JPWO2021038740A1 - Obstacle detection device - Google Patents

Abstract

The detection unit (11) detects whether or not the distance measuring sensor (2) provided in the own vehicle (20) has received the reflected wave. The recognition unit (12) divides the reflected wave detected by the detection unit (11) into a reference reflected wave group reflected by the reference obstacle or a target reflected wave group reflected by the target obstacle for which the type is to be determined. Classify. The normalization unit (13) sets the values correlated with the sizes of the plurality of reflected waves classified in the target reflected wave group to the values correlated with the sizes of the plurality of reflected waves classified into the reference reflected wave group. Normalize based on. The discriminating unit (14) calculates two feature quantities using values correlated with the normalized magnitudes of a plurality of reflected waves classified into the target reflected wave group, and is based on the two feature quantities. To determine the type of target obstacle.

Description

The present invention relates to an obstacle detection device.

For the following two reasons, there is a demand for technology for detecting the height of obstacles around the vehicle.
(1) To determine whether the obstacle is a high obstacle that collides with the bumper of the vehicle or a low obstacle that does not collide with the bumper when the vehicle advances or retreats, and suppresses unnecessary warnings or brakes for the low obstacle ( 2) To make it easier for occupants to get on and off by parking the vehicle with an appropriate clearance against low obstacles that do not collide with the bumper when the vehicle is parked.

The object detection device according to Patent Document 1 observes a time change in the reflected wave intensity when a vehicle approaches an object by using an ultrasonic sensor, and is a low object when the reflected wave intensity changes from an increase to a decrease when the vehicle approaches. It was judged that.

Japanese Unexamined Patent Publication No. 2016-80639

Since the object detection device according to Patent Document 1 is configured as described above, there is a problem that the height of the obstacle cannot be determined unless the vehicle approaches the obstacle directly. Therefore, for example, when parking, if a vehicle equipped with an object detection device does not enter the parking space, obstacles of different heights such as parked vehicles, walls, curbs, ring fasteners, and steps existing in the parking space will be encountered. Cannot be determined.

The present invention has been made to solve the above-mentioned problems, and provides an obstacle detection device that does not require a vehicle to approach an obstacle in front of the obstacle in order to determine the type of the obstacle. The purpose is.

The obstacle detection device according to the present invention has a detection unit that detects whether or not a distance measuring sensor provided in the vehicle has received a reflected wave, and a reference obstacle that reflects the reflected wave detected by the detection unit. There is a correlation between the recognition unit that classifies into the reference reflected wave group or the target reflected wave group reflected by the target obstacle for which the type is to be determined, and the size of multiple reflected waves classified into the target reflected wave group. The normalization section that normalizes the values based on the values correlated with the magnitude of the multiple reflected waves classified in the reference reflected wave group, and the normalization of the multiple reflected waves classified in the target reflected wave group. A discriminant unit that calculates two feature quantities using values that correlate with the later size and discriminates the type of the target obstacle based on the two feature quantities, and a discriminant unit that discriminates the target obstacle. It is provided with an output unit that outputs information indicating the type.

According to the present invention, two feature quantities are calculated using values correlated with the normalized magnitudes of a plurality of reflected waves classified into the target reflected wave group, and based on the two feature quantities. Since the type of the target obstacle is discriminated, it is possible to provide an obstacle detection device that does not require the vehicle to face and approach the obstacle in order to discriminate the type of the obstacle.

It is a block diagram which shows the structural example of the obstacle detection apparatus which concerns on Embodiment 1. FIG. It is a graph which shows the obstacle detection method by the detection part of Embodiment 1. FIG. FIG. 3A is a diagram showing the state of the exploration wave and the reflected wave when the surface roughness of the road surface is smooth, and FIG. 3B is a graph showing the reflection level of the road surface reflected wave and the obstacle reflected wave. FIG. 4A is a diagram showing the state of the exploration wave and the reflected wave when the surface roughness of the road surface is rough, and FIG. 4B is a graph showing the reflection level of the road surface reflected wave and the obstacle reflected wave. 5A, 5B, and 5C are graphs showing a method of discriminating the type of obstacle by the discriminating unit when the state of the road surface is not taken into consideration. It is a top view which shows an example of a parking space and a reflected wave group. It is a figure which looked at the parking space of FIG. 6 from the X-axis direction. FIG. 3 is a plan view showing another example of a parking space and a reflected wave group. It is a figure which looked at the parking space of FIG. 8 from the X-axis direction. It is a top view which shows an example of a parking space and a reflected wave group. It is a top view which shows an example of a parking space and a reflected wave group. It is a graph which shows the method of discriminating the type of an obstacle by a discriminating part when the condition of a road surface is taken into consideration. It is a flowchart which shows the operation example of the obstacle detection apparatus which concerns on Embodiment 1. It is a figure which shows an example of the hardware composition of the obstacle detection apparatus which concerns on Embodiment 1. FIG. It is a figure which shows another example of the hardware composition of the obstacle detection apparatus which concerns on Embodiment 1. FIG.

Hereinafter, in order to explain the present invention in more detail, embodiments for carrying out the present invention will be described with reference to the accompanying drawings.
Embodiment 1.
FIG. 1 is a block diagram showing a configuration example of the obstacle detection device 1 according to the first embodiment. The vehicle is equipped with an obstacle detection device 1, a distance measuring sensor 2, a transmission / reception unit 3, and an automatic parking device 4. A transmission / reception unit 3 and an automatic parking device 4 are connected to the obstacle detection device 1, and a distance measuring sensor 2 is connected to the transmission / reception unit 3.

The vehicle is provided with, for example, a distance measuring sensor 2 that transmits an exploration wave toward the side of the vehicle. The range-finding sensor 2 is a TOF (Time of Flyght) type sensor that transmits "exploration waves" such as ultrasonic waves, light, or radio waves, and receives "reflected waves" reflected by the exploration waves around the vehicle. ..

The transmission / reception unit 3 outputs a transmission signal to the distance measurement sensor 2, and causes the distance measurement sensor 2 to transmit an exploration wave corresponding to the transmission signal. Further, the transmission / reception unit 3 converts the reflected wave received by the distance measuring sensor 2 into a received signal and outputs the reflected wave to the obstacle detection device 1.

The obstacle detection device 1 includes a detection unit 11, a recognition unit 12, a normalization unit 13, a discrimination unit 14, and an output unit 15.
The obstacle detection device 1 detects obstacles around the vehicle and determines the type of obstacle corresponding to the height of the detected obstacle. Here, it is assumed that there are three types of obstacles: "driving obstacle", "road obstacle", and "road surface obstacle". Among the obstacles, the obstacle having a height high enough to come into contact with the bumper of the vehicle is called a "driving obstacle". The traveling obstacle is a wall or another vehicle parked. Further, among the obstacles, an obstacle having a height low enough not to come into contact with the bumper of the vehicle and having a high height so that the vehicle cannot get over it is called a "road obstacle". Obstacles on the road are curbs or ring clasps. Further, among obstacles, an obstacle having a height low enough not to come into contact with the bumper of the vehicle and having a height low enough to allow the vehicle to get over is called a "road surface obstacle". Road surface obstacles are steps and the like. That is, a traveling obstacle is an obstacle having a height higher than a road obstacle, and a road obstacle is an obstacle having a height higher than a road surface obstacle.

The detection unit 11 compares the magnitude (that is, the reflection level) of the reflected wave received by the ranging sensor 2 with the predetermined threshold value Th1, and determines whether or not the ranging sensor 2 has received the reflected wave. Detect.

FIG. 2 is a graph showing a method of detecting an obstacle by the detection unit 11 of the first embodiment. The horizontal axis of the graph is the propagation distance until the exploration wave transmitted from the ranging sensor 2 is reflected by an obstacle around the vehicle and received by the ranging sensor 2, and the vertical axis is output from the transmission / reception unit 3. The magnitude of the received signal, that is, the reflection level. When the reflection level of the reflected wave exceeds a predetermined threshold value Th1, the detection unit 11 detects the reflected wave. Further, the detection unit 11 calculates the distance between the distance measuring sensor 2 and the reflection point where the reflected wave is reflected based on the propagation distance when the reflection level exceeds the threshold Th1, and the two-dimensional coordinate position of the reflection point. Is calculated.
In addition, when the distance measuring sensor 2 transmits the exploration wave once, the detection unit 11 sets the reflected wave of the reflection level equal to or higher than the threshold Th1 detected first as the "first reflected wave", and then detects the threshold Th1. The reflected wave of the above reflection level is referred to as a "second reflected wave".

Further, the detection unit 11 detects the width of the portion where the reflection level of the reflected wave exceeds the threshold value Th1 as the “wave width”. Alternatively, the detection unit 11 may detect the waveform area of the portion where the reflection level of the reflected wave exceeds the threshold value Th1 as the “area”. Alternatively, the detection unit 11 may detect the maximum value of the reflection level of the reflected wave as the “peak value”. Alternatively, the detection unit 11 may detect as the "response time" from the time when the reflection level of the reflected wave reaches the maximum value to the time when it falls below the threshold value Th1. The wave width, area, crest value, and response time are values that correlate with the magnitude of the reflected wave. The value that correlates with the magnitude of the reflected wave is not limited to the above-mentioned wave width and the like. The detection unit 11 may detect the slope of the falling edge in the waveform of the reflected wave, the time constant in the waveform of the reflected wave, or the like as a value that correlates with the magnitude of the reflected wave.

The detection unit 11 outputs a value correlating with the magnitude of the detected reflected wave and information on the position of the reflection point to the recognition unit 12. The position of the reflection point is information expressed in two-dimensional coordinates.

FIG. 3A is a diagram showing the state of the exploration wave 52 and the reflected wave 53 when the surface roughness of the road surface 51 is smooth, and the arrow 211 shows the main propagation path of the exploration wave. FIG. 3B is a graph showing the reflection level of the road surface reflected wave 54. FIG. 4A is a diagram showing the state of the exploration wave 62 and the reflected wave 63 when the surface roughness of the road surface 61 is rough, and the arrow 212 shows the main propagation path of the exploration wave. FIG. 4B is a graph showing the reflection level of the road surface reflected wave 64. In FIGS. 3B and 4B, the vertical axis of the graph is the magnitude of the reflected wave (that is, the reflection level), and the horizontal axis is the distance measured as the reflected wave by reflecting the exploration wave transmitted from the distance measuring sensor 2 by an obstacle. It is the propagation distance until it is received by the sensor 2. When the surface roughness of the road surface 51 is smooth and flat like a concrete road surface (FIG. 3A), the reflection level of the reflected wave 53 reflected by the road surface 51, that is, the road surface reflected wave 54 is small (FIG. 3B). On the other hand, when the surface roughness of the road surface 61 is as rough as an asphalt road surface (FIG. 4A), the reflected level of the reflected wave 63 reflected by the road surface 61, that is, the road surface reflected wave 64 is large (FIG. 4B).

As shown by the arrow 211 in FIG. 3A, a part of the exploration wave 52 is specularly reflected by the smooth road surface 51 and then reflected by the obstacle 55 to return to the distance measuring sensor 2. As shown by the arrow 212 in FIG. 4A, a part of the exploration wave 62 is diffusely reflected by the rough road surface 61, and then a part of the diffusely reflected reflected wave is reflected by the obstacle 65 and returns to the distance measuring sensor 2. .. When the road surface 61 is rough as shown in FIG. 4A, the exploration wave 62 is diffused on the road surface 61, so that the reflected wave 63 reflected by the obstacle 65, that is, the reflection level of the obstacle reflected wave 66 is the reflection reflected by the obstacle 55. It is significantly lower than the reflection level of the wave 53, that is, the obstacle reflected wave 56. As described above, the reflection level of the reflected waves 53 and 63 reflected by the obstacles 55 and 65 at the same height changes depending on the surface roughness (that is, disturbance) of the road surfaces 51 and 61. Therefore, when the obstacle detection device 1 determines the type of obstacle using a value that correlates with the magnitude of the reflected wave, it is vulnerable to disturbance and it is difficult to accurately determine the type of obstacle. Therefore, in the first embodiment, the obstacle detection device 1 that is resistant to disturbance and can discriminate the type of obstacle with high accuracy is realized by considering the state of the road surface.

Here, a method of discriminating the type of obstacle by the discriminating unit 14 when the state of the road surface is not taken into consideration will be described.
When the road surface condition is not taken into consideration, the discriminating unit 14 calculates two feature quantities using a value correlating with the magnitude of the reflected wave, and on the feature quantity space having the calculated two feature quantities on two axes. , The type of obstacle is determined based on the predetermined first threshold value and second threshold value. As described above, the values that correlate with the magnitude of the reflected wave are the peak value, the wave width, the area, the response time, and the like. The feature amount is a statistical value of a value correlating with the magnitude of a plurality of reflected waves obtained by a plurality of explorations by the distance measuring sensor 2, and is an average value, a dispersion value, a standard deviation, or the like. The discrimination unit 14 uses, for example, an average value of values correlating with the magnitude of the reflected wave as the first feature amount. Further, the discriminating unit 14 uses a standard deviation of a value correlated with the magnitude of the reflected wave as the second feature amount. Here, it is assumed that the tall obstacle has a larger first feature amount and second feature amount than the short obstacle.

5A, 5B, and 5C are graphs showing a method of discriminating the type of obstacle by the discriminating unit 14 when the state of the road surface is not taken into consideration. An obstacle that is the target for the discrimination unit 14 to discriminate the type is referred to as a "target obstacle". The graph of FIG. 5B shows the feature space for the target obstacle on the reference road surface. The graph of FIG. 5A shows the feature space for the target obstacle on the road surface whose surface is rougher than the reference road surface. The graph of FIG. 5C shows the feature space for the target obstacle on the road surface whose surface is smoother than the reference road surface. The horizontal axis of the graph is the first feature amount, and the vertical axis is the second feature amount. In the discrimination unit 14, a first threshold value Th11 and a second threshold value Th12 corresponding to the combination of the first feature amount and the second feature amount are predetermined. The discriminating unit 14 determines whether the target obstacle is a road surface obstacle, a road obstacle, or a traveling obstacle by the first threshold value Th11 and the second threshold value Th12 larger than the first threshold value Th11. When the feature amount of the target obstacle is included in the range where the first feature amount and the second feature amount are smaller than the first threshold value Th11, the discrimination unit 14 discriminates the target obstacle as a road surface obstacle. Further, when the feature amount of the target obstacle is included in the range where the first feature amount and the second feature amount are equal to or more than the first threshold value Th11 and smaller than the second threshold value Th12, the discriminating unit 14 sets the target obstacle on the road. Determine as a thing. Further, when the feature amount of the target obstacle is included in the range where the first feature amount and the second feature amount are the second threshold value Th12 or more, the discrimination unit 14 discriminates the target obstacle as a traveling obstacle.

The value that correlates with the magnitude of the reflected wave and the first feature amount and the second feature amount calculated from the value that correlates with the magnitude of the reflected wave depend on the state of the road surface. Therefore, when the discriminating unit 14 discriminates the type of the target obstacle existing on the road surface whose surface is rougher than the reference road surface by using the first threshold value Th11 and the second threshold value Th12 tuned on the reference road surface, the reflected wave The values that correlate with the size and the first feature amount and the second feature amount become smaller than expected, and as shown in FIG. 5A, a tall target obstacle is erroneously determined to be a short target obstacle. On the contrary, when the discriminating unit 14 discriminates the type of the target obstacle existing on the road surface whose surface is smoother than the reference road surface by using the first threshold value Th11 and the second threshold value Th12 tuned on the reference road surface, the reflection The value that correlates with the magnitude of the wave and the first feature amount and the second feature amount become larger than expected, and as shown in FIG. 5C, a short object is erroneously determined to be a tall object.

Therefore, in the obstacle detection device 1 of the first embodiment, the normalization unit 13 determines the reflected wave of a known object (that is, a reference obstacle) existing on the road surface having the same surface roughness as the road surface on which the target obstacle exists. By normalizing the value that correlates with the magnitude of the reflected wave of the target obstacle by using the value that correlates with the magnitude, the influence of the road surface in the discrimination unit 14 is canceled. The reference obstacle is, for example, a parked vehicle (that is, a traveling obstacle) or a step (road surface obstacle) in a parking scene.

The recognition unit 12 receives information on the value correlating with the magnitude of the reflected wave and the position of the reflection point from the detection unit 11. The recognition unit 12 groups the reflected waves based on the positions of the reflection points. Based on the grouped reflected waves, the recognition unit 12 recognizes whether or not the reflected waves included in this group are reflected by the reference obstacle. The recognition unit 12 recognizes the type of the reference obstacle based on the position of the reflected wave of the group recognized as the reference obstacle. Further, the recognition unit 12 calculates the reference value Pref based on the value that correlates with the magnitude of the reflected wave of the group recognized as the reference obstacle. The recognition unit 12 outputs the information of the type of the reference obstacle, the reference value Ref, and the flag indicating that the reference value Pref has been calculated to the normalization unit 13.

FIG. 6 is a plan view showing an example of the parking space 22 and the reflected wave groups G1 and G2. FIG. 7 is a view of the parking space 22 of FIG. 6 as viewed from the X-axis direction. The example shown in FIGS. 6 and 7 is a scene in which the own vehicle 20 provided with the distance measuring sensor 2 is trying to parallel park in the parking space 22 between the parked vehicle 23 and the parked vehicle 24. Here, the traveling direction of the own vehicle 20 is the X-axis direction, and the direction orthogonal to the X-axis direction is the Y-axis direction. The back side of the parking space 22 is partitioned by a curb 21 which is an obstacle on the road. In this example, the obstacle detection device 1 uses the parked vehicle 23 as a reference obstacle to determine the type of the target obstacle on the back side of the parking space 22. In the examples of FIGS. 6 and 7, the target obstacle is the curb 21 which is a road obstacle, but the type of the target obstacle is not limited to this.

While the own vehicle 20 that has started the automatic parking mode is traveling in the direction indicated by the arrow, for example, the distance measuring sensor 2 transmits the exploration wave 11 times and receives the reflected wave reflected in the measuring range 2a. The detection unit 11 calculates the positions of the reflection points of the 11 first reflected waves indicated by white circles (◯) using the results of 11 searches. The detection unit 11 outputs to the recognition unit 12 the values correlating with the magnitude of each reflected wave obtained in the 11th exploration and the positions of the 11 reflection points. When the distance in the X-axis direction and the distance in the Y-axis direction between two adjacent reflection points are equal to or less than a predetermined distance, the recognition unit 12 determines that the two reflection points are the same reflected wave group. In this way, the recognition unit 12 groups the above 11 reflection points into the reflected wave group G1 and the reflected wave group G2. Then, the recognition unit 12 recognizes the reflected wave group G2 of the reflected wave groups G1 and G2 closer to the own vehicle 20 as the reference reflected wave group, and the length of the reflected wave group G2 in the X-axis direction is increased. If it is within a predetermined range, the type of the reference obstacle of the reflected wave group G2 is recognized as the parked vehicle 23. This predetermined range corresponds to the length of the side surface of the vehicle. When the recognition unit 12 recognizes that the reference obstacle is the parked vehicle 23, the recognition unit 12 calculates an average value of values (for example, peak values) that correlate with the magnitudes of the six reflected waves included in the reflected wave group G2. , The calculated average value is used as the reference value Pref. Further, when the reference value Pref can be calculated, the recognition unit 12 sets a flag indicating that the calculation can be performed.

In the above example, the recognition unit 12 recognizes whether or not the reference obstacle is the parked vehicle 23 by comparing the length of the reflected wave group G2 of the reference obstacle with the predetermined range. , The recognition method is not limited to this method.
For example, the recognition unit 12 may recognize whether or not the reference obstacle is the parked vehicle 23 based on the number of reflected waves obtained in one search by the distance measuring sensor 2. When the distance measuring sensor 2 transmits the exploration wave once and the detection unit 11 detects only the first reflected wave, the recognition unit 12 has the reflection of the reference obstacle corresponding to the reflection point where the first reflected wave is reflected. It recognizes that it is a tall obstacle that cannot explore an object deeper than the point, that is, a parked vehicle 23. On the other hand, when the distance measuring sensor 2 transmits the search wave once and the detection unit 11 detects the second reflected wave following the detection of the first reflected wave, the recognition unit 12 detects the reflection point where the first reflected wave is reflected. It is recognized that the reference obstacle corresponding to is not a short obstacle that can search for an object behind the reflection point, that is, a parked vehicle 23.
Further, for example, the recognition unit 12 may recognize whether or not the reference obstacle is the parked vehicle 23 by using a learning device that machine-learns the shape of the reflection point group reflected by the vehicle.

FIG. 8 is a plan view showing another example of the parking space 22 and the reflected wave groups G11 and G12. FIG. 9 is a view of the parking space 22 of FIG. 8 as viewed from the X-axis direction. In the example shown in FIGS. 8 and 9, similarly to the example shown in FIGS. 6 and 7, the own vehicle 20 provided with the distance measuring sensor 2 has a parking space 22 between the parked vehicle 23 and the parked vehicle 24. It is a scene where you are trying to parallel park. However, the parking space 22 shown in FIGS. 8 and 9 has a two-stage curb structure in which the front side is partitioned by a step 30 which is a road surface obstacle and the back side is partitioned by a curb 21 which is a road obstacle. In this example, the obstacle detection device 1 uses the step 30 as a reference obstacle to determine the type of the target obstacle on the back side of the parking space 22. In the examples of FIGS. 8 and 9, the target obstacle is the curb 21 which is a road obstacle, but the type of the target obstacle is not limited to this.

While the own vehicle 20 that has started the automatic parking mode is traveling in the direction indicated by the arrow, for example, the distance measuring sensor 2 transmits the exploration wave 11 times and receives the reflected wave reflected in the measuring range 2a. The detection unit 11 calculates the positions of the reflection points of the 11 first reflected waves indicated by white circles (◯) using the results of 11 searches. The detection unit 11 outputs to the recognition unit 12 the values correlating with the magnitude of each reflected wave obtained in the 11th exploration and the positions of the 11 reflection points. When the distance in the X-axis direction and the distance in the Y-axis direction between two adjacent reflection points are equal to or less than a predetermined distance, the recognition unit 12 determines that the two reflection points are the same reflected wave group. In this way, the recognition unit 12 groups the above 11 reflection points into the reflected wave group G11. Then, when the length of the reflected wave group G11 in the X-axis direction is out of the predetermined range, the recognition unit 12 recognizes the reflected wave group G11 as the reference reflected wave group, and the recognition unit 12 recognizes the reflected wave group G11 as the reference reflected wave group, and the reflected wave group G11. The type of the reference obstacle is recognized as the step 30. This predetermined range corresponds to the length of the side surface of the vehicle. When the recognition unit 12 recognizes that the reference obstacle is the step 30, it calculates an average value of values (for example, peak values) that correlate with the magnitude of the 11 reflected waves included in the reflected wave group G11. The calculated average value is used as the reference value Pref. Further, when the reference value Pref can be calculated, the recognition unit 12 sets a flag indicating that the calculation can be performed.

In the above example, the recognition unit 12 recognizes whether or not the reference obstacle is a step 30 by comparing the length of the reflected wave group G11 of the reference obstacle with a predetermined range. The recognition method is not limited to this method.
For example, the recognition unit 12 may recognize whether or not the reference obstacle is the step 30 based on the number of reflected waves obtained in one search by the distance measuring sensor 2. When the distance measuring sensor 2 transmits the exploration wave once and the detection unit 11 detects the first reflected wave and the second reflected wave, the recognition unit 12 has a reference obstacle corresponding to the reflection point reflected by the first reflected wave. It recognizes that the object is a short obstacle that can search for an object deeper than the reflection point, that is, a step 30.
Further, for example, the recognition unit 12 may recognize whether or not the reference obstacle is the step 30 by using a learning device that machine-learns the shape of the reflection point group reflected by the step.

In the example shown in FIGS. 8 and 9, the obstacle detection device 1 uses the parked vehicle 23 as a reference obstacle, and is a target obstacle on the back side (that is, a road obstacle) that partitions the parking space 22. A certain curb 21) may be discriminated. In this case, the detection unit 11 uses the results of 11 searches to determine the positions of the reflection points of the 11 first reflected waves indicated by white circles (○) and the 7th positions indicated by white triangles (Δ). 2 Calculate the position of the reflection point of the reflected wave. The recognition unit 12 determines grouping with respect to the first reflected wave, and groups the first reflected wave into the reflected wave group G11. Similarly, the recognition unit 12 determines grouping with respect to the second reflected wave, and groups the second reflected wave into the reflected wave group G12. Then, the recognition unit 12 recognizes whether or not the reference obstacle of the reflected wave group G12 of the second reflected wave is the parked vehicle 23 by using the recognition method shown in FIGS. 6 and 7. When the recognition unit 12 recognizes that the reference obstacle is the parked vehicle 23, the recognition unit 12 calculates an average value of values (for example, peak values) that correlate with the magnitudes of the six reflected waves included in the reflected wave group G12. , The calculated average value is used as the reference value Pref. Further, when the reference value Pref can be calculated, the recognition unit 12 sets a flag indicating that the calculation can be performed.

After the recognition of the reference obstacle is completed, the recognition unit 12 receives from the detection unit 11 the value corresponding to the magnitude of the reflected wave and the information on the position of the reflection point in order to determine the type of the target obstacle. The recognition unit 12 groups the reflected waves based on the positions of the reflection points. The recognition unit 12 recognizes whether or not the reflected waves included in this group are reflected by the target obstacle based on the grouped reflected waves. The recognition unit 12 outputs information on a value that correlates with the magnitude of the reflected wave of the group recognized as the target obstacle to the normalization unit 13.

The normalization unit 13 receives from the recognition unit 12 the information of the type of the reference obstacle, the reference value Ref, and the flag indicating that the reference value Pref has been calculated. Further, the normalization unit 13 receives from the detection unit 11 a value that correlates with the magnitude of the reflected wave reflected by the target obstacle. When the normalization unit 13 receives a flag indicating that the reference value Pref has been calculated, the normalization unit 13 sets a value (for example, a peak value) that correlates with the magnitude of the reflected wave reflected by the target obstacle to the reference value Ref (for example). , The average value of the peak value), and the value that correlates with the magnitude of the reflected wave reflected by the target obstacle after normalization, and the information of the above flag are output to the discrimination unit 14. On the other hand, when the normalization unit 13 does not receive the above flag, the normalization unit 13 outputs a value correlating with the magnitude of the reflected wave reflected by the target obstacle to the discrimination unit 14 without normalization.

FIG. 10 is a plan view showing an example of the parking space 22 and the reflected wave group G31. The parking space 22 shown in FIG. 10 is the same as the parking space 22 shown in FIGS. 6 and 7. After the automatic parking device 4 starts the automatic parking mode and the recognition period of the reference obstacle in the recognition unit 12 has elapsed, while the own vehicle 20 is traveling in the direction indicated by the arrow, for example, the distance measuring sensor 2 Transmits the exploration wave six times and receives the reflected wave reflected in the survey area 2a. In this example, the recognition unit 12 groups the reflection points of the six first reflected waves indicated by white circles (◯) into the reflected wave group G31, and recognizes the reflected wave group G31 as the target reflected wave group. do. The normalization unit 13 assumes that the road surface roughness in which the parked vehicle 23, which is the reference object, is present and the road surface roughness in which the target obstacle is present are the same. Then, the normalization unit 13 sets the values W1 to W6 that correlate with the magnitudes of the six reflected waves included in the reflected wave group G31 of the target obstacle, respectively, as reference values calculated from the reflected waves of the reference object. Normalize with Reflect. In the following equation, W1'to W6' are the normalized values of the values W1 to W6 that correlate with the magnitude of the reflected wave.

W1'= W1 / Ref
W2'= W2 / Ref
・ ・ ・ ・
W6'= W6 / Ref

FIG. 11 is a plan view showing an example of the parking space 22 and the reflected wave groups G41 and G42. The parking space 22 shown in FIG. 11 is the same as the parking space 22 shown in FIGS. 8 and 9. After the automatic parking device 4 starts the automatic parking mode and the recognition period of the reference obstacle in the recognition unit 12 has elapsed, while the own vehicle 20 is traveling in the direction indicated by the arrow, for example, the distance measuring sensor 2 Transmits the exploration wave six times and receives the reflected wave reflected in the survey area 2a. In this example, the recognition unit 12 reflects the six first reflected wave reflection points indicated by white circles (◯) in the reflected wave group G41 and the six second reflected waves indicated by white triangles (Δ). It is assumed that the points are grouped into the reflected wave group G42. Then, the recognition unit 12 recognizes the reflected wave group G42 on the side farther from the own vehicle 20 among the reflected wave groups G41 and G42 as the target reflected wave group. The normalization unit 13 assumes that the road surface roughness in which the parked vehicle 23, which is the reference object, is present and the road surface roughness in which the target obstacle is present are the same. Then, the normalization unit 13 normalizes the values that correlate with the magnitudes of the six reflected waves included in the reflected wave group G42 of the target obstacle with the reference value Pref calculated from the reflected waves of the reference object. To become.

Since the reference value Def when the reference obstacle is the tall parked vehicle 23 and the reference value Def when the short curb 21 is the short rim stone 21, the normalized values W1'to W6' are also the reference obstacles. It depends on the type of thing. Therefore, when the normalization unit 13 is premised on using a plurality of types of reference obstacles, the normalization unit 13 corrects each of the normalized values W1'to W6' by using the correction values according to the types of the reference obstacles. do it. It is assumed that the correction value according to the type of the reference obstacle is preset in the normalization unit 13.

The discrimination unit 14 receives information from the normalization unit 13 of values W1'to W6' that correlate with the magnitude of the reflected wave of the target obstacle after normalization and a flag indicating that the reference value Pref can be calculated. .. The discrimination unit 14 calculated two features using the values W1'to W6'that correlate with the magnitude of the reflected wave of the target obstacle after normalization, and had the calculated two features on two axes. On the feature space, the type of the target obstacle is determined based on the predetermined first threshold value Th11 and the second threshold value Th12. The feature amount is an average value, a variance value, a standard deviation, or the like. For example, the discrimination unit 14 calculates the average value of the values W1'to W6' that correlate with the magnitude of the reflected wave of the target obstacle after normalization as the first feature amount, and after normalization as the second feature amount. The standard deviation of the values W1 ′ to W6 ′ that correlate with the magnitude of the reflected wave of the target obstacle is calculated.

FIG. 12 is a graph showing a method of discriminating the type of obstacle by the discriminating unit 14 when the condition of the road surface is taken into consideration. The discrimination unit 14 plots the normalized feature amount 40 of the target obstacle in the normalized feature amount space shown in FIG. Since the feature amount 40 is included in the range of the first threshold value Th11 or more and smaller than the second threshold value Th12, the discrimination unit 14 discriminates the target obstacle as a road obstacle. The output unit 15 outputs information indicating that the target obstacle is a road obstacle to the automatic parking device 4.

The automatic parking device 4 receives information indicating the type of the target obstacle from the output unit 15, and parks the own vehicle 20 in the parking space 22 using this information. At that time, in the automatic parking device 4, is the obstacle high enough that the obstacle existing in the back side of the parking space 22 comes into contact with the bumper of the own vehicle 20 before the own vehicle 20 enters the parking space 22? Since it is possible to determine whether the obstacle is low enough not to come into contact with the bumper, it is possible to optimize the guidance route when controlling the own vehicle 20, and to optimize the parking position. Further, the automatic parking device 4 accurately detects the parking space 22 by using the information of the target obstacle even when the parking space 22 has a two-stage curb structure as shown in FIGS. 8 and 9. be able to.

Next, the operation of the obstacle detection device 1 will be described.
FIG. 13 is a flowchart showing an operation example of the obstacle detection device 1 according to the first embodiment. In step ST1, the transmission / reception unit 3 starts distance measurement by the distance measurement sensor 2.

When the detection unit 11 receives the notification that the automatic parking device 4 has started the automatic parking mode in step ST2 (step ST2 “YES”), the detection unit 11 uses the transmission / reception result received from the transmission / reception unit 3 in step ST3 to reflect the wave. Is detected. The recognition unit 12 recognizes the reflected wave reflected by the reference obstacle among the reflected waves detected by the detection unit 11, and classifies the reflected wave into the reference reflected wave group. On the other hand, when the automatic parking mode is not set (step ST2 “NO”), the operation of the obstacle detection device 1 ends.

In step ST4, the recognition unit 12 calculates a reference value using a value that correlates with the magnitude of a plurality of reflected waves classified into the reference reflected wave group.

In step ST5, the detection unit 11 detects the reflected wave using the transmission / reception result of the exploration wave. The recognition unit 12 recognizes the reflected wave reflected by the target obstacle among the reflected waves detected by the detection unit 11, and classifies the reflected wave into the target reflected wave group.
The recognition unit 12 may perform the search in this step ST5, for example, when a predetermined time has elapsed from the time when the automatic parking mode is started, or when the distance measuring sensor 2 performs a predetermined number of searches. Start operation.

In step ST6, the normalization unit 13 normalizes the values that correlate with the magnitudes of the plurality of reflected waves classified into the target reflected wave group by using the reference values calculated by the recognition unit 12.

In step ST7, the discriminating unit 14 calculates the first feature amount and the second feature amount using the values corresponding to the normalized magnitudes of the plurality of reflected waves classified into the target reflected wave group. The discrimination unit 14 discriminates the type of the target obstacle based on the first threshold value Th11 and the second threshold value Th12 on the feature amount space having the calculated first feature amount and the second feature amount on two axes.

In step ST8, the output unit 15 outputs information indicating the type of the target obstacle determined by the determination unit 14 to the automatic parking device 4.

Next, the hardware configuration of the obstacle detection device 1 according to the first embodiment will be described.
14 and 15 are diagrams showing a hardware configuration example of the obstacle detection device 1 according to the first embodiment. The functions of the detection unit 11, the recognition unit 12, the normalization unit 13, the discrimination unit 14, and the output unit 15 in the obstacle detection device 1 are realized by the processing circuit. That is, the obstacle detection device 1 includes a processing circuit for realizing the above function. The processing circuit may be a processing circuit 100 as dedicated hardware, or may be a processor 101 that executes a program stored in the memory 102.

As shown in FIG. 14, when the processing circuit is dedicated hardware, the processing circuit 100 may be, for example, a single circuit, a composite circuit, a programmed processor, a parallel programmed processor, or an ASIC (Application Specific Integrated Circuit). ), FPGA (Field Processor Metal Gate Array), or a combination thereof. The functions of the detection unit 11, the recognition unit 12, the normalization unit 13, the discrimination unit 14, and the output unit 15 may be realized by a plurality of processing circuits 100, or the functions of each unit may be realized by one processing circuit 100. You may.

As shown in FIG. 15, when the processing circuit is the processor 101, the functions of the detection unit 11, the recognition unit 12, the normalization unit 13, the discrimination unit 14, and the output unit 15 are software, firmware, or software and firmware. It is realized by the combination with. The software or firmware is described as a program and stored in the memory 102. The processor 101 realizes the functions of each part by reading and executing the program stored in the memory 102. That is, the obstacle detection device 1 includes a memory 102 for storing a program in which the step shown in the flowchart of FIG. 13 is eventually executed when executed by the processor 101. Further, it can be said that this program causes the computer to execute the procedure or method of the detection unit 11, the recognition unit 12, the normalization unit 13, the discrimination unit 14, and the output unit 15.

Here, the processor 101 is a CPU (Central Processing Unit), a processing device, an arithmetic unit, a microprocessor, or the like.
The memory 102 may be a non-volatile or volatile semiconductor memory such as a RAM (Random Access Memory), a ROM (Read Only Memory), an EPROM (Erasable Programmable ROM), or a flash memory, and may be a non-volatile or volatile semiconductor memory such as a hard disk or a flexible disk. It may be a magnetic disk of the above, or an optical disk such as a CD (Compact Disc) or a DVD (Digital Versaille Disc).

Regarding the functions of the detection unit 11, the recognition unit 12, the normalization unit 13, the discrimination unit 14, and the output unit 15, some of them are realized by dedicated hardware, and some of them are realized by software or firmware. It is also good. As described above, the processing circuit in the obstacle detection device 1 can realize the above-mentioned functions by hardware, software, firmware, or a combination thereof.

As described above, the obstacle detection device 1 according to the first embodiment includes a detection unit 11, a recognition unit 12, a normalization unit 13, a discrimination unit 14, and an output unit 15. The detection unit 11 detects whether or not the distance measuring sensor 2 provided in the own vehicle 20 has received the reflected wave. The recognition unit 12 classifies the reflected wave detected by the detection unit 11 into a reference reflected wave group reflected by the reference obstacle or a target reflected wave group reflected by the target obstacle for which the type is to be determined. The normalization unit 13 is based on the values correlated with the sizes of the plurality of reflected waves classified into the target reflected wave group and the values correlated with the sizes of the plurality of reflected waves classified into the reference reflected wave group. Normalize. The discriminating unit 14 calculates two feature quantities using values correlated with the normalized magnitudes of the plurality of reflected waves classified into the target reflected wave group, and the target is based on the two feature quantities. Determine the type of obstacle. The output unit 15 outputs information indicating the type of the target obstacle discriminated by the discriminating unit 14. Since the discriminating unit 14 discriminates the type of the target obstacle based on the two feature quantities calculated using the values correlated with the magnitude of the reflected wave reflected by the target obstacle, the target obstacle is discriminated. It is possible to provide an obstacle detection device 1 in which the own vehicle 20 does not need to face and approach the target obstacle in order to determine the type. Further, since the normalization unit 13 normalizes the value correlated with the magnitude of the reflected wave of the target reflected wave group with the value correlated with the magnitude of the reflected wave of the reference reflected wave group, it is resistant to disturbance. It is possible to provide an obstacle detection device 1 capable of discriminating the type of a target obstacle with high accuracy.

In the present invention, it is possible to modify any component of the embodiment or omit any component of the embodiment within the scope of the invention.

Since the obstacle detection device according to the present invention is designed to determine the type of obstacle, it is suitable for an obstacle detection device used for an automatic parking device or the like.

1 Obstacle detection device, 2 Distance measurement sensor, 2a range, 3 Transmission / reception unit, 4 Automatic parking device, 11 Detection unit, 12 Recognition unit, 13 Normalization unit, 14 Discrimination unit, 15 Output unit, 20 Own vehicle, 21 Edge stone (target obstacle), 22 parking space, 23 parked vehicle (standard obstacle), 24 parked vehicle, 30 steps (standard obstacle), 40 feature quantity, 51,61 road surface, 52,62 exploration wave, 53,63 Reflected wave, 54,64 Road surface reflected wave, 55,65 Obstacle, 56,66 Obstacle reflected wave, 100 processing circuit, 101 processor, 102 memory, 211,212 arrow, G1, G2, G11, G12, G31, G41 , G42 reflected wave group, Th1 threshold, Th11 first threshold, Th12 second threshold.

The obstacle detection device according to the present invention has a detection unit that detects whether or not a distance measuring sensor provided in the vehicle has received a reflected wave, and a reference obstacle that reflects the reflected wave detected by the detection unit. There is a correlation between the recognition unit that classifies into the reference reflected wave group or the target reflected wave group reflected by the target obstacle for which the type is to be determined, and the size of multiple reflected waves classified into the target reflected wave group. The normalization section that normalizes the values based on the values correlated with the magnitude of the multiple reflected waves classified in the reference reflected wave group, and the normalization of the multiple reflected waves classified in the target reflected wave group. A discriminant unit that calculates two feature quantities using values that correlate with the later size and discriminates the type of the target obstacle based on the two feature quantities, and a discriminant unit that discriminates the target obstacle. It is equipped with an output unit that outputs information indicating the type, and the recognition unit identifies the reference obstacle when the length of the reference reflected wave group is within a predetermined range. Recognized as the side of the vehicle, the target obstacle is the obstacle on the back side that divides the parking space.
Further, in the obstacle detection device according to the present invention, a detection unit that detects whether or not the distance measuring sensor provided in the vehicle has received the reflected wave and the reflected wave detected by the detection unit are used as a reference obstacle. Corresponds to the size of a plurality of reflected waves classified into the target reflected wave group and the recognition unit classified into the reference reflected wave group reflected by the above or the target reflected wave group reflected by the target obstacle for which the type is to be determined. A normalization unit that normalizes a certain value based on a value correlated with the magnitude of multiple reflected waves classified into the reference reflected wave group, and a plurality of reflected waves classified into the target reflected wave group. Two feature quantities are calculated using values correlated with the size after normalization, and a discriminant unit that discriminates the type of the target obstacle based on the two feature quantities and a target fault discriminated by the discriminant unit. It is equipped with an output unit that outputs information indicating the type of object, and when the length of the reference reflected wave group is outside the predetermined range, the recognition unit identifies the reference obstacle and divides the parking space. Recognizing that there is a step, the target obstacle is an obstacle on the back side that divides the parking space.

Claims (3)

A detector that detects whether or not the ranging sensor installed in the vehicle has received the reflected wave,
A recognition unit that classifies the reflected wave detected by the detection unit into a reference reflected wave group reflected by the reference obstacle or a target reflected wave group reflected by the target obstacle for which the type is to be determined.
The values correlated with the magnitudes of the plurality of reflected waves classified into the target reflected wave group are normalized based on the values correlated with the magnitudes of the plurality of reflected waves classified into the reference reflected wave group. Normalization part and
Two feature quantities are calculated using values correlated with the normalized magnitudes of the plurality of reflected waves classified into the target reflected wave group, and the target obstacle is based on the two feature quantities. A discriminator that discriminates the type and
An obstacle detection device including an output unit that outputs information indicating the type of the target obstacle determined by the discrimination unit. The first aspect of claim 1, wherein the reference obstacle is a side surface of a parked vehicle parked in a parking space, and the target obstacle is an obstacle on the back side of the parking space. Obstacle detection device. The first aspect of claim 1, wherein the reference obstacle is a step on the front side that divides the parking space, and the target obstacle is an obstacle on the back side that divides the parking space. Obstacle detection device.

Images