Object detection device

Abstract

To further appropriately update an unevenness map indicating road surface height.SOLUTION: An object detection device includes a laser radar which irradiates a monitoring area with a laser beam and receives reflection light of the emitted laser beam, a map generation unit which generates an unevenness map indicating the road surface height for each cell inside the monitoring area based on the received reflection light and updates the unevenness map, and an object detection unit which detects a detection object on a road surface inside the monitoring area based on the updated unevenness map and the reflection light received by the laser radar. The map generation unit updates the unevenness map so as to satisfy a constraint condition in which the change in the road surface height is determined in advance.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] The present invention relates to an object detection device. [Background technology]

[0002] For example, there is an object detection device that detects objects using a laser radar, as described in Patent Document 1. This object detection device detects obstacles and the like on the road surface within the monitoring area by irradiating laser light from the laser radar to multiple positions within a monitoring area set within a railroad crossing and calculating the three-dimensional positions of the reflection points of the irradiated laser light.

[0003] Furthermore, such object detection devices detect the road surface height in advance by irradiating a monitoring area with a laser beam and generate an unevenness map indicating the road surface height at each position, and then detect obstacles and the like on the road surface based on the generated unevenness map. [Prior art documents] [Patent documents]

[0004] [Patent Document 1] Japanese Patent Application Laid-Open No. 2010-249569 Summary of the Invention [Problem to be solved by the invention]

[0005] Here, for example, road surface height may change due to snow accumulation, etc. For this reason, it is considered that the object detection device updates the unevenness map used for detecting pedestrians, etc. at appropriate times. When updating this unevenness map, it is required to update the unevenness map indicating the road surface height more appropriately so that pedestrians on the road surface can be detected with high accuracy even after the update. [Means for solving the problem]

[0006] An object detection device according to one aspect of the present invention comprises a laser radar that irradiates a monitoring area set within a railroad crossing with laser light and receives reflected light of the irradiated laser light; a map generation unit that detects the road surface height for each of a plurality of cells that make up the monitoring area based on the received reflected light, generates an unevenness map showing the road surface height for each cell within the monitoring area, and updates the unevenness map using the generated unevenness map; and an object detection unit that detects a detected object on the road surface within the monitoring area based on the updated unevenness map and the reflected light received by the laser radar, and the map generation unit updates the unevenness map so that changes in road surface height satisfy predetermined constraints.

[0007] In this object detection device, when updating the unevenness map, a constraint is set on changes in road surface height. Therefore, the object detection device does not update the unevenness map in a way that would cause changes in road surface height to not satisfy the constraint, but updates the unevenness map in a way that satisfies the constraint. This allows the object detection device to more appropriately update the unevenness map that indicates road surface height.

[0008] In the object detection device, an upper limit road surface height in the case where the road surface height increases before and after updating the unevenness map is set as a constraint condition, the upper limit road surface height being a road surface height that is higher by a predetermined height than the road surface height of the cell indicated by a predetermined initial unevenness map, and the map generation unit may update the unevenness map using the upper limit of the road surface height as the upper limit road surface height in the case where the road surface height of the cell increases before and after updating the unevenness map. In this case, the object detection device can prevent the road surface height from becoming unnecessarily higher than the road surface height indicated by the initial unevenness map by updating the unevenness map.

[0009] In the object detection device, a lower limit road surface height is set as a constraint condition in the case where the road surface height becomes lower before and after updating the unevenness map, the lower limit road surface height is a road surface height that is equal to or lower than the road surface height of a cell indicated by a predetermined initial unevenness map, and the map generation unit may update the unevenness map using the lower limit road surface height as the lower limit road surface height in the case where the road surface height of a cell becomes lower before and after updating the unevenness map. In this case, the object detection device can prevent the road surface height from becoming unnecessarily lower than the road surface height indicated by the initial unevenness map by updating the unevenness map.

[0010] In the object detection device, the lower limit road surface height may be the road surface height of a cell indicated by the initial unevenness map. Here, the initial unevenness map can be considered to have been set correctly as the road surface height was confirmed when the map was set. Therefore, it is unlikely that the detected road surface height will be lower than the road surface height indicated by the initial unevenness map. This allows the object detection device to more appropriately update the unevenness map by setting the lower limit road surface height to the road surface height indicated by the initial unevenness map when updating the unevenness map.

[0011] In the object detection device, an upper limit of the change in road surface height before and after updating the roughness map is set as a constraint, and if the road surface height of a cell increases before and after updating the roughness map, the map generation unit may update the roughness map using the upper limit of the change in road surface height as the upper limit of change in road surface height. In this case, the object detection device can prevent the road surface height indicated by the current roughness map from becoming higher than necessary as a result of updating the roughness map.

[0012] The object detection device may further include an input receiving unit that receives an input operation for not applying the upper limit of change amount set as a constraint condition, and when the input receiving unit receives the input operation for not applying the upper limit of change amount, the map generation unit may update the unevenness map without applying the upper limit of change amount set as a constraint condition. In this case, even if there is a constraint condition for the upper limit of change amount, the object detection device can switch between applying and not applying this constraint condition depending on whether or not there is an input operation for not applying the upper limit of change amount, thereby more appropriately updating the unevenness map.

[0013] In the object detection device, if the road surface height of a cell decreases before and after updating the unevenness map, the map generation unit may update the unevenness map using the newly detected road surface height. In this case, the object detection device can set an upper limit on the change in road surface height as a constraint when the road surface height increases, and update the unevenness map without any constraint when the road surface height decreases. Note that the update of the unevenness map on the side where the road surface height decreases is an update in a direction that makes it easier to detect a detection target on the road surface. In this way, by updating the unevenness map, the object detection device can reduce the risk of missing a detection target.

[0014] In the object detection device, when the object detection unit detects a detection object, the map generation unit may update the unevenness map for a cell of the monitoring area in which the detection object is present using the road surface height detected before the detection object was detected, rather than using the newly detected road surface height. In this case, the object detection device can accurately update the unevenness map indicating the road surface height without updating the unevenness map using the height position of the detected detection object as the height position of the road surface.

[0015] In the object detection device, the object detection unit may set a lower mask at a position higher than the road surface height by a predetermined height based on the updated unevenness map, and detect an object that exists at a position higher than the lower mask as the detection target based on the reflected light received by the laser radar. In this case, by using the lower mask, the object detection device can prevent the detection of an object that is not intended to be detected, such as a fallen leaf blown into the monitoring area by the wind, as the detection target.

[0016] In the object detection device, an upper limit may be set for the height position of the lower mask. Because the lower mask is set based on the unevenness map, the lower mask may be set at a position higher than necessary depending on the road surface height in the updated unevenness map. Even in such a case, the object detection device will not set the lower mask higher than necessary because an upper limit is set for the height position of the lower mask.

[0017] In the object detection device, the map generation unit may extract, as a missing cell, a cell in which the number of received reflected light beams is equal to or less than a predetermined threshold, and estimate the road surface height of the missing cell based on the road surface heights of the cells surrounding the missing cell. Here, if the number of received reflected light beams is small, the road surface height may not be detected accurately due to the influence of noise, etc. Therefore, the object detection device can estimate the road surface height of the missing cell while suppressing the influence of noise, etc. by estimating the road surface height of the missing cell based on the surrounding road surface heights.

[0018] An object detection device according to another aspect of the present invention comprises a laser radar that irradiates a monitoring area set within a railroad crossing with laser light and receives reflected light of the irradiated laser light; a map generation unit that detects the road surface height for each of a plurality of cells that make up the monitoring area based on the received reflected light, generates an unevenness map showing the road surface height for each cell in the monitoring area, and updates the unevenness map using the generated unevenness map; and an object detection unit that detects a detected object on the road surface within the monitoring area based on the updated unevenness map and the reflected light received by the laser radar.When a detected object is detected by the object detection unit, the map generation unit updates the unevenness map for cells in the monitoring area in which the detected object is present, using the road surface height detected before the detected object was detected, rather than using the newly detected road surface height.

[0019] In this object detection device, for cells in which a detection object exists, the unevenness map is updated using the road surface height detected before the detection object is detected. Therefore, the object detection device does not update the unevenness map by using the height position of the detected detection object as the height position of the road surface, and can accurately update the unevenness map indicating the road surface height. This allows the object detection device to more appropriately update the unevenness map indicating the road surface height. [Effects of the Invention]

[0020] According to various aspects of the present invention, it is possible to more appropriately update the unevenness map indicating the road surface height. [Brief explanation of the drawings]

[0021] [Figure 1] FIG. 1 is a block diagram showing a schematic configuration of an object detection device according to an embodiment. [Figure 2] FIG. 2 is a diagram showing the monitoring area set within the railroad crossing as seen from above. [Figure 3] FIG. 3 is a schematic diagram for explaining the setting of the unevenness map and the lower mask. [Figure 4]FIG. 4 is a schematic diagram showing a lower mask set around the object to be detected. [Figure 5] 5(a) to 5(c) are schematic diagrams for explaining how histograms of the light reception results of reflected light are synthesized. [Figure 6] Fig. 6(a) is a side view showing how a laser beam is blocked by an obstacle and an interpolation target cell is generated, and Fig. 6(b) is a plan view showing how a laser beam is blocked by an obstacle and an interpolation target cell is generated. [Figure 7] 7(a) to 7(c) are schematic diagrams for explaining how the unevenness map is updated in accordance with the height of grass and the like, and how the lower mask is set. [Figure 8] 8(a) to 8(c) are schematic diagrams for explaining how the unevenness map is updated without using the measurement points of the detected detection object. [Figure 9] 9(a) is a schematic diagram for explaining the constraints on the change in road surface height when the road surface height decreases before and after the update, and FIG. 9(b) is a schematic diagram for explaining the constraints on the change in road surface height when the road surface height increases before and after the update. [Figure 10] Fig. 10(a) is a schematic diagram for explaining constraints on changes in road surface height when the road surface height decreases after updating, and Fig. 10(b) is a schematic diagram for explaining constraints on changes in road surface height when the road surface height increases after updating. [Figure 11] 11(a) and 11(b) are schematic diagrams for explaining the upper limit of the lower mask. [Figure 12] FIG. 12 is a flowchart showing the flow of the unevenness map generation and update process performed in the object detection device. DETAILED DESCRIPTION OF THE INVENTION

[0022] Hereinafter, embodiments of the present invention will be described with reference to the drawings. In each drawing, the same or corresponding elements are designated by the same reference numerals, and redundant description will be omitted.

[0023] The object detection device 1 shown in FIGS. 1 and 2 detects a detection target within a monitoring area X. In this embodiment, the object detection device 1 detects a detection target within a monitoring area X set within a railroad crossing F where a railroad track R and a road L for automobiles intersect. The detection target here is, for example, a vehicle. The detection result of the object detection device 1 is output to a higher-level device of the object detection device 1, and is used in the higher-level device for various controls, such as control of railway signals and operation control of trains traveling on the railroad track R. The object detection device 1 includes a laser radar 10, a calculation unit 20, a display unit 30, and an input unit 40.

[0024] The display unit 30 is a display device (e.g., a monitor) for displaying and presenting various information such as images to the operator of the object detection device 1. The display unit 30 can display, for example, the detection status of the detection target object. The input unit 40 is an input device operated by the operator of the object detection device 1. The type of input device used as the input unit 40 is not limited.

[0025] The laser radar 10 irradiates laser light at multiple positions within a monitoring area X and receives reflected light of the irradiated laser light. The laser radar 10 is also called a LIDAR (Light Detection and Ranging). The laser radar 10 is fixed to a support member such as a pole installed on the ground. The laser radar 10 may also be fixed to, for example, the wall of a building. The laser radar 10 is installed so that it can look down from above on a detection target moving within the railroad crossing F.

[0026] The laser radar 10 includes an irradiating unit 11, a reflecting unit 12, and a light receiving unit 13. The irradiating unit 11 includes a light emitting element that generates laser light. The irradiating unit 11 irradiates the laser light at a predetermined irradiation cycle.

[0027] The reflecting unit 12 includes a mirror that reflects the laser light. The reflecting unit 12 reflects the laser light emitted from the emitting unit 11 toward the monitoring area X using the mirror. The reflecting unit 12 can also change the angle of the mirror. The reflecting unit 12 changes the angle of the mirror to change the reflection angle of the laser light, thereby sequentially reflecting the laser light to multiple positions within the monitoring area X.

[0028] Furthermore, the reflecting unit 12 reflects the reflected light that has reflected off an object and returned to the laser radar 10 toward the light receiving unit 13. The light receiving unit 13 receives the reflected light that has reflected off an object via the reflecting unit 12. The laser radar 10 outputs the result of receiving the laser light to the calculating unit 20.

[0029] The calculation unit 20 generates an unevenness map showing the unevenness of the road surface within the monitoring area X, and detects the detection target based on the generated unevenness map and the result of receiving the reflected laser light at the light receiving unit 13. The calculation unit 20 may be provided integrally with the laser radar 10, or may be provided at a location separate from the laser radar 10.

[0030] The calculation unit 20 is an electronic control unit having, for example, a CPU (Central Processing Unit), a ROM (Read Only Memory), and a RAM (Random Access Memory). The calculation unit 20 realizes various functions by, for example, loading a program stored in the ROM into the RAM and executing the program loaded into the RAM with the CPU. The calculation unit 20 may be composed of multiple electronic control units.

[0031] The calculation unit 20 functionally comprises a map generation unit 21, an input reception unit 22, and an object detection unit 23. The map generation unit 21 detects the road surface height for each of the multiple cells that make up the monitoring area X based on the reflected light received by the light receiving unit 13. Then, the map generation unit 21 generates an unevenness map that indicates the road surface height for each cell in the monitoring area X based on the detection results. The map generation unit 21 can detect the height (unevenness) of the road surface based on the height positions of the reflection points of the reflected light of the laser light.

[0032] As an example, the map generating unit 21 uses a reference plane H set within the monitoring area X as a reference as shown in FIG. 3, and calculates the height of each point on the road surface G using the reference plane H as a reference. The unevenness map indicates the unevenness of each cell on the road surface G, i.e., the height of each cell on the road surface G. The map generating unit 21 also generates the unevenness map at predetermined intervals and updates the unevenness map with the newly generated unevenness map. The generation and updating of the unevenness map performed by the map generating unit 21 will be described in detail below.

[0033] The input receiving unit 22 receives various input operations inputted into the input unit 40 by the operator of the object detection device 1. Here, the input receiving unit 22 can receive input operations from the operator related to the generation of a concavo-convex map performed by the map generating unit 21. The input operations received by the input receiving unit 22 will be described in detail later.

[0034] The object detection unit 23 detects an object to be detected on the road surface G within the monitoring area X based on the unevenness map generated by the map generation unit 21 and the reflected light received by the light receiving unit 13 of the laser radar 10. Here, the object detection unit 23 detects the object to be detected using a lower mask. Note that various well-known methods can be used as a method for detecting an object to be detected based on the detection result of the laser radar 10 performed by the object detection unit 23.

[0035] Specifically, as shown in Fig. 3, the object detection unit 23 sets a lower mask M at a position that is a predetermined height higher than the road surface height of the road surface G based on the generated (updated) unevenness map. In other words, the lower mask M has a shape that follows the unevenness of the road surface G. Then, the object detection unit 23 detects an object that exists at a position higher than the lower mask M as a detection target based on the reflected light received by the light receiving unit 13 of the laser radar 10. Note that the object detection unit 23 executes a detection process for a detection target at each predetermined detection period using the unevenness map generated (updated) by the map generation unit 21, etc.

[0036] 4, when a detection target P is detected, the object detection unit 23 sets the height position of the lower mask M low around the detection target P. As a result, even if the detection target P falls over and its height decreases, for example, as shown by the dashed line in FIG. 4, the object detection unit 23 can continue to detect the detection target P. In this way, when a detection target P is detected, the object detection unit 23 sequentially sets the lower mask M, which has been set to a low height position around the detection target P, so as to follow the movement of the detection target P.

[0037] Next, the details of generating and updating the unevenness map performed by the map generating unit 21 will be described. First, the generation of the unevenness map will be described. As described above, the map generating unit 21 detects the road surface height for each cell that constitutes the monitoring area X. Furthermore, the map generating unit 21 determines the road surface height of a cell that is the target of detection based on the light reception results of multiple reflected lights reflected by the cell. Here, the map generating unit 21 can determine the road surface height using various methods, such as setting the average or median value as the road surface height based on the multiple light reception results in the target cell, or setting the height with the greatest number of light reception results as the road surface height.

[0038] Here, for example, in a cell located far from the laser radar 10, the number of measurement points (received reflected light) at which heights are detected within the cell may be small. In this case, the light reception results may be easily affected by noise, and the road surface height within the cell may not be detected correctly. Therefore, as an example, the map generation unit 21 extracts cells in which the number of measurement points (received reflected light) is equal to or less than a predetermined threshold as missing cells. Then, the map generation unit 21 estimates the road surface height of the missing cell based on the road surface heights of the cells surrounding the missing cell. Note that the map generation unit 21 may estimate the road surface heights of all cells or a predetermined portion of cells based on the road surface heights of the cells surrounding the missing cell, regardless of whether the missing cell is one in which the number of measurement points is equal to or less than the threshold.

[0039] As an example of this estimation, the map generation unit 21 generates a histogram with the number of measurement points on the vertical axis and the road surface height on the horizontal axis based on the light reception results of the reflected light in the deficient cell, as shown in FIG. 5(a). Then, as shown in FIG. 5(b), the map generation unit 21 combines histograms of the cells surrounding the deficient cell with the histogram of the deficient cell. Here, the map generation unit 21 adds together bins in the histogram that indicate the same road surface height. Note that the hatched columnar portion in FIG. 5(b) is the combined portion of the histograms of the surrounding cells. The map generation unit 21 estimates the road surface height of the deficient cell based on the combined histogram (the histogram shown in FIG. 5(b)).

[0040] Furthermore, when combining histograms, the map generating unit 21 may combine a total of eight histograms of cells C2 to C9 surrounding the shortage cell C1 into the histogram of the shortage cell, as shown in Fig. 5(c). Note that the combination of histograms performed by the map generating unit 21 is not limited to simply adding together bins that indicate the same road surface height. Furthermore, although the histograms of the eight cells surrounding the shortage cell are combined into the histogram of the shortage cell, the number of histograms to be combined and the positions of the cells are not limited to this.

[0041] Here, for example, as shown in Figures 6(a) and 6(b), if there is an obstacle T (curb, snow, etc.) on the road surface G that blocks the laser light, it becomes difficult to detect the road surface height of a cell C located behind the obstacle as seen from the laser radar 10. Therefore, the map generation unit 21 extracts, from among the multiple cells C in the monitoring area X, a cell C whose received reflected light does not satisfy predetermined light reception conditions as an interpolation target cell Ca. Note that in Figures 6(a) and 6(b), the hatched cells C are interpolation target cells Ca.

[0042] Here, the light receiving condition is satisfied when the road surface height can be properly detected based on the received reflected light. The light receiving condition is not satisfied when the road surface height cannot be properly detected, for example, because the laser light is blocked by an obstacle T. As an example, the light receiving condition may be set based on the state of the measurement points detected in the target cell. This state of the measurement points may be the number of measurement points, or may be various states other than the number.

[0043] Then, the map generation unit 21 interpolates the road surface height of the interpolation target cell Ca based on the road surface height of the cell C adjacent to the interpolation target cell Ca. Specifically, the map generation unit 21 identifies a first cell, which is a cell C adjacent to the interpolation target cell Ca on the laser radar 10 side, and a second cell, which is a cell C adjacent to the interpolation target cell Ca on the opposite side from the laser radar 10 side. These identified first and second cells are different from the interpolation target cell Ca and are cells for which a road surface height has been detected. Then, the map generation unit 21 interpolates the road surface height of the interpolation target cell Ca based on the identified road surface height of the first cell and the road surface height of the second cell.

[0044] 6(a) and 6(b), there are cases where two or more interpolation target cells Ca are adjacent to each other to form an interpolation target cell group. In this case, the map generating unit 21 interpolates the road surface height of each of the interpolation target cells Ca based on the road surface height of a first cell adjacent to the interpolation target cell group on the laser radar 10 side and the road surface height of a second cell adjacent to the interpolation target cell group on the opposite side to the laser radar side.

[0045] Specifically, as shown in FIG. 6(b), the map generation unit 21 draws a straight line A from the laser radar 10 to the interpolation target cell Ca. Among the interpolation target cells Ca, the interpolation target cells Ca1 to Ca6 located on the straight line A are set as an interpolation target cell group CA for estimating the road surface heights of the interpolation target cells Ca1 to Ca6. The map generation unit 21 then identifies a first cell C11, which is a cell C adjacent to the interpolation target cell group CA on the laser radar 10 side, and a second cell C12, which is a cell C adjacent to the interpolation target cell group CA on the opposite side from the laser radar 10 side. The map generation unit 21 then estimates the road surface heights of each of the interpolation target cells Ca1 to Ca6 that make up the interpolation target cell group CA based on the road surface heights of the first cell C11 and the second cell C12. The map generation unit 21 also estimates the road surface heights of the other interpolation target cells Ca other than the interpolation target cells Ca1 to Ca6 using the same method.

[0046] The map generating unit 21 can interpolate the road surface height of the interpolation target cell Ca so that it is equal to or less than the linear interpolation height obtained by linearly interpolating the road surface height of the first cell C11 and the road surface height of the second cell C12. In this way, the map generating unit 21 can interpolate the road surface height of the interpolation target cell Ca by various methods based on the road surface heights of the first cell C11 and the second cell C12, as long as it is equal to or less than the linear interpolation height.

[0047] Also, here, an example is given in which an interpolation target cell Ca occurs due to being blocked by an obstacle T. However, the present invention is not limited to this, and even if an interpolation target cell Ca occurs that does not satisfy the light receiving conditions due to a cause other than the obstacle T, the road surface height can be interpolated using the above-described method.

[0048] Next, updating of the unevenness map will be described. For example, from the state of the road surface G shown in FIG. 7(a), grass W may grow on the road surface G as shown in FIG. 7(b), or snow S may accumulate on the road surface G. In such a case, when the object detection unit 23 performs a detection process for the detection object using the lower mask M set based on the unevenness map of the road surface G shown in FIG. 7(a), the object detection unit 23 may erroneously detect the grass W or the like as the detection object. Note that the grass W and snow S are only examples, and other objects may be detected as the detection object. For this reason, the map generation unit 21 generates an unevenness map at predetermined time intervals based on the reflected light received by the light receiving unit 13, and updates the unevenness map with the newly generated unevenness map.

[0049] In other words, the road surface height of the road surface G indicated by the unevenness map is not the height of only the road surface G excluding the grass W, snow S, etc., but is a height including the heights of objects other than the detection target, such as the grass W and snow S. For example, if grass W exists on the road surface G, the height of the upper end of the grass W is set as the road surface height of the portion where this grass W exists.

[0050] In this case, the object detection unit 23 sets the lower mask M based on an unevenness map that takes into account the grass W and snow S, as shown in Fig. 7(b). That is, as shown in Fig. 7(b), the lower mask M is set at a position that is a predetermined height higher than the upper ends of the grass W and snow S in the area where the grass W and snow S are present. As a result, even if the grass W on the road surface G grows and becomes taller, or the snow S accumulates and becomes taller, the lower mask M is set in accordance with this change, so the grass W and snow S will not be detected as detection objects.

[0051] 7(c), the height position of the lower mask M is set low around the detection object P. However, even if the lower mask M is set low, it is higher than the grass W and snow S, so the grass W and snow S are not detected as the detection object.

[0052] When the detection object P is detected, the map generation unit 21 does not use the measurement point of the detection object P (light reflected by the detection object P) to update the unevenness map. Specifically, for example, assume that an unevenness map of the road surface G has been generated as shown in FIG. 8(a). Then, assume that the detection object P moves and the object detection unit 23 detects the detection object P on the road surface G as shown in FIG. 8(b).

[0053] In this state, when the map generation unit 21 updates the unevenness map, the map generation unit 21 does not use the newly detected height as the road surface height for cells in the monitoring area X where the detection object P is present. In other words, the map generation unit 21 does not use the height detected based on the measurement point K1 reflected by the detection object P, among the multiple measurement points K detected by the laser radar 10, as the road surface height. Note that the map generation unit 21 acquires information about the detection object P from the object detection unit 23. Then, for cells in which the detection object P is present, the map generation unit 21 generates a new unevenness map using the road surface height detected before the detection object P was detected, and updates the unevenness map.

[0054] As a result, even if the detection object P is present, the newly generated unevenness map is an unevenness map for a state in which the detection object P is not present. For this reason, as shown in FIG. 8(c), even if the detection object P is present, a lower mask M is set according to the road surface height of the actual road surface G, and the detection object P is detected based on the lower mask M. For example, in the example shown in FIG. 8(c), the detection object P is in a state in which it has fallen over, but the road surface height indicated by the unevenness map is the height of the actual road surface G. For this reason, even if the detection object P falls while walking and the unevenness map is updated immediately thereafter, the object detection unit 23 can continue to detect the detection object P.

[0055] Furthermore, if the road surface height of a cell changes before and after the update, the map generation unit 21 updates the roughness map so that the change in road surface height for that cell satisfies predetermined constraint conditions. In this embodiment, the following constraint conditions (1) to (4) are predetermined as constraint conditions. The map generation unit 21 adjusts the road surface height as necessary and updates the roughness map so that these constraint conditions (1) to (4) are satisfied.

[0056] Constraints (1) and (2) are constraints on the change from the road surface height before updating. In other words, they are constraints on the amount of road surface height update per update. Constraint (1) will be explained using FIG. 9(a). In FIG. 9(a), the left side of the figure shows the road surface height before updating with a dashed line, and the right side of the figure shows the road surface height detected this time with a dashed line. The same illustration method as FIG. 9(a) is used for FIGS. 9(b), 10(a), and 10(b).

[0057] 9(a), if the road surface height of a cell is lower before and after updating the unevenness map, the map generation unit 21 updates the unevenness map using the newly detected road surface height. That is, as constraint condition (1), a condition is set that if the road surface height of a cell is lower before and after updating the unevenness map, there is no constraint on updating the road surface height.

[0058] As shown in Fig. 9(b), an upper limit of the change in road surface height before and after updating the roughness map is set as constraint condition (2). If the road surface height of a cell increases before and after updating the roughness map, the map generation unit 21 updates the roughness map using the upper limit of the change in road surface height as the upper limit change amount. In other words, if the road surface height increases, the maximum value of the change in road surface height per update becomes the upper limit change amount.

[0059] The operator of the object detection device 1 can input an input operation to not apply the upper limit change amount set as the constraint condition (2) through the input unit 40. The input receiving unit 22 receives the input operation to not apply the upper limit change amount input through the input unit 40. When the input receiving unit 22 receives the input operation to not apply the upper limit change amount, the map generation unit 21 updates the unevenness map without applying the upper limit change amount set as the constraint condition (2).

[0060] For example, if the road surface height has changed significantly due to snow accumulation, the operator inputs a command not to apply the upper limit change amount, which causes the map generator 21 to generate an unevenness map indicating the current road surface height when updating the unevenness map.

[0061] Constraints (3) and (4) are constraints on the change from the road surface height indicated by the initial unevenness map. As shown in FIG. 10(a), constraint (3) sets a lower limit road surface height in the case where the road surface height decreases before and after updating the unevenness map. This lower limit road surface height is equal to or lower than the road surface height of the cell indicated by the predetermined initial unevenness map. If the road surface height of the cell decreases before and after updating the unevenness map, the map generation unit 21 updates the unevenness map with the lower limit of the road surface height set to the lower limit road surface height. In other words, even if the road surface height decreases before and after updating, the road surface height will not be lower than the lower limit road surface height. Note that this lower limit road surface height may be the road surface height of the cell indicated by the initial unevenness map.

[0062] As shown in FIG. 10(b), a maximum road surface height is set as constraint condition (4) in the case where the road surface height increases before and after updating the unevenness map. This maximum road surface height is a road surface height that is a predetermined height higher than the road surface height of the cell indicated by the predetermined initial unevenness map. If the road surface height of the cell increases before and after updating the unevenness map, the map generation unit 21 updates the unevenness map using the upper limit of the road surface height as the maximum road surface height. In other words, even if the road surface height increases before and after updating, the road surface height will not exceed the maximum road surface height. The height difference between the maximum road surface height in constraint condition (4) and the road surface height of the cell indicated by the initial unevenness map is greater than the maximum change amount of the road surface height in constraint condition (2).

[0063] In this way, if the road surface height decreases before and after the update, the map generation unit 21 updates the unevenness map by setting the road surface height as necessary to satisfy the constraints (1) and (3). Also, if the road surface height increases before and after the update, the map generation unit 21 updates the unevenness map by setting the road surface height as necessary to satisfy the constraints (2) and (4).

[0064] Next, the upper limit of the lower mask will be described. As shown in FIG. 11(a), an upper limit M1 is set as the height position of the lower mask M. The upper limit M1 is set, for example, at a position that is a predetermined height higher than a reference plane H set within the monitoring area X. However, the upper limit M1 may also be set at a position that is a predetermined height higher than the road surface height indicated by the initial unevenness map so as to follow the unevenness of the road surface height indicated by the preset initial unevenness map. As shown in FIG. 11(b), the road surface height indicated by the updated unevenness map may increase due to grass W growing on the road surface G or due to the accumulation of snow S. Even in this case, the object detection unit 23 sets the lower mask M based on the road surface height indicated by the updated unevenness map so that the lower mask M does not exceed the upper limit M1.

[0065] Next, the flow of the unevenness map generation and update process performed in the object detection device 1 will be described using the flowchart in Fig. 12. The flowchart shown in Fig. 12 is repeatedly executed at predetermined time intervals. In other words, the unevenness map is updated at predetermined time intervals.

[0066] 12, the map generating unit 21 synthesizes histograms for the missing cells in which the number of measurement points is equal to or less than the threshold value, as described with reference to FIGS. 5(a) to 5(c) (S101). Next, the map generating unit 21 detects the road surface height for each cell constituting the monitoring area X (S102). Furthermore, for an interpolation target cell that does not satisfy the light receiving condition due to the presence of an obstacle or the like, the map generating unit 21 interpolates the road surface height based on the road surface heights of the first and second cells adjacent to the interpolation target cell (S103).

[0067] The map generating unit 21 generates a roughness map based on the detected and interpolated road surface height (S104). At that time, the map generating unit 21 generates a new roughness map so as to satisfy the above-mentioned constraints (1) to (4). Then, the map generating unit 21 updates the roughness map with the generated new roughness map (S105).

[0068] Here, the map generating unit 21 generates and updates the unevenness map at predetermined time intervals, but the generation and updating are not limited to these timings. For example, the operator can input an input operation for forcibly updating the unevenness map through the input unit 40. When this input operation for forcibly updating is received by the input receiving unit 22, the map generating unit 21 may perform the process of generating and updating the unevenness map.

[0069] In this object detection device 1, constraints are set on changes in road surface height when updating the unevenness map. Therefore, the object detection device 1 does not update the unevenness map in a way that would cause changes in road surface height to not satisfy the constraints, but updates the unevenness map so that the constraints are satisfied. This allows the object detection device 1 to more appropriately update the unevenness map that indicates the road surface height.

[0070] In the object detection device 1, an upper limit on the amount of change in road surface height before and after updating the unevenness map is set as constraint condition (2). This enables the object detection device 1 to prevent the road surface height indicated by the current unevenness map from becoming higher than necessary as a result of updating the unevenness map.

[0071] When an input operation for not applying the upper limit change amount is received by the input receiving unit 22, the map generating unit 21 updates the unevenness map without applying the upper limit change amount set as the constraint condition (2). In this case, even if the constraint condition (2) for the upper limit change amount exists, the object detection device 1 can switch between applying and not applying this constraint condition (2) depending on whether or not there is an input operation for not applying the upper limit change amount, thereby more appropriately updating the unevenness map.

[0072] In the object detection device 1, a constraint (1) is set such that if the road surface height of a cell decreases before and after updating the unevenness map, there is no constraint on updating the road surface height. In this case, the map generation unit 21 sets an upper limit on the amount of change in road surface height as constraint (2) when the road surface height increases, but can update the unevenness map without any constraint when the road surface height decreases. Note that the update of the unevenness map on the side where the road surface height decreases is an update in a direction that makes it easier to detect the detection target P on the road surface. As a result, by updating the unevenness map in this way, the object detection device 1 can reduce the risk of missing a detection target P.

[0073] In the object detection device 1, a lower limit road surface height in the case where the road surface height decreases before and after updating the unevenness map is set as constraint condition (3). This enables the object detection device 1 to prevent the road surface height from becoming unnecessarily lower than the road surface height indicated by the initial unevenness map due to updating the unevenness map.

[0074] Furthermore, the lower limit road surface height in constraint condition (3) is the road surface height of the cell indicated by the initial unevenness map. Here, the initial unevenness map can be considered to have been set correctly by performing height confirmation work etc. when it was set. Therefore, it is difficult to imagine that the detected road surface height will be lower than the road surface height indicated by the initial unevenness map. As a result, the object detection device 1 can more appropriately update the unevenness map by setting the lower limit road surface height to the road surface height indicated by the initial unevenness map when updating the unevenness map.

[0075] In the object detection device 1, an upper limit road surface height in the case where the road surface height increases before and after updating the unevenness map is set as constraint condition (4). This enables the object detection device 1 to prevent the road surface height from becoming unnecessarily higher than the road surface height indicated by the initial unevenness map when the unevenness map is updated.

[0076] For cells in the monitoring area X where a detection object exists, the map generating unit 21 updates the unevenness map using the road surface height detected before the detection object is detected. In this case, the map generating unit 21 does not update the unevenness map by using the height position of the detected detection object P as the height position of the road surface, and can accurately update the unevenness map that indicates the road surface height.

[0077] The object detection unit 23 sets a lower mask M based on the updated unevenness map, and detects an object that is located higher than the lower mask M as the detection object P. In this case, by using the lower mask M, the object detection unit 23 can prevent the detection of an object that is not intended to be detected, such as a fallen leaf blown down into the monitoring area X by the wind, as the detection object P.

[0078] An upper limit M1 is set for the height position of the lower mask M. Because the lower mask M is set based on the unevenness map, depending on the road surface height in the updated unevenness map, the lower mask M may be set at a position higher than necessary. Even in such a case, the object detection unit 23 will not set the lower mask higher than necessary because an upper limit is set for the height position of the lower mask.

[0079] The map generating unit 21 estimates the road surface height of a deficient cell for which the number of received reflected light beams is equal to or less than a predetermined threshold, based on the road surface heights of the cells surrounding the deficient cell. Here, if the number of received reflected light beams (the number of measurement points) is small, the road surface height may not be detected accurately due to the influence of noise, etc. For this reason, the map generating unit 21 estimates the road surface height of the deficient cell based on the surrounding road surface heights, thereby making it possible to estimate the road surface height of the deficient cell while suppressing the influence of noise, etc.

[0080] Although the embodiments of the present invention have been described above, the present invention is not limited to the above-described embodiments.

[0081] In the object detection device 1 of the above embodiment, the lower mask M is set in accordance with the concavities and convexities of the concavity and convexity map, and the height position of the reflected light received by the light receiving unit 13 is compared with the lower mask M to detect the detection target P. This means that, for example, the lower mask M of a fixed height is set at a position higher by a predetermined height than the reference plane H. Then, the detection target P is detected by comparing the value obtained by subtracting the height corresponding to the concavities and convexities of the concavity and convexity map from the height position of the reflected light received by the light receiving unit 13 with the lower mask M of the fixed height.

[0082] The gist of the present invention will be described below. [Invention 1] a laser radar that irradiates a laser beam within a monitoring area set within the railroad crossing and receives reflected light of the irradiated laser beam; a map generating unit that detects a road surface height for each of a plurality of cells constituting the monitoring area based on the received reflected light, generates an unevenness map indicating the road surface height for each of the cells in the monitoring area, and updates the unevenness map using the generated unevenness map; an object detection unit that detects an object to be detected on a road surface within the monitoring area based on the updated unevenness map and the reflected light received by the laser radar; Equipped with The map generation unit updates the unevenness map so that the change in road surface height satisfies a predetermined constraint condition. [Invention 2] an upper limit road surface height in a case where the road surface height increases before and after updating the unevenness map is set as the constraint condition; the upper limit road surface height is the road surface height that is higher by a predetermined height than the road surface height of the cell indicated by a predetermined initial unevenness map, The object detection device described in Invention 1, wherein the map generation unit updates the unevenness map by setting the upper limit of the road surface height as the upper limit road surface height when the road surface height of the cell becomes higher before and after updating the unevenness map. [Invention 3] a lower limit road surface height in a case where the road surface height decreases before and after updating the unevenness map is set as the constraint condition; the lower limit road surface height is the road surface height that is equal to or lower than the road surface height of the cell indicated by a predetermined initial unevenness map, The object detection device according to invention 1 or 2, wherein the map generation unit updates the unevenness map by setting the lower limit of the road surface height as the lower limit road surface height when the road surface height of the cell becomes lower before and after updating the unevenness map. [Invention 4] 4. The object detection device according to claim 3, wherein the lower limit road surface height is the road surface height of the cell indicated by the initial unevenness map. [Invention 5] an upper limit of a change amount of the road surface height before and after updating the unevenness map is set as the constraint condition; An object detection device described in any one of Inventions 1 to 4, wherein the map generation unit updates the unevenness map when the road surface height of the cell increases before and after updating the unevenness map, using the upper limit of the change in road surface height as the upper limit change amount. [Invention 6] an input receiving unit that receives an input operation for not applying the upper limit change amount set as the constraint condition, The object detection device according to invention 5, wherein the map generation unit updates the unevenness map without applying the upper limit change amount set as the constraint condition when the input operation that does not apply the upper limit change amount is accepted by the input acceptance unit. [Invention 7] The object detection device according to invention 5 or 6, wherein the map generation unit updates the unevenness map using the newly detected road surface height if the road surface height of the cell becomes lower before and after updating the unevenness map. [Invention 8] An object detection device as described in any one of Inventions 1 to 7, wherein when the object to be detected is detected by the object detection unit, the map generation unit updates the unevenness map for the cell in the monitoring area in which the object to be detected is present, using the road surface height detected before the object to be detected, rather than using the newly detected road surface height. [Invention 9] An object detection device as described in any one of Inventions 1 to 8, wherein the object detection unit sets a lower mask at a position that is a predetermined height higher than the road surface height based on the updated unevenness map, and detects an object that exists at a position higher than the lower mask as the detection object based on the reflected light received by the laser radar. [Invention 10] 10. The object detection device according to claim 9, wherein an upper limit is set for the height position of the lower mask. [Invention 11] The map generation unit extracting the cell in which the number of received reflected lights is equal to or less than a predetermined threshold as a deficient cell; 11. The object detection device according to any one of inventions 1 to 10, wherein the road surface height of the missing cell is estimated based on the road surface heights of the cells surrounding the missing cell. [Invention 12] a laser radar that irradiates a laser beam within a monitoring area set within the railroad crossing and receives reflected light of the irradiated laser beam; a map generating unit that detects a road surface height for each of a plurality of cells constituting the monitoring area based on the received reflected light, generates an unevenness map indicating the road surface height for each of the cells in the monitoring area, and updates the unevenness map using the generated unevenness map; an object detection unit that detects an object to be detected on a road surface within the monitoring area based on the updated unevenness map and the reflected light received by the laser radar; Equipped with When the object detection unit detects the detection object, the map generation unit updates the unevenness map for the cell in the monitoring area in which the detection object is present, using the road surface height detected before the detection object was detected, rather than using the newly detected road surface height. [Explanation of symbols]

[0083] 1. Object detection device 10 Laser radar 21 Map generation unit 22 Input reception section 23 Object detection unit C, C2~C9 cells C1 Missing Cell F railroad crossing M Lower Mask M1 upper limit P Object to be detected X monitoring area

Claims

1. a laser radar that irradiates a laser beam within a monitoring area set within the railroad crossing and receives reflected light of the irradiated laser beam; a map generating unit that detects a road surface height for each of a plurality of cells constituting the monitoring area based on the received reflected light, generates an unevenness map indicating the road surface height for each of the cells in the monitoring area, and updates the unevenness map using the generated unevenness map; an object detection unit that detects an object to be detected on a road surface within the monitoring area based on the updated unevenness map and the reflected light received by the laser radar; Equipped with The map generation unit updates the unevenness map so that the change in road surface height satisfies a predetermined constraint condition.

2. an upper limit road surface height in a case where the road surface height increases before and after updating the unevenness map is set as the constraint condition; the upper limit road surface height is the road surface height that is higher by a predetermined height than the road surface height of the cell indicated by a predetermined initial unevenness map, 2. The object detection device according to claim 1, wherein, when the road surface height of the cell increases before and after updating the unevenness map, the map generation unit updates the unevenness map by setting an upper limit of the road surface height as the upper limit road surface height.

3. a lower limit road surface height in a case where the road surface height decreases before and after updating the unevenness map is set as the constraint condition; the lower limit road surface height is the road surface height that is equal to or lower than the road surface height of the cell indicated by a predetermined initial unevenness map, 2. The object detection device according to claim 1, wherein, when the road surface height of the cell becomes lower before and after updating the unevenness map, the map generation unit updates the unevenness map by setting a lower limit of the road surface height as the lower limit road surface height.

4. The object detection device according to claim 3 , wherein the lower limit road surface height is the road surface height of the cell indicated by the initial unevenness map.

5. an upper limit of a change amount of the road surface height before and after updating the unevenness map is set as the constraint condition; 2. The object detection device according to claim 1, wherein, when the road surface height of the cell increases before and after updating the unevenness map, the map generation unit updates the unevenness map using an upper limit of the amount of change in the road surface height as the upper limit change amount.

6. an input receiving unit that receives an input operation for not applying the upper limit change amount set as the constraint condition, 6. The object detection device according to claim 5, wherein, when the input operation to which the upper limit change amount is not applied is accepted by the input accepting unit, the map generating unit updates the unevenness map without applying the upper limit change amount set as the constraint condition.

7. 6. The object detection device according to claim 5, wherein, when the road surface height of the cell becomes lower before and after updating the unevenness map, the map generation unit updates the unevenness map using the newly detected road surface height.

8. 2. The object detection device according to claim 1, wherein, when the object to be detected is detected by the object detection unit, the map generation unit updates the unevenness map for the cell in the monitoring area in which the object to be detected is present, using the road surface height detected before the object to be detected, rather than using the newly detected road surface height.

9. 2. The object detection device according to claim 1, wherein the object detection unit sets a lower mask at a position that is a predetermined height higher than the road surface height based on the updated unevenness map, and detects an object that exists at a position higher than the lower mask as the detection object based on the reflected light received by the laser radar.

10. The object detection device according to claim 9 , wherein an upper limit is set for the height position of the lower mask.

11. The map generation unit extracting the cell in which the number of received reflected lights is equal to or less than a predetermined threshold as a deficient cell; 11. The object detection device according to claim 1, wherein the road surface height of the missing cell is estimated based on the road surface heights of the cells surrounding the missing cell.

12. a laser radar that irradiates a laser beam within a monitoring area set within the railroad crossing and receives reflected light of the irradiated laser beam; a map generating unit that detects a road surface height for each of a plurality of cells constituting the monitoring area based on the received reflected light, generates an unevenness map indicating the road surface height for each of the cells in the monitoring area, and updates the unevenness map using the generated unevenness map; an object detection unit that detects an object to be detected on a road surface within the monitoring area based on the updated unevenness map and the reflected light received by the laser radar; Equipped with When the object detection unit detects the detection object, the map generation unit updates the unevenness map for the cell in the monitoring area in which the detection object is present, using the road surface height detected before the detection object was detected, rather than using the newly detected road surface height.

Images