threshold th den , it is determined that there is no distance region where the point group density of the detection points is low (step 407), and the process ends.

[0042] Next, we will explain in detail the process of determining the degree of danger (step 303) within the track obstacle detection 1. Figure 9 is a flowchart of the degree of danger determination. The degree of danger determination determines the degree of danger for distance regions where the point cloud density is determined to be low in the point cloud density determination. Possible reasons for low point cloud density include the fact that the point cloud was not detected because the distance region is beyond the detection range 31 of the radar device 10, and the presence of track obstacles such as steps, ditches, and cliffs. The degree of danger determination distinguishes between the two and detects only the presence of track obstacles. That is, as shown in Figure 10, if the point cloud density is determined to be low in distance region 42 within the detection range 31 of the radar device 10, it is highly likely that the low point cloud density was determined to be due to the presence of track obstacles, and therefore the degree of danger is determined to be high. On the other hand, if the point cloud density is determined to be low in distance region 43 outside the detection range 31 of the radar device 10, it is not possible to determine which of the two causes is responsible. However, at present, the mobile object 2 to which the radar device 10 is attached is still sufficiently far from the distance region 43 where the point cloud density has been determined to be low. If the distance region 43 comes within the detection range 31 of the radar device 10 as the mobile object 2 moves in the future, the cause can be determined, so the level of danger is judged to be low at this time.

[0043] First, the distance r between the distance region d, which was determined to have a low point cloud density, and the moving object 2. d Calculate (step 501). Next, determine the risk threshold th for determining the degree of risk. dist Set the risk threshold (step 502). dist This is set, for example, based on the detection range 31 of the radar device 10. This setting allows detection only of distance regions where road obstacles exist among distance regions where the point cloud density is determined to be low. The detection range 31 of the radar device 10 may be set based on the specifications of the radar device 10, or it may be defined as the point where the point cloud density has decreased by a predetermined percentage relative to the maximum point cloud density in a distance histogram with the position of the radar device as the origin. However, it is necessary that the distance is such that the moving object 2 can stop after making a decision and can also detect the obstacle.

[0044] Next, distance r d and the risk threshold th dist Compare the distance r (step 503). d The risk threshold dist If it is smaller than this, the risk is determined to be high (step 504), and the process is terminated. d The risk threshold dist In the above case, since there is no immediate danger to the mobile body 2, the level of danger at this time is determined to be low (step 505), and the process is terminated.

[0045] Based on the above, it is possible to determine whether or not obstacles exist on the track based on point cloud density. This makes it possible to accurately detect obstacles on the track even on tracks with many recesses or protrusions, using sensor devices such as radar with a wide detection range.

[0046] Figure 11 shows a flowchart of a modified version of the point cloud density determination process. In this modified version, the point cloud density threshold is set based on the average number of point clouds of detected points detected over a predetermined period, which differs from the point cloud density determination process shown in Figure 7.

[0047] First, the point cloud information of the detected points stored in the memory of the detection device 1 is retained (step 421). At this time, not only the point cloud information of the detected points in the current frame, but also the point cloud information of the detected points over a predetermined period in the past, for example, the point cloud information of the most recent n frames (n is a natural number) of detected points is retained. Furthermore, the track is divided into multiple distance regions (step 422), and the point cloud of the current frame from the retained point cloud is distributed to the divided distance region based on the position information of each region, resulting in a point cloud count cnt for each distance region d. d Calculate (step 423). At this time, divide the width of each distance region into equal intervals. Then, the number of points in the point cloud of detected points assigned to each distance region is cnt. d This allows us to show the point cloud density for each distance region d of the current frame.

[0048] Next, a point cloud density threshold th is used to determine whether the point cloud density is high or low. cntSet the point cloud density threshold (step 424). cnt This value is obtained by distributing the points held in step 421 for the most recent n frames based on their respective position information into divided distance regions, and normalizing the average number of points to the number of points per unit frame. In other words, the number of points held for the past n frames is cnt p Therefore, the point cloud density threshold th cnt It can be calculated as follows: th cnt = cnt p / n ···(2)

[0049] Next, for each distance region, the number of points in the point cloud is cnt. d and point cloud density threshold th cnt Compare with (Step 425). Number of points in the point cloud: cnt d The point cloud density threshold is th cnt If a distance region d smaller than exists, it is determined that a distance region with a low point cloud density of detected points exists (step 426), and the process is terminated. In contrast, the total number of point clouds in all distance regions is cnt. d If the point cloud density is greater than or equal to the point cloud density threshold thcnt, it is determined that there are no distance regions with low point cloud density for the detected points (step 427), and the process is terminated.

[0050] Next, referring to Figures 12 and 13, we will provide a detailed explanation of the process for Track Obstacle Detection 2, one of the two track obstacle detection methods. Figure 12 is a flowchart of Track Obstacle Detection 2. Figures 13(a) and (b) are examples of Track Obstacle Detection 2, where the horizontal axis represents the distance in the direction of movement of the moving body 2, and the vertical axis represents the height relative to the height of the upper track 51. Each figure also shows the positions of a type of track obstacle, namely a step 52, the upper track 51 above the step 52, and the lower track 53 below the step 52. Specifically, the step 52 is located 3.0 m away from the moving body 2 in the direction of movement of the moving body 2, the upper track 51 is located before the step 52, and the lower track 53, located 0.25 m lower than the upper track 51, is located on the track after the step 52.

[0051] Track obstacle detection 2 divides the track into multiple distance regions and compares the average height of the detection point with a predetermined height threshold for each distance region. If there is a distance region with an average height smaller than the height threshold, it is determined that a track obstacle exists. Track obstacle detection 2 is performed by the computer (processor) constituting the height determination unit 27 executing a program stored in memory.

[0052] First, the point cloud information of the detected points is stored in the memory of the detection device 1 (step 601). Then, the track is divided into multiple distance regions (step 602), and the stored point cloud is distributed to the divided distance region based on the position information of each region, and the average height h for each distance region d is calculated. ave The value is calculated (step 603). In detection device 1, the width of each distance region was set to 0.1m. Figure 13(a) shows the distribution of detection points 32 and the average height 70 for each distance region d, and Figure 13(b) shows the average height 70 of the detection points for each distance interval.

[0053] Next, a height threshold th is used to determine the presence or absence of obstacles on the track. h Set the height threshold (step 604). h Although this can be set as appropriate, the detection device 1 uses the average value h of the height of the track in a predetermined section. ref and a predetermined offset h offset Based on this, the following formula was set. th h = h ref - h offset ...(3)

[0054] Each figure in Figure 13 shows the height threshold th h This is shown by line 76. Note that the average height h of the track in the designated section. ref The size of the predetermined section used when calculating the value may be changed according to the speed of the moving object 2. The starting point of the predetermined section may be based on the moving object 2, or on the starting position. Furthermore, a predetermined offset h may be used. offset This value may be determined based on the standard deviation in the height direction of the detected point cloud, or it may be set to an arbitrary value.

[0055] Next, for each distance region, the average height h ave 70 and height threshold th h Compare 76 (step 605). Average height h ave 70 is the height threshold th h If there is a distance region smaller than 76, it is determined that there is a track obstacle (step 606), and the process is terminated. In contrast, the average height h of all distance regions ave 70 is the height threshold th h If the value is 76 or higher, it is determined that there are no obstacles on the track (step 607), and the process is terminated. In the example shown in Figure 13, the average height h of the distance region of the upper track 51 is... ave 70 is the height threshold th h The average height h of the distance region of the lower track 53 is 76 or higher. ave 70 is the height threshold th h Since it is smaller than 76, it is determined that there is an obstacle on the track.

[0056] With the detection method described above, it becomes possible to accurately detect obstacles on a track, even on a track with many recesses or protrusions, using sensor devices such as radar with a wide detection range.

[0057] Next, with reference to Figures 14-17, modified examples of the detection device, detection method, and program according to the present invention will be described. Figure 14 is a schematic diagram of the modified detection device 6, and Figure 15 is a flowchart showing the operation of the detection device 6. The detection device 6 differs from the detection device 1 in that the track obstacle determination unit 25' includes a track data monitoring unit 28 in addition to the point cloud density determination unit 26' and the height determination unit 27. For this reason, components of the detection device 6 that have the same functions as the detection device 1 are given the same reference numerals as those of the detection device 1.

[0058] The track data monitoring unit 28 detects changes in the track environment based on detection information detected by the radar device 10 and provides the results to the point cloud density determination unit 26'. The point cloud density determination unit 26' is the same as the point cloud density determination unit 26 of the detection device 1 in that it compares the point cloud density of detected points with a predetermined point cloud density threshold for each distance region obtained by dividing the track around the moving object 2 into multiple distance regions, and determines that a track obstacle exists when there is a distance region with a point cloud density smaller than the point cloud density threshold, and the distance between the distance region with a point cloud density smaller than the point cloud density threshold and the moving object 2 is smaller than a predetermined danger threshold. However, it differs from the point cloud density determination unit 26' in that it updates the point cloud density threshold when an environmental change is detected. The point cloud density determination unit 26', height determination unit 27, and track data monitoring unit 28 of the detection device 6 perform their functions by describing their functions in a program and executing it on a computer, but they may also be configured by individual hardware and software. The program is stored in the computer's storage device.

[0059] Next, the operation of the detection device 6, that is, the detection method and program which are embodiments of the present invention, will be described. The operation of the detection device 6 differs from that of the detection device 1 shown in Figure 6, specifically in the operation of point cloud density determination (step 301) in the track obstacle determination 1 process. For this reason, the point cloud density determination process performed by the point cloud density determination unit 26' of the detection device 6 will be described below.

[0060] Figure 15 is a flowchart of the point cloud density determination process of the detection device 6. Steps 401 to 407 shown in Figure 15 are the same as steps 401 to 407 shown in Figure 7 above. First, the point cloud information of the detected points stored in the memory of the detection device 6 is retained and stored (step 401). Next, the road is divided into multiple distance regions (step 402), and the retained point clouds are distributed to the divided distance regions based on their respective position information, determining the point cloud density den for each distance region d. d Calculate (Step 403).

[0061] Next, a point cloud density threshold th is used to determine whether the point cloud density is high or low. den Set the point cloud density threshold th. den Determine whether the settings are initial settings (step 408). If it is initial settings, the existing point cloud density threshold th den Since there is no point cloud density threshold th den Set the specific point cloud density threshold (step 404). den The setting method is the same as the setting method for the detection device 1 described in step 404 of Figure 7. In contrast, if it is not the initial setting, the track data monitoring unit 28 performs an environmental change determination process (step 409). In this process, if an environmental change occurs, such as a change in the condition of the track (material, pavement condition, etc.), the detection information of the point cloud detected using a predetermined index is determined to be different from that of normal conditions, i.e., whether or not there is an environmental change. If it is determined that there is an environmental change (step 410), the point cloud density threshold th den Reset (update) (step 404). On the other hand, if it is determined that there is no change in the environment, the point cloud density threshold th den Without making any changes to the settings, the existing point cloud density threshold th den Use it.

[0062] Next, for each distance region, the point cloud density den d and point cloud density threshold th den Compare (Step 405). Point cloud density den d The point cloud density threshold is th den If a distance region smaller than den exists, it is determined that a distance region with a low point cloud density of detected points exists (step 406), and the process is terminated. In contrast, the point cloud density of all distance regions den d The point cloud density threshold is th den If the above conditions are met, it is determined that there are no distance regions with a low point cloud density of detected points (step 407), and the process is terminated.

[0063] Next, referring to the flowchart in Figure 16, we will explain the process of determining environmental changes in detail. First, point cloud information of detected points over a predetermined period in the past, for example, the environmental change index env for the most recent n frames (n is a natural number) fac Keep and store (Step 701). Environmental change indicator env fac This is detection information from the radar device 10 that depends on changes in the track environment, and for example, the received signal strength is used. That is, when traveling on a track with the same conditions, there is little change in the received signal strength. On the other hand, if the track being traveled on switches to a track that is more reflective, the received signal strength will increase, and if it switches to a track that is less reflective, the received signal strength will decrease. When such environmental changes occur, resetting the point cloud density threshold used in the point cloud density determination process can improve the accuracy of determining track obstacles.

[0064] Environmental change indicators (env) fac Once the retention is complete, the accumulated environmental change indicators (env) for a predetermined number of frames are stored. fac Jud judgment index ind Calculate (Step 702). Judgment index ind While the appropriate selection is possible, the detection device 6 utilizes the standard deviation of the received signal strength. Specifically, the standard deviation of the received signal strength is monitored at predetermined frame intervals. The judgment index jud is calculated in Figure 17. ind The time change is plotted with time (frames) on the horizontal axis and the judgment index "jud" on the vertical axis. ind The magnitude is shown. In the figure, the moving object 2 is traveling from track 3 to track 4, where the conditions (material, pavement condition, etc.) are different. That is, the radar device 10 determines whether the conditions are different by determining whether the amplitude intensity of the reflected waves received by the radar device 10 differs by a certain amount or more. When the received intensity detected by the radar device 10 is divided into frame periods of a predetermined period, in period 85, the moving object 2 is traveling only on track 3, which is in the same condition, so the judgment index jud ind 72, that is, the standard deviation of the received signal strength becomes smaller. During period 86, the mobile object 2 moves from track 3 to track 4, which has a different state, so the judgment index jud during the period ind72 increases. During period 87, since the moving body 2 is traveling only on the road surface 4 with the same state, the determination index jud ind 72 decreases.

[0065] Next, an environmental change threshold th std is set (step 703). The environmental change threshold th std can be appropriately set according to the selected determination index jud ind or the assumed determination index jud ind variation, and a relative threshold based on the magnitude of the standard deviation of the past frames may be set. In the example of FIG. 17, the environmental change threshold th std is set to the value shown by line 77. Then, for each frame period, the determination index jud ind and the environmental change threshold th std are compared (step 704). And when the determination index jud ind exceeds the environmental change threshold th std , it is determined that there is an environmental change (step 705). On the other hand, when the determination index jud ind is less than or equal to the environmental change threshold th std , it is determined that there is no environmental change (step 706).

[0066] Note that the determination index jud ind may use the average value of the reception intensity. When using the standard deviation described above, the temporal change of the reception intensity was observed, but when using the average value, the change is captured on the distance axis. That is, from the point cloud information for each predetermined distance, the reception intensity is evaluated for each distance region. The magnitude of the change when the reception intensity at the initial position is set to 1 is observed. When it is determined that the amount of change is large based on the environmental change threshold, it is determined that there is an environmental change. The environmental change threshold may be an arbitrary set value, or a relative threshold based on the average reception intensity of the distance already traveled may be set.

[0067] By the detection method of monitoring the environmental change of the road surface as described above and updating the point cloud density threshold when a predetermined environmental change occurs, it is possible to accurately detect road surface obstacles even on a road surface with many irregularities.

[0068] Although the detection device, detection method, and program according to the present invention have been described above, the present invention is not limited to the embodiments described above, but includes all aspects included in the concept and claims of the present invention. [Explanation of symbols]

[0069] 1.6 Detection device 2 Mobile Units 3, 4 Runway 10. Sensor equipment (radar equipment) 11 Transmitting antenna 12 Receiving antenna 13 Transmission Control Unit 14 Signal Generation Unit 15. Oscillator 16 switches 17 ADC 18 Distance calculation section 19 Speed ​​calculation section 20. Threshold setting section 21 Detection unit 22 Angle calculation unit 23 Correction section 24 Filter section 25, 25' Track obstacle detection section 26, 26' Point cloud density determination section 27 Height determination unit 28 Track Data Monitoring Department 29 Higher-level equipment 31 Detection range 32, 33 detection points 40 Target (Route) 42, 43 area 51 Upper track 52. Track obstacles (steps, concave obstacles) 53 Lower track Average value of 70 detection points 72 Judgment index Thresholds 75, 76, 77 80 Warning display 85, 86, 87 period

Claims

1. A sensor device that periodically detects detection points on the path around a moving object, A track obstacle determination unit determines whether or not a track obstacle exists for each distance region obtained by dividing the track into multiple distance regions, based on the detection points. Equipped with, The aforementioned track obstacle determination unit includes a point cloud density determination unit, and the point cloud density determination unit is For each distance region, the point cloud density of the detected points is compared with a predetermined point cloud density threshold. When a distance region having a point cloud density smaller than the point cloud density threshold exists, and the distance between the distance region and the moving object is smaller than a predetermined risk threshold, it is determined that the obstacle on the track exists. The point cloud density threshold is set based on the average value of the point cloud density of detected points in a predetermined interval. The aforementioned track obstacle detection unit includes a track data monitoring unit that detects changes in the track environment based on the detection information of the sensor device. The point cloud density determination unit updates the point cloud density threshold when the environmental change is detected. Detection device.

2. A sensor device that periodically detects detection points on the path around a moving object, A track obstacle determination unit determines whether or not a track obstacle exists for each distance region obtained by dividing the track into multiple distance regions, based on the detection points. Equipped with, The aforementioned track obstacle determination unit includes a point cloud density determination unit, and the point cloud density determination unit is For each distance region, the point cloud density of the detected points is compared with a predetermined point cloud density threshold. When a distance region having a point cloud density smaller than the point cloud density threshold exists, and the distance between the distance region and the moving object is smaller than a predetermined risk threshold, it is determined that the obstacle on the track exists. The point cloud density threshold is set based on the average number of point clouds of detected points detected over a predetermined period. Detection device.

3. The aforementioned track obstacle detection unit includes a track data monitoring unit that detects changes in the track environment based on the detection information of the sensor device. The point cloud density determination unit updates the point cloud density threshold when the environmental change is detected. The detection device according to claim 2.

4. The aforementioned track obstacle detection unit includes a height determination unit, and the height determination unit is For each of the distance regions, the average value of the height of the detection point is compared with a predetermined height threshold. When there exists a distance region having the average value smaller than the height threshold, it is determined that the obstacle on the track exists. The detection device according to any one of claims 1 to 3.

5. The detection device according to any one of claims 1 to 4, wherein the track obstacle determination unit issues a warning when a track obstacle is present.

6. The sensor device periodically detects detection points on the path around the moving object, The steps include: determining whether or not an obstacle exists on the track based on the detection points for each distance region obtained by dividing the track into multiple distance regions; Includes, The aforementioned determination step is, For each distance region, the step is to compare the point cloud density of the detected points with a predetermined point cloud density threshold. The steps include determining that a track obstacle exists when there is a distance region having a point cloud density smaller than the point cloud density threshold, and the distance between the distance region and the moving object is smaller than a predetermined risk threshold, Includes, The point cloud density threshold is set based on the average value of the point cloud density of detected points in a predetermined interval. The aforementioned determination step is, Based on the detection information from the sensor device, the steps include detecting changes in the environment of the track, When the aforementioned environmental change is detected, the point cloud density threshold is updated. Further including, Detection method.

7. The sensor device periodically detects detection points on the path around the moving object, The steps include: determining whether or not an obstacle exists on the track based on the detection points for each distance region obtained by dividing the track into multiple distance regions; Includes, The aforementioned determination step is, For each distance region, the step is to compare the point cloud density of the detected points with a predetermined point cloud density threshold. The steps include determining that a track obstacle exists when there is a distance region having a point cloud density smaller than the point cloud density threshold, and the distance between the distance region and the moving object is smaller than a predetermined risk threshold, Includes, The point cloud density threshold is set based on the average number of point clouds of detected points detected over a predetermined period. Detection method.

8. On the computer, The sensor device has the function of periodically detecting detection points on the path around the moving object, The function includes determining whether or not obstacles exist on the track based on the detection points for each distance region into which the track is divided, This is a program to achieve this. The function for making the determination is, For each distance region, the point cloud density of the detected points is compared with a predetermined point cloud density threshold. If there is a distance region having a point cloud density smaller than the point cloud density threshold, and the distance between that distance region and the moving object is smaller than a predetermined risk threshold, then it is determined that the obstacle on the track exists. Includes functions, The point cloud density threshold is set based on the average value of the point cloud density of detected points in a predetermined interval. The function for making the determination is, Based on the detection information from the aforementioned sensor device, changes in the environment of the track are detected. When the aforementioned environmental change is detected, the point cloud density threshold is updated. Includes further features, program.

9. On the computer, The sensor device has the function of periodically detecting detection points on the path around the moving object, The function includes determining whether or not obstacles exist on the track based on the detection points for each distance region into which the track is divided, This is a program to achieve this. The function for making the determination is, For each distance region, the point cloud density of the detected points is compared with a predetermined point cloud density threshold. If there is a distance region having a point cloud density smaller than the point cloud density threshold, and the distance between that distance region and the moving object is smaller than a predetermined risk threshold, then it is determined that the obstacle on the track exists. Includes functions, The point cloud density threshold is set based on the average number of point clouds of detected points detected over a predetermined period. program.

Images