JP7134344B2 - Obstacle detection device - Google Patents

Description

The present invention relates to an obstacle detection device.

Conventionally, a device for detecting objects such as obstacles (hereinafter simply referred to as "objects") around a vehicle by using a TOF (Time of Flight) type distance measuring sensor provided in the vehicle, that is, an obstacle detection device. is being developed. Patent Document 1 discloses a technique for determining whether an object is a person or an object based on the waveform of a portion (hereinafter referred to as an "echo") corresponding to a reflected wave from an object in a signal received by a ranging sensor. is disclosed.

That is, by detecting a substantially plate-shaped echo (hereinafter referred to as a “plate-shaped echo”; see Fig. 2 of Patent Document 1, etc.), it is determined that the object is an object. Further, when a substantially forest-like echo (hereinafter referred to as "forest-like echo"; see Fig. 3 of Patent Document 1, etc.) is detected, it is determined that the object is a person. This is based on the fact that when the object is a person, the surface shape of the object is complicated, so there is a high probability that forest-like echoes will be detected.

U.S. Pat. No. 8,432,770

When the object is a pedestrian, there is a timing at which the plate echo is detected. For example, when the pedestrian's arms are swung so that both arms are positioned right beside the trunk, the pedestrian's outline may resemble the outline of a single pole. Thereby, a plate-like echo is detected. In the conventional determination method, based on such a plate-like echo, an object may be erroneously determined to be a stationary object even though the object is a pedestrian.

Also, when the object is a stationary object, a forest-like echo may be detected. For example, when the object is a plurality of poles and the plurality of poles are arranged close to each other, the probe waves are reflected by the plurality of poles. Thus, forest-like echoes are detected. In the conventional determination method, due to such forest-like echoes, the object is sometimes erroneously determined to be a pedestrian even though the object is a stationary object.

SUMMARY OF THE INVENTION The present invention has been made to solve the problems described above, and an object of the present invention is to improve accuracy in determining whether an object is a pedestrian or not.

The obstacle detection device according to the technology disclosed herein uses a signal received by a ranging sensor provided in a vehicle to detect an echo group within a window having a predetermined time width or distance width as a width corresponding to a pedestrian. a frequency analysis unit for calculating the frequency distribution in the echo group by performing frequency analysis on the echo group; and whether or not the object corresponding to the echo group is a pedestrian based on the frequency distribution. and an obstacle discrimination unit that discriminates whether the object is a pedestrian and outputs the result of the discrimination , the obstacle discrimination unit determines whether the object is a pedestrian based on the number of echoes in the echo group and the spectrum width in the frequency distribution If the condition for determining that the object is a pedestrian is satisfied and the number of echoes is equal to or greater than a predetermined number, the obstacle determination unit determines that the object is a bicycle .

According to the present invention, since it is configured as described above, it is possible to improve the accuracy of determining whether an object is a pedestrian or not.

1 is a block diagram showing main parts of an obstacle detection system including an obstacle detection device according to Embodiment 1; FIG. FIG. 4 is an explanatory diagram showing details of processing executed by a transmission/reception control unit, an echo detection unit, and a frequency analysis unit; FIG. 4 is an explanatory diagram showing an example of an echo group, a synthesized power spectrum, and the like when a search wave is reflected by one pole; FIG. 4 is an explanatory diagram showing an example of an echo group, a synthesized power spectrum, and the like when a search wave is reflected by two poles; FIG. 2 is an explanatory diagram including a side view of a pedestrian, a top view of the pedestrian, and a characteristic diagram showing the moving speed with respect to each position of the left arm, body and right arm of the pedestrian. FIG. 11 is an explanatory diagram showing an example of an echo group, a synthesized power spectrum, and the like when a survey wave is reflected by a pedestrian in a state immediately before or after the third state; FIG. 10 is an explanatory diagram showing an example of an echo group, a synthesized power spectrum, and the like when an investigation wave is reflected by a pedestrian in a second state; FIG. 10 is an explanatory diagram showing an example of a correspondence relationship between the number of echoes and spectrum width when an investigation wave is reflected by a pedestrian; FIG. 3 is an explanatory diagram showing an example of a threshold for determining whether an object is a pedestrian or a stationary object, and an example of a threshold for determining whether an object is a bicycle; FIG. 4 is an explanatory diagram showing an example of a threshold value for determining whether an object is a bicycle; FIG. 4 is a Venn diagram showing the relationship between a pedestrian discrimination condition and a bicycle discrimination condition. FIG. 5 is an explanatory diagram showing the details of processing executed by a first trunk determination unit; FIG. 5 is an explanatory diagram showing an example of a correspondence relationship between Doppler shift amounts and relative velocities; FIG. 5 is an explanatory diagram showing examples of transverse movement, oblique movement, and forward-backward movement; FIG. 4 is an explanatory diagram including characteristic diagrams showing spectral width versus time and characteristic diagrams indicating the number of echoes versus time in each of transverse movement, oblique movement, and back-and-forth movement; FIG. 4 is an explanatory diagram showing an example of a threshold for determining whether an object is a pedestrian or a stationary object, and an example of a threshold for determining the moving direction of the pedestrian; FIG. 5 is an explanatory diagram showing details of processing executed by a collision determination unit; 2 is an explanatory diagram showing a hardware configuration of a control device including the obstacle detection device according to Embodiment 1; FIG. 4 is an explanatory diagram showing another hardware configuration of a control device including the obstacle detection device according to Embodiment 1; FIG. 4 is a flow chart showing the operation of a control device including the obstacle detection device according to Embodiment 1; It is a flowchart which shows the detail of a pedestrian discrimination|determination process. 9 is a flowchart showing details of mode determination processing; 4 is a flowchart showing details of collision determination processing; FIG. 4 is an explanatory diagram showing an example of a threshold value for determining whether an object is a pedestrian or a stationary object; FIG. 4 is an explanatory diagram showing an example of a threshold value for determining whether an object is a bicycle; FIG. 4 is a Venn diagram showing the relationship between a pedestrian discrimination condition and a bicycle discrimination condition. It is a flowchart which shows the detail of another pedestrian discrimination|determination process. FIG. 4 is an explanatory diagram showing an example of a composite power spectrum and the like when the sampling rate or frequency resolution is low; FIG. 4 is an explanatory diagram showing an example of partial widths in a combined power spectrum; FIG. 5 is an explanatory diagram showing an example of an echo-to-echo distance and an example of an echo duration time; FIG. 12 is a block diagram showing the essential parts of an obstacle detection system including an obstacle detection device according to Embodiment 8; FIG. 11 is an explanatory diagram showing the details of processing executed by a second body determination unit; FIG. 11 is a flowchart showing details of another mode determination process; FIG.

DESCRIPTION OF THE PREFERRED EMBODIMENTS Hereinafter, in order to describe 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 essential parts of an obstacle detection system including an obstacle detection device according to Embodiment 1. FIG. An obstacle detection system including an obstacle detection device according to Embodiment 1 will be described with reference to FIG.

The vehicle 1 has N distance measuring sensors 2 (N is an integer equal to or greater than 2). Specifically, for example, four ranging sensors 2 are provided at the front end of the vehicle 1 . These ranging sensors 2 constitute a ranging sensor group 3 . Each distance measuring sensor 2 is composed of a TOF type distance measuring sensor. Each distance measuring sensor 2 is capable of transmitting, for example, ultrasonic waves, radio waves, or light (hereinafter collectively referred to as "searching waves"). Further, each distance measuring sensor 2 can receive the search wave reflected by the object O, that is, the reflected wave.

The vehicle 1 has a control device 4 . The control device 4 is configured by, for example, an ECU (Electronic Control Unit). Vehicle 1 has sensors 5 . The sensors 5 include, for example, wheel speed sensors, GPS (Global Positioning System) receivers, yaw rate sensors, and gyro sensors. The vehicle 1 has a vehicle control device 6 . The vehicle control device 6 is configured by an ECU, for example.

The control device 4 has a transmission control section 21 and a reception control section 22 . A transmission/reception control unit 11 is configured by the transmission control unit 21 and the reception control unit 22 . The control device 4 has an echo detector 12 and a frequency analyzer 13 . The control device 4 has an object discrimination section 31 , a first body discrimination section 32 , a speed calculation section 33 , a position calculation section 34 and a mode discrimination section 35 . The object discrimination section 31 , the first body discrimination section 32 , the speed calculation section 33 , the position calculation section 34 and the mode discrimination section 35 constitute an obstacle discrimination section 14 . The control device 4 has a vehicle information acquisition section 15 , a collision determination section 16 and a warning signal output section 17 .

The echo detection unit 12, the frequency analysis unit 13, and the obstacle discrimination unit 14 constitute a main part of the obstacle detection device 100. FIG. In this way, the essential part of the obstacle detection system 200 is configured.

Next, operations of the transmission/reception control unit 11, the echo detection unit 12, and the frequency analysis unit 13 will be described with reference to FIG.

The transmission control unit 21 sequentially supplies an electrical signal (hereinafter referred to as a “transmission signal”) TS to the N ranging sensors 2, thereby controlling the N ranging sensors 2 to sequentially transmit a search wave (hereinafter referred to as “transmission signal”). (referred to as "probing wave transmission control"). Specifically, for example, when the vehicle 1 is moving forward, the transmission control unit 21 sequentially supplies the transmission signal TS to the four ranging sensors 2 provided at the front end of the vehicle 1, thereby The four ranging sensors 2 are caused to sequentially transmit search waves.

The search wave transmitted by each ranging sensor 2 is, for example, a pulse wave modulated with a predetermined carrier frequency. From the viewpoint of avoiding interference between any two ranging sensors 2 out of the N ranging sensors 2, the transmission control unit 21 uses so-called "time division multiplexing", "frequency division multiplexing" or "code division multiplexing". It is preferable to control the ranging sensor group 3 by multiplexing.

When an electric signal is output by each ranging sensor 2, the reception control unit 22 acquires the output electric signal (hereinafter referred to as "received signal") RS. The reception control unit 22 performs a process of detecting the object O (hereinafter referred to as "object detection process") by comparing the amplitude of the received signal RS with a predetermined threshold value Th1. RS' in the figure indicates a signal corresponding to the result of threshold determination (hereinafter referred to as "determination result signal"). The reception control unit 22 outputs the reception signal RS to the echo detection unit 12 when the object O is detected by the object detection process.

Based on the result of the object detection process, the echo detection unit 12 executes a process (hereinafter referred to as "echo detection process") for detecting a portion of the received signal RS corresponding to the reflected wave from the object O, that is, the echo E. . More specifically, the echo detector 12 detects one echo E when one echo E exists within a time window W_T having a predetermined width. On the other hand, when multiple echoes E exist within the time window W_T, the echo detector 12 detects the multiple echoes E. FIG. A group EG composed of one or more detected echoes E is hereinafter referred to as an "echo group". The echo detector 12 calculates the number NE of echoes E included in the echo group EG (hereinafter referred to as "echo number").

Here, the time window W_T corresponds to the distance window W_D having a predetermined width. Hereinafter, the time window W_T and the distance window W_D are collectively referred to as "windows". The width of the window W is a reference value for the height of a general pedestrian (hereinafter referred to as "reference height"), a reference value for the arm swing width of a general pedestrian, or a reference value for the stride length of a general pedestrian. It is set in advance based on at least one of the values (hereinafter referred to as "reference stride length"). For example, when the reference height is 180 cm and the reference stride length is 80 cm, the width of the distance window W_D is set to 100 cm. This value includes the so-called "margin". Thus, the window W has a width corresponding to the pedestrian.

The echo group EG contains one frequency component FC or a plurality of frequency components FC. The frequency analysis unit 13 performs frequency analysis on the echo group EG to obtain one power spectrum PS1 corresponding to one frequency component FC, or a plurality of power spectra PS1 corresponding to a plurality of frequency components FC. to calculate The frequency analysis unit 13 adds up the one or more calculated power spectra PS1 to calculate one power spectrum (hereinafter sometimes referred to as "combined power spectrum") PS2. Hereinafter, these power spectra PS1 and PS2 may be collectively referred to as "frequency distribution".

The frequency analysis unit 13 calculates a width (hereinafter referred to as “spectrum width”) SW of a portion where the intensity in the synthesized power spectrum PS2 exceeds a predetermined threshold Th2. Specifically, for example, the frequency analysis unit 13 calculates the difference between the value of the lowest frequency whose intensity exceeds the threshold Th2 and the value of the highest frequency whose intensity exceeds the threshold Th2. Calculate SW.

When the object O is detected by the object detection process, the reception control unit 22 calculates the distance D is calculated (hereinafter referred to as "ranging process"). The distance D corresponds to the distance between the vehicle 1 and the object corresponding to the individual echo E (ie object O or part of object O).

D=(PV×PT)/2 (1)

Here, PV indicates the propagation velocity of the probe wave in air. The value of the propagation velocity PV is pre-stored in the reception control unit 22, for example. PT indicates the round-trip propagation time of the probe wave. Therefore, PV×PT corresponds to the round-trip propagation distance PD of the probe wave.

In the example shown in FIG. 2, two echoes E_1, E_2 are detected. The echo group EG includes two frequency components FC_1 and FC_2 corresponding to two echoes E_1 and E_2. Therefore, two power spectra PS1_1 and PS1_2 corresponding to the two frequency components FC_1 and FC_2 are calculated. Two distances D_1 and D_2 (not shown) corresponding to the two echoes E_1 and E_2 are calculated based on the two round trip times PT_1 and PT_2 corresponding to the two echoes E_1 and E_2. Here, PV×PT_1 corresponds to the round-trip propagation distance PD_1. Also, PV×PT_2 corresponds to the round-trip propagation distance PD_2.

Next, the operation of the object discrimination section 31 will be described with reference to FIGS. 3 to 11. FIG.

FIG. 3 shows an example of an echo group EG and a synthesized power spectrum PS2 etc. when the object O is a single pole. In this case, as shown in FIG. 3, an echo group EG including one echo E is detected. That is, plate echoes are detected. The echo group EG includes one frequency component FC corresponding to one echo E. FIG. Here, a so-called "Doppler shift" occurs according to the running speed VV of the vehicle 1. FIG. ΔF in the figure indicates the Doppler shift amount of one echo E with respect to the reference frequency F_ref. The value of the reference frequency F_ref is, for example, a value corresponding to the frequency of the transmission signal TS.

The Doppler shift amount ΔF varies according to the relative velocity RV of the object O with respect to the vehicle 1. Normally, when the object O is a stationary object, the relative speed RV is equivalent to the running speed VV of the vehicle 1.

FIG. 4 shows an example of an echo group EG and a synthesized power spectrum PS2 etc. when the object O is two poles (O_1 and O_2 in the figure) and the two poles are arranged close to each other. is shown. In this case, as shown in FIG. 4, an echo group EG including two echoes E_1 and E_2 is detected. That is, a forest echo is detected. The echo group EG includes two frequency components FC_1 and FC_2 corresponding to two echoes E_1 and E_2.

Here, the Doppler shift amount ΔF_1 of echo E_1 is equivalent to the Doppler shift amount ΔF_2 of echo E_2. This is because the relative velocity RV_1 of one pole (O_1 in the figure) with respect to the vehicle 1 is equal to the relative velocity RV_2 of another pole (O_2 in the figure) with respect to the vehicle 1 . This is because each of the relative speeds RV_1 and RV_2 is equivalent to the traveling speed VV. As a result, the value of frequency component FC_1 is equal to the value of frequency component FC_2. As a result, the spectral width SW in the example shown in FIG. 4 is equivalent to the spectral width SW in the example shown in FIG.

On the other hand, when the object O is a pedestrian, it is as follows.

Pedestrian shapes are usually more complex than pole shapes. Therefore, when the pedestrian is irradiated with the exploration wave, one or more of the plurality of parts of the pedestrian (for example, the head, the body, the right arm, the left arm, the right leg, and the left leg) emits the probe wave. reflected. For example, the mounting height of the group of ranging sensors 3 is higher than the waist position of the pedestrian, and the irradiation direction of the search wave by each ranging sensor 2 is parallel or substantially parallel to the road surface (hereinafter referred to as "road surface parallel direction”), the probe wave is mainly reflected by at least one of the body, right arm, and left arm. Alternatively, for example, when the mounting height of the range-finding sensor group 3 is lower than the waist position of the pedestrian, and the irradiation direction of the search wave by each range-finding sensor 2 is set parallel to the road surface, the main In addition, the probe wave is reflected by at least one of the body, right leg and left leg. An example in which the probe wave is reflected by at least one of the torso, right arm, and left arm will be mainly described below.

As shown in FIG. 5, the pedestrian is in a state in which the right arm is raised forward by arm swing and the left leg is stepped forward accordingly (hereinafter referred to as a "first state"); A state in which the left arm is raised forward by the arm swing and the right leg is stepped forward accordingly (hereinafter referred to as the "third state"), and an intermediate state between these states, that is, both states. The state in which the arm is placed right beside the body (hereinafter referred to as the "second state") is repeated.

A characteristic line I_1 in FIG. 5 indicates an example of the moving speed MV_1 of the left arm with respect to the position of the left arm when the pedestrian transits in the order of the first state, the second state, and the third state. A characteristic line I_2 indicates an example of the body movement speed MV_2 with respect to the body position at this time. A characteristic line I_3 indicates an example of the moving speed MV_3 of the right arm with respect to the position of the right arm at this time.

Normally, the absolute value of the swing speed of each arm due to arm swing is maximum in the second state and minimum in each of the first and third states. Therefore, as shown in FIG. 5, each of the movement speed MV_1 of the left arm and the movement speed MV_3 of the right arm periodically fluctuates with respect to the position. These moving velocities MV_1 and MV_3 are in an opposite phase relationship to each other. On the other hand, the moving speed MV_2 of the trunk is substantially constant.

Therefore, the absolute value of the difference value ΔV_1 between the movement speed MV_2 of the body and the movement speed MV_1 of the left arm is maximum in the second state and minimum in each of the first and third states. . The absolute value of the difference value ΔV_2 between the body moving speed MV_2 and the right arm moving speed MV_3 is also maximum in the second state and minimum in each of the first and third states. Also, the absolute value of the difference value ΔV_3 between the moving speed MV_1 of the left arm and the moving speed MV_3 of the right arm has a maximum value in the second state and has a minimum value in each of the first state and the third state.

FIG. 6 shows an example of an echo group EG and a synthesized power spectrum PS2, etc., when the object O is a pedestrian (O_C in the drawing) immediately before or after the third state. When the pedestrian is in the third state, the round-trip propagation distance PD_1 (not shown) between the vehicle 1 and the left arm and the round-trip propagation distance PD_2 between the vehicle 1 and the body are compared to when the pedestrian is in the second state. (not shown) and the round-trip propagation distance PD_3 (not shown) between the vehicle 1 and the right arm. As a result, three echoes E_1, E_2, E_3 are detected as shown in FIG. That is, a forest echo is detected. More specifically, an echo E_1 corresponding to the left arm, an echo E_2 corresponding to the torso, and an echo E_3 corresponding to the right arm are detected.

Here, as described with reference to FIG. 5, the movement speed MV_1 of the left arm, the movement speed MV_2 of the trunk, and the movement speed MV_3 of the right arm can take different values depending on the swing speed of each arm due to arm swing. Accordingly, the relative velocity RV_1 of the left arm with respect to the vehicle 1, the relative velocity RV_2 of the torso with respect to the vehicle 1, and the relative velocity RV_3 of the right arm with respect to the vehicle 1 can also have different values.

As a result, the Doppler shift amount ΔF_1 of echo E_1, the Doppler shift amount ΔF_2 of echo E_2, and the Doppler shift amount ΔF_3 of echo E_3 are different values. That is, the frequency component FC_1 corresponding to echo E_1, the frequency component FC_2 corresponding to echo E_2, and the frequency component FC_3 corresponding to echo E_3 have different values. Therefore, the spectral width SW in the example shown in FIG. 6 is greater than the spectral width SW in the examples shown in FIGS.

FIG. 7 shows an example of an echo group EG and a synthesized power spectrum PS2 when the object O is a pedestrian in the second state (O_B in the drawing). When the pedestrian is in the second state, compared to when the pedestrian is in the first state or the third state, the distance between any two distances among the round-trip propagation distances PD_1, PD_2, PD_3 (not shown) Difference value is small. As a result, one echo E is detected as shown in FIG. That is, plate echoes are detected.

However, as described above, the left arm relative velocity RV_1 to the vehicle 1, the torso relative velocity RV_2 to the vehicle 1, and the right arm relative velocity RV_3 to the vehicle 1 have different values. Therefore, as shown in FIG. 7, the echo group EG contains three frequency components FC_1, FC_2, and FC_3. That is, the echo group EG includes a frequency component FC_1 corresponding to the left arm, a frequency component FC_2 corresponding to the trunk, and a frequency component FC_3 corresponding to the right arm. Therefore, the spectral width SW in the example shown in FIG. 7 is greater than the spectral width SW in the examples shown in FIGS.

In particular, as described with reference to FIG. 5, when the pedestrian is in the second state, each of the difference values ΔV_1, ΔV_2, and ΔV_3 is higher than when the pedestrian is in the first state or the third state. is large. Therefore, when the pedestrian is in the second state, the difference value between any two of the frequency components FC_1, FC_2, and FC_3 is higher than when the pedestrian is in the first state or the third state. is large. As a result, the spectral width SW in the example shown in FIG. 7 is even larger than the spectral width SW in the example shown in FIG.

That is, when the object O is a pedestrian, when the pedestrian is in the first state or the third state, the number of echoes NE increases and the spectrum The width SW becomes smaller. On the other hand, when the object O is a pedestrian, when the pedestrian is in the second state, the echo number NE decreases and the spectrum Width SW increases. Therefore, when the object O is a pedestrian, the echo number NE and the spectral width SW are inversely correlated with each other. A characteristic line II in FIG. 8 shows an example of the correspondence relationship between the number of echoes NE and the spectrum width SW when the object O is a pedestrian.

Based on the above contents, the threshold value Th3_1 for determining whether the object O is a pedestrian or a stationary object is set as follows before the vehicle 1 is shipped at the latest.

That is, for example, measured values of the echo number NE and the spectrum width SW are collected when the object O is one pole. Also, measured values of the number of echoes NE and the spectrum width SW are collected when the object O is two poles. Also, measured values of the echo number NE and the spectrum width SW are collected when the object O is a pedestrian in the first state or the third state. Also, measured values of the echo number NE and the spectral width SW are collected when the object O is a pedestrian in the second state.

These values are plotted in a coordinate system CS1 having a first axis corresponding to echo number NE and a second axis corresponding to spectral width SW. By clustering the plotted point cloud, an area A1_1 corresponding to one pole, an area A1_2 corresponding to two poles, an area A1_3 corresponding to the pedestrian in the second state, and the first state or the third state area A1_4 corresponding to the pedestrian is set (see FIG. 9).

Next, based on the areas A1_1 and A_2 corresponding to the stationary object and the areas A1_3 and A1_4 corresponding to the pedestrian, a curve corresponding to the threshold Th3_1 (hereinafter referred to as "discrimination curve") is set (see FIG. 9). A so-called "machine learning" technique may be used for setting the areas A1_1, A1_2, A1_3, and A1_4 and for setting the discrimination curve corresponding to the threshold value Th3_1.

As a result, the threshold Th3_1 will be set by the time the vehicle 1 is shipped at the latest. With the threshold Th3_1 set, the object determination unit 31 uses the set threshold Th3_1 to determine whether the object O is a pedestrian or a stationary object as follows.

That is, the object discrimination section 31 acquires the value of the number of echoes NE calculated by the echo detection section 12 and acquires the value of the spectrum width SW calculated by the frequency analysis section 13 . The object discrimination unit 31 compares the acquired value of the spectrum width SW with a threshold Th3_1 corresponding to the acquired value of the number of echoes NE. The object determination unit 31 determines that the object O is a pedestrian when the value of the spectrum width SW is equal to or greater than the threshold Th3_1 (see FIG. 9). On the other hand, when the value of the spectral width SW is less than the threshold Th3_1, the object discriminating section 31 discriminates that the object O is a stationary object (see FIG. 9).

By the determination using the threshold value Th3_1, for example, when the object O is two poles (see FIG. 4), it can be correctly determined that the object O is a stationary object even though forest-like echoes are detected. can be done. Also, for example, when the object O is a pedestrian in the second state (see FIG. 7), it can be correctly determined that the object O is a pedestrian even though the plate echo is detected. As a result, it is possible to improve the accuracy of determining whether the object O is a pedestrian or not compared with the conventional determination method that does not use the frequency distribution.

Note that the threshold Th3_2 for determining whether the object O is a bicycle may be set as described below before the vehicle 1 is shipped at the latest. The threshold Th3_2 is the value of the echo number NE when the value of the spectrum width SW is equal to or greater than the threshold Th3_1 corresponding to the value of the echo number NE, that is, when the condition for determining that the object O is a pedestrian is satisfied. is compared with

That is, for example, measured values of the echo number NE and the spectral width SW are collected when the object O is a bicycle. These values are plotted in the coordinate system CS1. By clustering the plotted point cloud, an area A1_5 corresponding to the bicycle is set (see FIG. 9).

Next, a discrimination curve corresponding to the threshold Th3_2 is set based on the areas A1_3 and A1_4 corresponding to pedestrians and the area A1_5 corresponding to bicycles (see FIG. 9). A machine learning technique may be used to set the area A1_5 and to set the discrimination curve corresponding to the threshold value Th3_2.

As a result, the thresholds Th3_1 and Th3_2 are set by the time the vehicle 1 is shipped at the latest. With the thresholds Th3_1 and Th3_2 set, the object determination unit 31 uses the set thresholds Th3_1 and Th3_2 to determine whether the object O is a pedestrian, a bicycle, or a stationary object. Determine whether the

That is, the object discrimination section 31 acquires the value of the number of echoes NE calculated by the echo detection section 12 and acquires the value of the spectrum width SW calculated by the frequency analysis section 13 . The object discrimination unit 31 compares the acquired value of the spectrum width SW with a threshold Th3_1 corresponding to the acquired value of the number of echoes NE. The object determination unit 31 determines that the object O is a pedestrian when the value of the spectrum width SW is equal to or greater than the threshold Th3_1 (see FIG. 9). On the other hand, when the value of the spectrum width SW is less than the threshold Th3_1, the object discriminating section 31 discriminates that the object O is a pedestrian (see FIG. 9).

When the object O is determined to be a pedestrian using the threshold value Th3_1, then the object determination unit 31 sets the value of the acquired echo number NE to a threshold value according to the value of the acquired spectrum width SW. Compare with Th3_2. The object determination unit 31 determines that the object O is a bicycle when the value of the number of echoes NE is equal to or greater than the threshold Th3_2 (see FIG. 9 or 10). On the other hand, when the value of the number of echoes NE is less than the threshold Th3_2, the object discriminating section 31 discriminates that the object O is a pedestrian (see FIG. 9 or 10). Generally, when the object O is a bicycle, compared to when the object O is a pedestrian, the area of the part of the object O that reflects the search wave is larger, so that the echo number NE is It utilizes the fact that it increases.

Thus, in the object discrimination section 31, it can be said that the condition for determining whether the object O is a bicycle is included or included in the condition for determining whether the object O is a pedestrian. FIG. 11 is a Venn diagram showing the relationship between these determination conditions.

Next, with reference to FIG. 12, the operation of the first body determining section 32 will be described.

When the object determination unit 31 determines that the object O is a pedestrian, the first body determination unit 32 performs processing to determine a frequency component FC corresponding to the body out of the plurality of frequency components FC. In this case, when the group of echoes EG includes a plurality of echoes E, the first body discriminating section 32 executes processing for discriminating an echo E corresponding to the body among the plurality of echoes E. do. Hereinafter, these processes are collectively referred to as "first body determination process".

Specifically, for example, when the echo group EG includes a plurality of echoes E, the first torso determination unit 32 calculates the peak value (not shown) of the amplitude of each of the plurality of echoes E. . The first trunk discriminator 32 discriminates the echo E having the maximum peak value among the plurality of echoes E as the echo E corresponding to the trunk. Further, the first body discriminating section 32 discriminates that the frequency component FC corresponding to the discriminated echo E is the frequency component FC corresponding to the body.

Alternatively, for example, when a plurality of echoes E are included in the echo group EG, the first body determination unit 32 calculates the duration (not shown) of each of the plurality of echoes E. The first body discriminating unit 32 determines that the echo E having the longest duration among the plurality of echoes E is the echo E corresponding to the body. Also, the first body discriminating section 32 discriminates that the frequency component FC corresponding to the discriminated echo E is the frequency component FC corresponding to the body.

Alternatively, for example, the first body determination unit 32 calculates the maximum value of intensity (hereinafter referred to as “maximum intensity”) P_max in the combined power spectrum PS2. The first torso determination unit 32 determines that the frequency component FC corresponding to the maximum intensity P_max among the plurality of frequency components FC is the frequency component FC corresponding to the torso. Further, when the echo group EG includes a plurality of echoes E, the first body discriminating section 32 discriminates that the echo E corresponding to the discriminated frequency component FC is the echo E corresponding to the body. (See FIG. 12).

Alternatively, for example, the first body determination unit 32 calculates an average value of frequencies (hereinafter referred to as "average frequency") F_ave in the combined power spectrum PS2. The first torso determination unit 32 determines that the frequency component FC corresponding to the average frequency F_ave among the plurality of frequency components FC is the frequency component FC corresponding to the torso. Further, when the echo group EG includes a plurality of echoes E, the first body discriminating section 32 discriminates that the echo E corresponding to the discriminated frequency component FC is the echo E corresponding to the body. (See FIG. 12). As shown in FIG. 12, the frequency component FC corresponding to the average frequency F_ave can be said to be the frequency component FC located in the center of the composite power spectrum PS2.

In the example shown in FIG. 12, the frequency component FC_2 out of the three frequency components FC_1, FC_2, FC_3 corresponds to the maximum intensity P_max. Also, the frequency component FC_2 among the three frequency components FC_1, FC_2, and FC_3 corresponds to the average frequency F_ave. Therefore, the first body determining unit 32 determines that the frequency component FC_2 among the three frequency components FC_1, FC_2, and FC_3 corresponds to the body. As a result, the first body determining unit 32 determines that the echo E_2 among the three echoes E_1, E_2, and E_3 corresponds to the body.

Next, referring to FIG. 13, the operation of the speed calculator 33 will be described. In addition, the operation of the vehicle information acquisition unit 15 will be explained.

As described above, the Doppler shift amount ΔF of the echo E varies depending on the relative velocity RV of the object O with respect to the vehicle 1. A characteristic line III in FIG. 13 shows an example of the correspondence relationship between the Doppler shift amount ΔF and the relative velocity RV.

Therefore, the velocity calculator 33 calculates the Doppler shift amount ΔF based on the result of frequency analysis by the frequency analyzer 13 . The velocity calculator 33 calculates the relative velocity RV of the object O with respect to the vehicle 1 based on the calculated Doppler shift amount ΔF.

The vehicle information acquisition unit 15 also acquires information indicating the traveling speed VV of the vehicle 1 using the sensors 5 . The speed calculator 33 calculates the moving speed MV of the object O using the traveling speed VV indicated by the acquired information and the calculated relative speed RV by the following equation (2).

MV=VV+RV (2)

Here, when the object determination unit 31 determines that the object O is a pedestrian, the velocity calculation unit 33 calculates the Doppler shift amount of the frequency component FC corresponding to the body based on the determination result of the first body determination unit 32. Calculate ΔF. As a result, when a plurality of echoes E are included in the echo group EG, the Doppler shift amount ΔF of the echoes E corresponding to the trunk is calculated. The velocity calculator 33 calculates the relative velocity RV based on the calculated Doppler shift amount ΔF. Thereby, the relative velocity RV of the trunk with respect to the vehicle 1 is calculated. Also, the moving speed MV of the trunk is calculated by the above equation (2).

Hereinafter, the process of calculating the relative velocity RV by the velocity calculator 33 and the process of calculating the moving velocity MV by the velocity calculator 33 are collectively referred to as "velocity calculation process".

Next, the operation of the position calculator 34 will be described. In addition, the operation of the vehicle information acquisition section 15 will be described.

The vehicle information acquisition unit 15 uses the sensors 5 to acquire information indicating the position coordinates of the vehicle 1 when the search waves are transmitted by the individual ranging sensors 2 . The position calculator 34 also acquires information indicating the distance D calculated by the reception controller 22 . The position calculator 34 calculates the position of the object O using these pieces of information. More specifically, the position calculation unit 34 locates the object O , the position coordinates PC are calculated. Various known calculation methods can be used to calculate the position coordinates PC. Description of these calculation methods is omitted.

Here, when the object O is determined to be a pedestrian by the object determination unit 31 and a plurality of echoes E are included in the echo group EG, the position calculation unit 34 determines that the first body determination unit 32 The position coordinates PC of the torso are calculated based on the result of the determination by . That is, in this case, the distance D corresponding to the left arm, the distance D corresponding to the trunk, and the distance D corresponding to the right arm are calculated by the reception control unit 22 . The position calculation unit 34 calculates the position coordinates PC based on the distance D corresponding to the trunk out of the calculated distances D. FIG.

Hereinafter, the process of calculating the position coordinates PC by the position calculator 34 will be referred to as "position calculation process".

Next, the operation of the mode determination unit 35 will be described with reference to FIGS. 14 to 16. FIG.

The directions in which pedestrians move relative to the vehicle 1 are roughly classified into the following three directions. That is, the moving directions are the left-right direction with respect to the vehicle 1 (hereinafter referred to as the "transverse direction"), the diagonal direction with respect to the vehicle 1 (hereinafter simply referred to as the "oblique direction"), and the front-rear direction with respect to the vehicle 1 (hereinafter simply referred to as the "front-rear direction"). ”). Hereinafter, movement of a pedestrian in the transverse direction is referred to as "transverse movement". Also, movement of a pedestrian in an oblique direction is referred to as "oblique movement". A pedestrian's movement in the transverse direction is referred to as "transverse movement". FIG. 14 shows examples of transverse movement, diagonal movement and back-and-forth movement.

When the pedestrian moves back and forth, the spectrum width SW tends to increase due to the Doppler shift amount ΔF becoming larger than when the pedestrian moves transversely. Further, the search wave is transmitted M times or more (M is an integer of 2 or more), and the echo group EG is detected M times, whereby M echo groups EG are detected, and the echo number NE is M. When M echo numbers NE are calculated by repeating the calculation, the maximum echo number NE in a group composed of these echo numbers NE (hereinafter referred to as "echo number group") tends to increase There is

On the other hand, when the movement mode of the pedestrian is crossing movement, the spectrum width SW tends to be smaller due to the Doppler shift amount ΔF being smaller than when the pedestrian movement mode is back-and-forth movement. . Also, the value of the maximum echo number NE in the echo number group tends to decrease.

In addition, when the pedestrian's movement mode is oblique movement, a tendency between the tendency when the pedestrian's movement mode is forward/backward movement and the tendency when the pedestrian's movement mode is crosswise movement appears. .

As shown in FIG. 15, when the object O is a pedestrian, the spectral width SW varies periodically with time. A characteristic line IV_1 in FIG. 15 shows an example of the spectrum width SW with respect to time when the pedestrian's movement mode is crossing movement. A characteristic line IV_2 shows an example of the spectrum width SW with respect to time when the pedestrian's movement mode is oblique movement. A characteristic line IV_3 shows an example of the spectrum width SW with respect to time when the pedestrian's movement mode is back-and-forth movement.

As shown in FIG. 15, when the pedestrian moves back and forth, the amount of change in spectral width SW with respect to time is the largest. Therefore, the search wave is transmitted M times or more, and the composite power spectrum PS2 is calculated M times, whereby M composite power spectra PS2 are calculated, and the spectrum width SW is calculated M times, whereby M When the spectrum widths SW are calculated, the variance value in the group (hereinafter referred to as "spectrum width group") formed by these spectrum widths SW is the largest. On the other hand, when the movement mode of the pedestrian is crossing movement, the change amount of the spectrum width SW with respect to time is the smallest. Therefore, the variance value in the spectral width group is the smallest.

Further, as shown in FIG. 15, when the object O is a pedestrian, the echo number NE periodically fluctuates with time. A characteristic line V_1 in FIG. 15 shows an example of the number of echoes NE with respect to time when the pedestrian's movement mode is crossing movement. A characteristic line V_2 shows an example of the number of echoes NE with respect to time when the pedestrian moves obliquely. A characteristic line V_3 shows an example of the number of echoes NE with respect to time when the pedestrian's movement mode is back-and-forth movement.

As shown in FIG. 15, when the pedestrian moves back and forth, the amount of change in the number of echoes NE with respect to time is the largest. Therefore, the variance value in the echo number group is the largest. On the other hand, when the pedestrian's movement mode is traverse movement, the amount of change in the number of echoes NE with respect to time is the smallest. Therefore, the variance value in the echo number group is the smallest.

Based on the above contents, a threshold value for determining whether the object O is a pedestrian or a stationary object (hereinafter referred to as "first threshold value") is set as follows before the vehicle 1 is shipped at the latest. Th4_1, a threshold for determining whether the moving direction of the pedestrian is the transverse direction or the diagonal direction (hereinafter referred to as "second threshold") Th4_2, and whether the pedestrian's moving direction is the diagonal direction or the front-back direction. A threshold (hereinafter referred to as “third threshold”) Th4_3 for determining whether or not there is is set.

That is, measured values of the echo number group and the spectral width group are collected when the object O is one pole. Also, measured values of the echo number group and the spectrum width group when the object O is two poles are collected. Also, actual measurements of echo numbers and spectral widths are collected when the object O is a pedestrian moving across. Also, measured values of the echo number group and the spectrum width group are collected when the object O is a pedestrian moving diagonally. Also, measured values of the echo number group and the spectrum width group are collected when the object O is a pedestrian moving back and forth.

Then, statistical processing is performed on each echo number group to calculate a variance value in each echo number group. Also, the variance value in each spectrum width group is calculated by executing statistical processing for each spectrum width group.

These values are plotted in a coordinate system CS3 having a first axis corresponding to the variance of the echo number NE and having a second axis corresponding to the variance of the spectral width SW. By clustering the plotted point cloud, an area A2_1 corresponding to one pole, an area A2_2 corresponding to two poles, an area A2_3 corresponding to a pedestrian moving across, and an area corresponding to a pedestrian moving diagonally and an area A2_5 corresponding to the pedestrian moving back and forth are set (see FIG. 16).

Next, a discrimination curve corresponding to the first threshold Th4_1 is set based on the areas A2_1 and A2_2 corresponding to the stationary object and the area A2_3 corresponding to the pedestrian crossing. Further, a discrimination curve corresponding to the second threshold Th4_2 is set based on the area A2_3 corresponding to the pedestrian moving across and the area A2_4 corresponding to the pedestrian moving diagonally. Further, a discrimination curve corresponding to the third threshold Th4_3 is set based on the area A2_4 corresponding to the pedestrian moving diagonally and the area A2_5 corresponding to the pedestrian moving back and forth (see FIG. 16).

The areas A2_1, A2_2, A2_3, A2_4, and A2_5 are set, the discrimination curve is set corresponding to the first threshold Th4_1, the discrimination curve is set corresponding to the second threshold Th4_2, and the discrimination curve is set corresponding to the third threshold Th4_3. A machine learning technique may be used for the setting.

In the example shown in FIG. 16, the second threshold Th4_2 is set such that the variance of the spectrum width SW gradually decreases as the variance of the number of echoes NE increases. The third threshold Th4_3 is also set such that the variance of the spectrum width SW gradually decreases as the variance of the number of echoes NE increases.

As a result, the first threshold Th4_1, the second threshold Th4_2, and the third threshold Th4_3 are set by the time the vehicle 1 is shipped at the latest. With the first threshold Th4_1, the second threshold Th4_2, and the third threshold Th4_3 set, the mode determination unit 35 uses the set first threshold Th4_1, the second threshold Th4_2, and the third threshold Th4_3, In the following manner, it is determined whether the object O is a pedestrian or a stationary object, and the moving direction of the pedestrian is determined.

That is, when the search wave is transmitted M times or more, the echo group EG is detected M times, and the number of echoes NE is calculated M times, and the number of echoes NE is calculated M times, the mode determination unit 35 , to calculate the variance in the group of echo numbers. That is, the mode determination unit 35 calculates the variance of the M number of echoes NE. Further, when the composite power spectrum PS2 is calculated M times and the spectrum width SW is calculated M times to calculate M spectrum widths SW, the mode determination unit 35 calculates the variance value in the spectrum width group. do. That is, the mode determination unit 35 calculates the variance value of the M spectral widths SW.

The mode determination unit 35 compares the calculated dispersion value of the spectrum width SW with a first threshold Th4_1 corresponding to the calculated dispersion value of the number of echoes NE. Further, the mode determination unit 35 compares the calculated dispersion value of the spectrum width SW with a second threshold Th4_2 corresponding to the calculated dispersion value of the number of echoes NE. Further, the mode determination unit 35 compares the calculated dispersion value of the spectrum width SW with a third threshold Th4_3 corresponding to the calculated dispersion value of the number of echoes NE.

When the variance value of the spectrum width SW is less than the first threshold Th4_1, the mode determination unit 35 determines that the object O is a stationary object (see FIG. 16). When the variance value of the spectrum width SW is greater than or equal to the first threshold Th4_1 and the variance value of the spectrum width SW is less than the second threshold Th4_2, the mode determination unit 35 determines that the object O is a pedestrian and that the object O is walking. It is determined that the moving direction of the person is the transverse direction (see FIG. 16). When the variance value of the spectrum width SW is greater than or equal to the second threshold Th4_2 and the variance value of the spectrum width SW is less than the third threshold Th4_3, the mode determination unit 35 determines that the object O is a pedestrian and that the object O is walking. It is determined that the moving direction of the person is oblique (see FIG. 16).

Thus, the mode of the object O can be determined by using the variance values of M echo numbers NE and M spectrum widths SW. More specifically, it is possible to determine whether the object O is a pedestrian or a stationary object, and to determine the movement direction of the pedestrian.

When the object O is determined to be a bicycle by the object determination unit 31, the mode determination unit 35 determines the movement direction of the bicycle by a determination method similar to the determination method described above (that is, the method of determining the movement direction of the pedestrian). may be used to determine the

Further, the determination result by the mode determination unit 35 usually corresponds to the determination result by the object determination unit 31 . That is, when the object determination unit 31 determines that the object O is a stationary object, the mode determination unit 35 also determines that the object O is a stationary object. Further, when the object determination unit 31 determines that the object O is a pedestrian, the mode determination unit 35 also determines that the object O is a pedestrian.

However, depending on circumstances around the vehicle 1 , exceptionally, the determination result by the mode determination unit 35 may not correspond to the determination result by the object determination unit 31 . That is, even though the object determination unit 31 determines that the object O is a stationary object, the mode determination unit 35 may determine that the object O is a pedestrian. Alternatively, even though the object determination unit 31 determines that the object O is a pedestrian, the mode determination unit 35 may determine that the object O is a stationary object. In such a case, the determination result by the mode determination unit 35 may be preferentially used over the determination result by the object determination unit 31 .

Hereinafter, the processes executed by the echo detector 12, frequency analyzer 13, object discriminator 31, first body discriminator 32, velocity calculator 33, and position calculator 34 are collectively referred to as "pedestrian discrimination processing". Further, in Embodiments 1 to 7, the processing executed by mode determination section 35 is generically referred to as "mode determination processing". Note that when a plurality of objects O are detected by the object detection process, the pedestrian determination process and the mode determination process are performed for each individual object O.

Next, operations of the collision determination section 16 and the warning signal output section 17 will be described with reference to FIG. In addition, the operation of the vehicle information acquisition unit 15, the vehicle control device 6, etc. will be described.

The collision determination unit 16 determines the possibility of the vehicle 1 colliding with each object O based on the result of the obstacle detection process, the result of the distance measurement process, the result of the pedestrian determination process, and the result of the mode determination process. It is. More specifically, the collision determination unit 16 determines the possibility of a collision at each time (hereinafter referred to as "future time") t+n, which is a predetermined time after the current time t, in the following manner.

First, the vehicle information acquisition unit 15 acquires information indicating the traveling direction DT of the vehicle 1 using the sensors 5 . The collision determination unit 16 uses the acquired information to set a collision determination area (hereinafter referred to as a “collision determination area”) A3 of the vehicle 1 . As shown in FIG. 17, the collision determination area A3 is an area in a coordinate system CS4 with the position of the vehicle 1 as a reference. Also, the collision determination area A3 is an area including a margin according to the traveling direction DT.

Next, the collision determination unit 16 estimates an area A4 where each object O exists at the current time t (hereinafter referred to as "existence area") based on the position coordinates PC of each object O calculated by the position calculation process. do. As shown in FIG. 17, the existence area A4 of each object O is an area in the coordinate system CS4. The existence area A4 of each object O includes a circular margin having a predetermined radius (hereinafter referred to as "margin diameter") r.

At this time, the collision determination unit 16 determines whether each object O is a pedestrian, a bicycle, or a stationary object based on the determination result by the object determination unit 31 (or the determination result by the mode determination unit 35). to set the margin radius r in each existing area A4. For example, the collision determination unit 16 makes the margin radius r of the presence area A4 corresponding to the bicycle larger than the margin radius r of the presence area A4 corresponding to the pedestrian. Further, the collision determination unit 16 makes the margin radius r of the presence area A4 corresponding to the pedestrian larger than the margin radius r of the presence area A4 corresponding to the stationary object. This is because bicycles are usually larger than pedestrians, and pedestrians are larger than poles.

Next, the collision determination unit 16 calculates a vector Vec corresponding to each object O based on the determination result by the object determination unit 31, the determination result by the mode determination unit 35, and the relative velocity RV calculated by the velocity calculation unit 33. do. The vector Vec associated with the pedestrian or bicycle is a vector having a direction corresponding to the moving direction determined by the mode determination unit 35 and having a magnitude corresponding to the relative speed RV. A vector Vec associated with a stationary object is a vector having a direction corresponding to the traveling direction DT of the vehicle 1 and having a magnitude corresponding to the running speed VV.

Next, the collision determination unit 16 uses the calculated vector Vec to predict the existence area A5 of each object O in the time from the current time t to the future time t+n (hereinafter referred to as "future time"). As shown in FIG. 17, the existence area A5 of each object O is an area in the coordinate system CS4. Further, the existence area A5 of each object O is an area having a predetermined opening angle θ. The opening angle θ corresponds to the expansion amount of the margin diameter r in future time.

At this time, the collision determination unit 16 determines whether each object O is a pedestrian, a bicycle, or a stationary object based on the determination result by the object determination unit 31 (or the determination result by the mode determination unit 35). to set the opening angle θ in each existing area A5. For example, the collision determination unit 16 increases the opening angle θ of the presence area A5 corresponding to pedestrians compared to the opening angle θ of the presence area A5 corresponding to bicycles. Further, the collision determination unit 16 makes the opening angle θ of the existence area A5 corresponding to the bicycle larger than the opening angle θ of the existence area A5 corresponding to the stationary object.

That is, usually, the uncertainty of the relative movement direction and the relative velocity RV at the future time t+n is highest in the order of pedestrian, bicycle and stationary object. In other words, the uncertainty is lower in the order of stationary objects, bicycles, and pedestrians. Therefore, the collision determination unit 16 sets the opening angle θ in each existing area A5 according to the uncertainty.

Next, the collision determination unit 16 determines whether each object O will reach the collision determination area A3 within a predetermined time based on whether each existing area A5 overlaps with the collision determination area A3. Thereby, the possibility that each object O will collide with the vehicle 1 within a predetermined time is determined.

Here, when it is determined that a plurality of objects O will reach the collision determination area A3 within a predetermined time, the collision determination unit 16 determines whether the plurality of objects O reach the collision determination area A3 first. Identify an object O that For example, when it is determined that a plurality of pedestrians will reach the collision determination area A3 within a predetermined time, the collision determination unit 16 first among the plurality of pedestrians reaches the collision determination area A3. Identify pedestrians.

In the example shown in FIG. 17, existence areas A4_1 to A4_4 at current time t are estimated for four objects O_1 to O_4 (not shown) based on position coordinates PC_1 to PC_4, respectively. Also, vectors Vec_1 to Vec_4 are calculated respectively, and existence areas A5_1 to A5_4 in the future time are predicted respectively.

Here, each of objects O_1 and O_2 is a stationary object (more specifically, one pole), object O_3 is a pedestrian, and object O_4 is a bicycle. Therefore, the margin diameter r_4 in the existing area A4_4 is larger than the margin diameter r_3 in the existing area A4_3. Also, the margin diameter r_3 in the existing area A4_3 is larger than the margin diameter r_1 in the existing area A4_1 and larger than the margin diameter r_2 in the existing area A4_2. Also, the opening angle θ_3 in the existing area A5_3 is larger than the opening angle θ_4 in the existing area A5_4. Further, the opening angle θ_4 in the existing area A5_4 is larger than the opening angle θ_1 in the existing area A5_1 and larger than the opening angle θ_2 in the existing area A5_2.

As the time progresses, the width of each existing area A5_1 to A5_2 expands. As a result, the existing area A5_4 first overlaps the collision determination area A3, and then the existing area A5_3 overlaps the collision determination area A3. FIG. 17 shows the state after the presence area A5_4 overlaps the collision determination area A3 and before the presence area A5_3 overlaps the collision determination area A3. The collision determination unit 16 determines that there is a possibility that the two objects O_3 and O_4 will collide with the vehicle 1 within a predetermined time. Further, the collision determination unit 16 specifies that the object O_4 among the two objects O_3 and O_4 is the one that collides with the vehicle 1 first.

Hereinafter, the processes executed by the collision determination unit 16 are collectively referred to as "collision determination processing".

The warning signal output unit 17 outputs a warning signal (hereinafter referred to as "warning signal") to a display device (not shown), an audio output device (not shown), the vehicle control device 6 or Output to at least one of wireless communication devices (not shown). The display device is provided, for example, on the dashboard of the vehicle 1, and is composed of a liquid crystal display or an organic EL (Electro Luminescence) display. The audio output device is composed of, for example, a speaker. A wireless communication device is composed of, for example, a transmitter and a receiver for wireless communication.

That is, the warning signal output unit 17 outputs a warning signal when the collision determination unit 16 determines that at least one object O may collide with the vehicle 1 within a predetermined period of time. The display device displays a warning image when a warning signal is output by the warning signal output unit 17 . The audio output device outputs a warning sound when the warning signal is output by the warning signal output unit 17 . When the warning signal is output from the warning signal output unit 17, the vehicle control device 6 controls the brakes and torque of the vehicle 1 to reduce collision damage. Alternatively, when the warning signal is output from the warning signal output unit 17, the vehicle control device 6 controls the brake, torque, steering, etc. of the vehicle 1 to perform collision avoidance control. When the warning signal is output by the warning signal output unit 17, the wireless communication device notifies a mobile information terminal (not shown) or a server device (not shown) of that fact.

Here, when it is determined that there is a possibility that a plurality of objects O will collide with the vehicle 1 within a predetermined time, the warning signal is specified as the first collision with the vehicle 1 among the plurality of objects O. It is based on object O. For example, in the example shown in FIG. 17 , the warning signal output unit 17 outputs a warning signal based on the determination result regarding the object O_4 by the collision determination unit 16 .

Note that when various processes by the transmission/reception control unit 11, the echo detection unit 12, the frequency analysis unit 13, and the obstacle determination unit 14 are executed a plurality of times within a predetermined time period related to the collision determination process, the collision determination unit 16, for example, , may adopt a majority decision based on the results of the processes executed a plurality of times. Thereby, the accuracy of the collision determination process can be improved.

Further, the collision determination unit 16 may execute so-called "tracking" for each object O based on the results of the position calculation processing executed multiple times. The collision determination unit 16 may use the tracking result when calculating the vector Vec. Thereby, the calculation accuracy of the vector Vec can be improved. As a result, it is possible to improve the prediction accuracy of the existence area A5.

Next, with reference to FIG. 18, the hardware configuration of main parts of the control device 4 will be described.

As shown in FIG. 18A, the control device 4 has a processor 41 and a memory 42 . Programs for realizing functions of the transmission/reception control unit 11, the echo detection unit 12, the frequency analysis unit 13, the obstacle determination unit 14, the vehicle information acquisition unit 15, the collision determination unit 16, and the warning signal output unit 17 are stored in the memory 42. is stored. By the processor 41 reading out and executing such a program, the transmission/reception control unit 11, the echo detection unit 12, the frequency analysis unit 13, the obstacle determination unit 14, the vehicle information acquisition unit 15, the collision determination unit 16, and the warning signal output unit 17 function is realized.

Alternatively, the control device 4 has a processing circuit 43 as shown in FIG. 18B. In this case, the functions of the transmission/reception control unit 11, the echo detection unit 12, the frequency analysis unit 13, the obstacle determination unit 14, the vehicle information acquisition unit 15, the collision determination unit 16, and the warning signal output unit 17 are realized by the dedicated processing circuit 43. be done.

Alternatively, the control device 4 has a processor 41, a memory 42 and a processing circuit 43 (not shown). In this case, some of the functions of the transmission/reception control unit 11, the echo detection unit 12, the frequency analysis unit 13, the obstacle determination unit 14, the vehicle information acquisition unit 15, the collision determination unit 16, and the warning signal output unit 17 are It is implemented by processor 41 and memory 42 and the remaining functions are implemented by dedicated processing circuitry 43 .

The processor 41 is composed of one or more processors. The individual processors are, for example, CPUs (Central Processing Units), GPUs (Graphics Processing Units), microprocessors, microcontrollers or DSPs (Digital Signal Processors).

The memory 42 is composed of one or more nonvolatile memories. Alternatively, the memory 42 is composed of one or more nonvolatile memories and one or more volatile memories. Each volatile memory uses, for example, RAM (Random Access Memory). Individual non-volatile memories are, for example, ROM (Read Only Memory), flash memory, EPROM (Erasable Programmable Read Only Memory), EEPROM (Electrically Erasable Programmable Read-Only Memory), SSD (Solid Disk) or SSD Drive).

The processing circuit 43 is composed of one or more digital circuits. Alternatively, the processing circuit 43 is composed of one or more digital circuits and one or more analog circuits. That is, the processing circuit 43 is composed of one or a plurality of processing circuits. Individual processing circuits are, for example, ASIC (Application Specific Integrated Circuit), PLD (Programmable Logic Device), FPGA (Field-Programmable Gate Array), SoC (System-on-a-Chip) or system Integrated LSI (Large-Scale) ) is used.

Next, the operation of the control device 4 will be described with reference to the flowchart of FIG.

First, the transmission control unit 21 executes probe wave transmission control (step ST1). Next, the reception control unit 22 executes object detection processing and distance measurement processing (step ST2). The details of the search wave transmission control, object detection processing, and distance measurement processing have already been described with reference to FIG.

If no object O is detected by the current object detection process (step ST3 "NO"), then the obstacle detection device 100 detects data Da related to the object O detected by the previous object detection process. It is determined whether or not it is stored in the storage section (for example, memory 42) of the control device 4 (step ST4). If such data Da is stored (step ST4 "YES"), then the obstacle detection device 100 erases such data Da (step ST5).

On the other hand, if one or more objects O are detected by the object detection processing this time (“YES” in step ST3), subsequent processing is executed for each individual object O. FIG. First, the obstacle detection device 100 determines whether or not the object O detected by the current object detection process is the same as the object O detected by the previous object detection process (step ST6). In other words, the obstacle detection device 100 determines whether or not the object O detected by the previous object detection process is the same as the object O detected by the current object detection process.

If the object O detected by the current object detection process is different from the object O detected by the previous object detection process (step ST6 "NO"), then the obstacle detection device 100 detects the object O detected by the previous object detection process. It is determined whether or not the data Da regarding the object O detected by the object detection process is stored in the storage unit (for example, the memory 42) of the control device 4 (step ST4). If such data Da is stored (step ST4 "YES"), then the obstacle detection device 100 erases such data Da (step ST5).

If it is determined that the object O detected by the current object detection process is the same as the object O detected by the previous object detection process (step ST6 "YES"), then the pedestrian discrimination process is executed. be. Details of the pedestrian determination process will be described later with reference to the flowchart of FIG. 20 .

Next, the obstacle detection device 100 stores the data Da regarding the object O detected by the current object detection processing in the storage section (for example, the memory 42) of the control device 4 (step ST7). That is, the data Da includes data indicating the distance D, data indicating the number of echoes NE, data indicating the spectrum width SW, data indicating whether the object O is a pedestrian, a bicycle, or a stationary object, and data indicating whether the object O is a pedestrian, a bicycle, or a stationary object. It includes data indicating the frequency component FC corresponding to the body of a pedestrian, data indicating the relative velocity RV, data indicating the moving velocity MV, data indicating the position coordinates PC, and the like.

Next, the obstacle detection device 100 determines whether or not the accumulated number of data Da regarding the object O detected by the current object detection process is a predetermined number (for example, M) or more (step ST8). When it is determined that the accumulated number of data Da is less than the predetermined number (step ST8 "NO"), the process of the control device 4 proceeds to step ST1.

If it is determined that the accumulated number of data Da is equal to or greater than the predetermined number ("YES" in step ST8), then mode determination processing and collision determination processing are executed. Details of the mode determination process will be described later with reference to the flowchart of FIG. 21 . Details of the collision determination process will be described later with reference to the flowchart of FIG.

Next, the warning signal output unit 17 outputs a warning signal according to the result of the collision determination process (step ST9). A specific example of when the warning signal is output and a specific example of the output destination of the warning signal have already been described with reference to FIG.

Next, pedestrian determination processing will be described with reference to the flowchart of FIG. 20 .

First, the echo detector 12 executes echo detection processing (step ST11). Next, the echo detector 12 calculates the number of echoes NE (step ST12). Then, the frequency analysis section 13 performs frequency analysis (step ST13). Then, the frequency analysis unit 13 calculates the spectrum width SW (step ST14). The details of the echo detection processing, the frequency analysis, and the method of calculating the spectral width SW have already been described with reference to FIG. 2, and therefore will not be described again.

Next, the object discrimination section 31 discriminates whether the object O detected by the current object detection processing is a pedestrian, a bicycle, or a stationary object (step ST15). The details of the determination method by the object determination unit 31 have already been described with reference to FIGS. 3 to 11, and therefore will not be described again.

When the object determination unit 31 determines that the object O is a pedestrian (“YES” in step ST16), then the first body determination unit 32 executes first body determination processing (step ST17). The details of the first body determination process have already been described with reference to FIG. 12, and therefore will not be described again. If the object determination unit 31 determines that the object O is not a pedestrian ("NO" in step ST16), the process of step ST17 is skipped.

Next, the speed calculator 33 executes speed calculation processing (step ST18). The details of the speed calculation process have already been described with reference to FIG. 13, and therefore will not be described again. Next, the position calculator 34 executes position calculation processing (step ST19). Since the details of the position calculation process have already been explained, a repeated explanation will be omitted.

Next, the details of the mode determination process will be described with reference to the flowchart of FIG. 21 .

First, the mode determination unit 35 acquires data Da regarding the object O detected by the current object detection process (step ST21). That is, the mode determination unit 35 acquires a predetermined number (for example, M) or more data Da stored in the storage unit (for example, the memory 42) of the control device 4. FIG. Next, the mode determination unit 35 uses the data Da acquired in step ST21 to calculate the variance value in the echo number group (step ST22). Also, the mode determination unit 35 uses the data Da acquired in step ST21 to calculate the variance value in the spectrum width group (step ST23).

Next, the mode determination unit 35 determines the mode of the object O using the variance values calculated in steps ST22 and ST23 (step ST24). More specifically, the mode determination unit 35 determines whether the object O is a pedestrian or a stationary object, and determines the movement direction of the pedestrian. The details of the determination method by the mode determination unit 35 have already been described with reference to FIGS. 14 to 16, and therefore will not be described again.

Next, the details of the collision determination process will be described with reference to the flowchart of FIG. 22 .

First, the collision determination section 16 sets a collision determination area A3 for the vehicle 1 (step ST31). Next, the collision determination unit 16 estimates the existence area A4 of the object O at the current time t (step ST32). Next, the collision determination unit 16 calculates the vector Vec and predicts the existence area A5 of the object O in the future time (step ST33).

Next, the collision determination section 16 determines whether or not the object O will reach the collision determination area A3 within a predetermined time based on whether the presence area A5 overlaps with the collision determination area A3 (step ST34). Thereby, the possibility that the object O will collide with the vehicle 1 within a predetermined time is determined.

Next, a modified example of the obstacle detection system 200 will be described.

The installation position of the range-finding sensor group 3 in the vehicle 1 is not limited to the front end portion of the vehicle 1 . For example, four ranging sensors 2 may be provided at the rear end of the vehicle 1 . The transmission control unit 21 may cause the four ranging sensors 2 provided at the rear end of the vehicle 1 to sequentially transmit search waves when the vehicle 1 is moving backward.

Also, the collision determination unit 16 may be provided in the obstacle detection device 100 . That is, the echo detection unit 12, the frequency analysis unit 13, the obstacle determination unit 14, and the collision determination unit 16 may constitute a main part of the obstacle detection device 100. FIG.

As described above, the obstacle detection device 100 uses the received signal RS from the ranging sensor 2 provided in the vehicle 1 to detect a window having a predetermined time width or distance width corresponding to a pedestrian. an echo detection unit 12 for detecting the echo group EG within W; a frequency analysis unit 13 for calculating the frequency distribution in the echo group EG by performing frequency analysis on the echo group EG; and based on the frequency distribution, the echo group EG and an obstacle discrimination unit 14 that discriminates whether or not the object O corresponding to is a pedestrian and outputs the result of the discrimination. Thereby, it can be determined whether or not the object O is a pedestrian. In particular, it is possible to improve the accuracy of determining whether the object O is a pedestrian or not compared to the conventional determination method that does not use the frequency distribution.

More specifically, as shown in FIG. 7, even when plate echoes are detected based on the waveform of the received signal RS (that is, the waveform in the time domain), the frequency distribution (that is, the distribution in the frequency domain) is used. As a result, the object O, which has been erroneously determined to be a stationary object by the conventional determination method, can be correctly determined to be a pedestrian. Further, as shown in FIG. 4, even when forest-like echoes are detected based on the waveform of the received signal RS (that is, the waveform in the time domain), by using the frequency distribution (that is, the distribution in the frequency domain), the conventional The object O, which has been erroneously determined to be a pedestrian by the determination method of (1), can be correctly determined to be a stationary object. In this way, by using the waveform in the time domain and the distribution in the frequency domain in combination, it is possible to achieve discrimination with higher accuracy than the conventional discrimination method that does not use the distribution in the frequency domain.

Further, the obstacle detection device 100 includes a first body determination unit 32 that determines the frequency component FC corresponding to the body of the pedestrian in the frequency distribution, and the Doppler shift amount ΔF of the frequency component FC corresponding to the body of the vehicle 1 and a speed calculation unit 33 for calculating the relative speed RV of the pedestrian with respect to. Thereby, when the object O is a pedestrian, the relative velocity RV of the pedestrian with respect to the vehicle 1 can be calculated accurately.

In addition, the first torso determination unit 32 determines that the frequency component FC having the maximum intensity in the frequency distribution or the frequency component FC located in the center of the frequency distribution is the frequency component FC corresponding to the torso. Thereby, the frequency component FC corresponding to the torso can be determined (see FIG. 12).

Further, the obstacle discriminating section 14 discriminates whether or not the object O is a pedestrian based on the number of echoes NE in the echo group EG and the spectrum width SW in the frequency distribution. Thereby, for example, using the threshold value Th3_1, it is possible to determine whether the object O is a pedestrian or a stationary object (see FIG. 9).

Further, the obstacle discrimination unit 14 discriminates the object O as a bicycle when the number of echoes NE is equal to or greater than a predetermined number when the condition for discriminating the object O as a pedestrian is satisfied. Thereby, for example, using the threshold value Th3_2, it is possible to determine whether the object O is a pedestrian or a bicycle (see FIG. 9 or 10).

Further, when the search wave is transmitted and received multiple times by the distance measuring sensor 2 and the frequency distribution is calculated multiple times by the frequency analysis unit 13, and a plurality of frequency distributions are calculated, statistical processing is performed on the plurality of frequency distributions. and a mode determination unit 35 that determines the mode of the object O by executing For example, it is possible to determine whether the object O is a pedestrian or a stationary object and determine the moving direction of the pedestrian based on the variance values of the M spectral widths SW by statistical processing of the spectrum width group. can.

Further, the mode determination unit 35 determines whether the object O is a pedestrian or a stationary object based on the clustering result of the variance of the number of echoes NE in the echo group EG and the variance of the spectrum width SW in the frequency distribution. Along with this, the moving direction of the pedestrian is discriminated. Thereby, for example, it is possible to determine whether the object O is a pedestrian or a stationary object using the first threshold Th4_1 (see FIG. 16). Further, for example, the movement direction of the pedestrian can be determined using the second threshold Th4_2 and the third threshold Th4_3 (see FIG. 16).

Further, the mode determination unit 35 determines that the object O is a stationary object when the variance value of the spectrum width SW is less than the first threshold value Th4_1. By using the first threshold Th4_1 based on the clustering result, it is possible to determine whether the object O is a pedestrian or a stationary object.

Further, when the variance value of the spectrum width SW is equal to or greater than the first threshold value Th4_1 and the variance value of the spectrum width SW is less than the second threshold value Th4_2, the mode determination unit 35 determines that the object O is a pedestrian, and , the direction of movement is transverse. By using the second threshold Th4_2 based on the result of clustering, the moving direction of the pedestrian can be determined.

Further, when the variance value of the spectrum width SW is equal to or greater than the second threshold value Th4_2 and the variance value of the spectrum width SW is less than the third threshold value Th4_3, the mode determination unit 35 determines that the object O is a pedestrian, and , the direction of movement is oblique. By using the second threshold Th4_2 and the third threshold Th4_3 based on the result of clustering, it is possible to determine the moving direction of the pedestrian.

Further, when the variance value of the spectrum width SW is greater than or equal to the third threshold value Th4_3, the mode determination unit 35 determines that the object O is a pedestrian and that the movement direction is the front-rear direction. By using the third threshold Th4_3 based on the result of clustering, the direction of movement of the pedestrian can be determined.

The second threshold Th4_2 and the third threshold Th4_3 are set to values that gradually decrease as the variance of the echo number NE increases. Thereby, for example, the second threshold Th4_2 and the third threshold Th4_3 shown in FIG. 16 can be set.

Further, when the plurality of objects O are determined to be a plurality of pedestrians by the obstacle determination unit 14, one of the plurality of pedestrians reaches the collision determination area A3 of the vehicle 1 within a predetermined time. A collision determination unit 16 that identifies a pedestrian is provided. As a result, for example, collision damage mitigation control or collision avoidance control can be realized.

Embodiment 2.
Another modification of the obstacle detection system 200 will be described.

The object discrimination unit 31 acquires the value of the number of echoes NE calculated by the echo detection unit 12 and acquires the value of the spectrum width SW calculated by the frequency analysis unit 13 . The object discrimination unit 31 compares the acquired value of the spectrum width SW with a threshold value Th3_1 corresponding to the acquired value of the echo number NE to determine whether the object O is a pedestrian or a stationary object. (see FIG. 23). The threshold Th3_1 shown in FIG. 23 is the same as the threshold Th3_1 shown in FIG.

Here, the object determination unit 31 determines that the moving speed of the object O is When the absolute value of MV is greater than or equal to a predetermined threshold value Th5, it is determined that object O is a bicycle (see FIG. 24). The moving speed MV is calculated by the speed calculator 33 . Also in this case, it can be said that the condition for determining whether the object O is a bicycle is included or included in the condition for determining whether the object O is a pedestrian (see FIG. 25).

Next, pedestrian discrimination processing in the obstacle detection device 100 will be described with reference to the flowchart of FIG.

First, the processing of steps ST41 to ST44 is executed. Since the processing contents of steps ST41 to ST44 are the same as the processing contents of steps ST11 to ST14 shown in FIG. 20, description thereof is omitted.

Next, the object discrimination section 31 discriminates whether the object O is a pedestrian or a stationary object (step ST45). More specifically, the object determination unit 31 uses the threshold value Th3_1 to determine whether the object O is a pedestrian or a stationary object (see FIG. 23).

When it is determined that the object O is a pedestrian using the threshold value Th3_1 (“YES” in step ST46), then the speed calculator 33 executes speed calculation processing (step ST47). Next, the object discrimination section 31 discriminates whether or not the object O is a bicycle (step ST48). More specifically, the object discriminating section 31 discriminates whether or not the object O is a bicycle using the threshold value Th5 (see FIG. 24).

If it is determined that the object O is not a bicycle using the threshold value Th5 (step ST49 "NO"), that is, if it is determined that the object O is a pedestrian using the threshold value Th5, then the first body determination unit 32 executes the first trunk determination process (step ST50). If it is determined that the object O is a bicycle using the threshold value Th5 ("YES" in step ST49), the process of step ST50 is skipped. Next, the position calculator 34 executes position calculation processing (step ST51).

On the other hand, if the object O is determined to be a stationary object using the threshold value Th3_1 (“NO” in step ST46), then the speed calculator 33 executes speed calculation processing (step ST52). Next, the position calculator 34 executes position calculation processing (step ST51).

As described above, when the conditions for determining that the object O is a pedestrian are satisfied and the absolute value of the moving speed MV of the object O is equal to or greater than a predetermined value, the obstacle determination unit 14 Determine that O is a bicycle. Thereby, for example, it is possible to determine whether the object O is a pedestrian or a bicycle using the threshold value Th5 (see FIG. 24).

Embodiment 3.
Another modification of the obstacle detection system 200 will be described.

Even if the frequency analysis unit 13 calculates the number of frequency components FC included in the echo group EG (hereinafter referred to as "the number of frequency components") instead of the echo detection unit 12 calculating the number of echoes NE. good. For example, in the example shown in FIG. 7, the number of echoes NE is one. In contrast, the number of frequency components is three. The obstacle discrimination unit 14 may use the number of frequency components in each process instead of the number of echoes NE.

Embodiment 4.
Another modification of the obstacle detection system 200 will be described.

The frequency analysis unit 13 may calculate the spectrum width of each power spectrum PS1 instead of calculating the spectrum width SW of the combined power spectrum PS2. The obstacle discriminating section 14 may use the spectrum width of the individual power spectrum PS1 for each process instead of the spectrum width SW of the combined power spectrum PS2.

Further, the frequency analysis unit 13 may use the received signal RS within the window W to directly calculate the combined power spectrum PS2 without calculating the power spectrum PS1 corresponding to each echo E.

Embodiment 5.
Another modification of the obstacle detection system 200 will be described.

In the examples shown in FIGS. 6 and 7, three frequency components FC_1, FC_2, and FC_3 are detected as independent components. This is because the sampling rate in the obstacle detection device 100 is sufficiently high, and the frequency resolution in the obstacle detection device 100 is sufficient for the difference value between any two of the frequency components FC_1, FC_2, and FC_3. because it is expensive.

However, in the actual obstacle detection device 100, it is assumed that the sampling rate is low or the frequency resolution is low. In such a case, as shown in FIG. 27, three frequency components FC_1, FC_2, and FC_3 are detected so as to be continuous with each other.

Therefore, the object discrimination section 31 may discriminate whether the object O is a pedestrian or a stationary object by the following discrimination method. That is, when the number of frequency components FC (that is, the number of frequency components) having an intensity equal to or greater than the threshold Th2 in the synthesized power spectrum PS2 (or the individual power spectra PS1) is a predetermined number or more, or when the synthesized power When the width of the frequency component FC having an intensity equal to or greater than the threshold Th2 in the spectrum PS2 (or individual power spectrum PS1) (i.e. spectrum width) is equal to or greater than a predetermined width, it is determined that the object O is a pedestrian. Also good.

By discrimination based on the number of frequency components, when the obstacle detection device 100 has a high sampling rate and high frequency resolution (see FIG. 6 or 7), it can be determined whether the object O is a pedestrian or a stationary object. can be discriminated. Further, as described in Embodiment 1, the first body determining unit 32 can determine the frequency component corresponding to the body among the plurality of mutually independent frequency components FC.

On the other hand, even if the obstacle detection device 100 has a low sampling rate, or even if the obstacle detection device 100 has a low sampling rate (see FIG. 27), the determination based on the spectral width , whether the object O is a pedestrian or a stationary object.

As described above, when the number of frequency components FC having an intensity equal to or greater than the threshold Th2 in the frequency distribution is equal to or greater than a predetermined number, or when the number of frequency components FC having an intensity equal to or greater than the threshold Th2 in the frequency distribution When the width is equal to or larger than the predetermined width, it is determined that the object O is a pedestrian. This makes it possible to determine whether the object O is a pedestrian or a stationary object regardless of whether the sampling rate is high or low and regardless of the frequency resolution. Also, when the sampling rate and frequency resolution are high, discrimination of the frequency component FC corresponding to the torso can be realized.

Embodiment 6.
Another modification of the obstacle detection system 200 will be described.

When there are a plurality of parts having an intensity exceeding the threshold Th2 in the synthesized power spectrum PS2, the frequency analysis unit 13 calculates the width (hereinafter referred to as "partial width") PW of each of these parts. Also good. The frequency analysis unit 13 may calculate the spectrum width SW by totaling the calculated partial widths PW.

For example, in the example shown in FIG. 28, the frequency analysis unit 13 calculates three partial widths PW_1, PW_2, PW_3. The frequency analysis unit 13 calculates the spectrum width SW by totaling these partial widths PW_1, PW_2, PW_3.

Embodiment 7.
Another modification of the obstacle detection system 200 will be described.

Instead of calculating the number of echoes NE, the echo detector 12 calculates the distance between the echo E corresponding to the minimum distance D and the echo E corresponding to the maximum distance D in the echo group EG (hereinafter referred to as "inter-echo It is also possible to calculate the DE. In this case, the obstacle discrimination section 14 may use the inter-echo distance DE in each process instead of the number of echoes NE.

For example, in the example shown in FIG. 29, the inter-echo distance DE in the echo group EG is between the echo E_1 corresponding to the minimum distance D_1 (not shown) and the echo E_3 corresponding to the maximum distance D_3 (not shown). is the distance of The echo detector 12 calculates the inter-echo distance DE.

Alternatively, instead of calculating the number of echoes NE, the echo detector 12 calculates the duration of each echo E in the echo group EG (hereinafter referred to as "echo duration") TE1, and calculates these echo durations. A total value of TE1 (hereinafter referred to as "total echo duration") TE2 may be calculated. In this case, the obstacle discriminating section 14 may use the total echo duration TE2 in each process instead of the number of echoes NE.

For example, in the example shown in FIG. 29, the echo detector 12 calculates three echo duration times TE1_1, TE1_2, TE1_3. The echo detection unit 12 calculates the total echo duration TE2 by totaling these echo durations TE1_1, TE1_2, and TE1_3.

Alternatively, the echo detector 12 may calculate the inter-echo distance DE in addition to calculating the number of echoes NE. In this case, the obstacle discriminating section 14 may use feature amounts based on the number of echoes NE and the distance between echoes DE in each process instead of the number of echoes NE.

Alternatively, the echo detector 12 may calculate the total echo duration TE2 in addition to calculating the number of echoes NE. In this case, the obstacle discriminating section 14 may use a feature amount based on the number of echoes NE and the total echo duration TE2 for each process instead of the number of echoes NE.

Alternatively, the echo detector 12 may calculate the inter-echo distance DE and the total echo duration TE2 instead of calculating the number of echoes NE. In this case, the obstacle discriminating section 14 may use feature amounts based on the inter-echo distance DE and the total echo duration TE2 for each process instead of the number of echoes NE.

Alternatively, the echo detector 12 may calculate the inter-echo distance DE and the total echo duration TE2 in addition to calculating the number of echoes NE. In this case, instead of the echo number NE, the obstacle discriminating section 14 may use feature amounts based on the echo number NE, the echo-to-echo distance DE, and the total echo duration TE2 for each process.

Embodiment 8.
Another modification of the obstacle detection system 200 will be described.

As shown in FIG. 30 , the obstacle detection device 100 may have a second body discrimination section 36 . When the object O is determined to be a pedestrian by the object determination unit 31 (or the mode determination unit 35), the second body determination unit 36 selects the frequency component FC corresponding to the body from among the plurality of frequency components FC. Execute the process to determine. In this case, when the group of echoes EG includes a plurality of echoes E, the second body discriminating section 36 performs processing for discriminating an echo E corresponding to the body among the plurality of echoes E. do. Hereinafter, these processes are collectively referred to as "second body determination process".

Here, the determination method by the second body determination section 36 is different from the determination method by the first body determination section 32 . Further, the determination timing by the second body determination section 36 is different from the determination timing by the first body determination section 32 .

That is, when the second torso determination unit 36 calculates M composite power spectra PS2 by transmitting the search wave M times or more and calculating the composite power spectrum PS2 M times, the composite power spectrum PS2 is calculated. By performing statistical processing on PS2, variations in individual frequency components FC in these combined power spectra PS2 are calculated. The second body discriminating section 36 discriminates that, among the plurality of frequency components FC in each combined power spectrum PS2, the frequency component FC having a smaller variation than the other frequency components FC is the body. More specifically, the second torso determination unit 36 determines that the frequency component FC with the smallest variation among the plurality of frequency components FC in each combined power spectrum PS2 is the frequency component FC corresponding to the torso. . For example, an intensity variance value is used as the index of variation.

For example, when M=3, as shown in FIG. 31, an echo group EG including three echoes E_1, E_2, and E_3 is detected at the first time t1 (this echo group EG has three frequencies components FC_1, FC_2, and FC_3 are included.), and a synthesized power spectrum PS2_1 corresponding to this echo group EG is calculated. Next, at a second time t2, an echo group EG including two echoes E_4 and E_5 is detected (this echo group EG includes two frequency components FC_4 and FC_5), and this echo It is assumed that the synthetic power spectrum PS2_2 corresponding to the group EG has been calculated. Next, at the third time t3, an echo group EG including two echoes E_6 and E_7 is detected (this echo group EG includes two frequency components FC_6 and FC_7), and this echo It is assumed that the synthetic power spectrum PS2_3 corresponding to the group EG has been calculated.

The second torso determination unit 36 performs statistical processing on these combined power spectra PS2_1, PS2_2, and PS2_3 to calculate the frequency F_std_min at which the dispersion value of intensity is the minimum (hereinafter referred to as "minimum dispersion frequency"). .

The second body determination unit 36 determines that the frequency component FC corresponding to the minimum variance frequency F_std_min among the plurality of frequency components FC included in each combined power spectrum PS2 is the frequency component FC corresponding to the body. Also, the second body determining unit 36 determines that the echo E corresponding to the determined frequency component FC is the echo E corresponding to the body.

That is, the second torso discrimination unit 36 determines that, of the three frequency components FC_1, FC_2, and FC_3 included in the synthesized power spectrum PS2_1, the frequency component FC_2 corresponding to the minimum dispersion frequency F_std_min is the frequency component FC corresponding to the torso. and discriminate. Further, the second torso determination unit 36 determines that the echo E_2 corresponding to the frequency component FC_2 among the three echoes E_1, E_2, and E_3 detected at time t1 is the echo E corresponding to the torso ( See Figure 31).

Further, the second body discriminating unit 36 discriminates that the frequency component FC_5 corresponding to the minimum dispersion frequency F_std_min is the frequency component FC corresponding to the body among the two frequency components FC_4 and FC_5 included in the synthesized power spectrum PS2_2. do. Further, the second body discriminating unit 36 discriminates that the echo E_5 corresponding to the frequency component FC_5 is the echo E corresponding to the body among the two echoes E_4 and E_5 detected at time t2 (see FIG. 31). reference).

Further, the second body discriminating unit 36 discriminates that the frequency component FC_6 corresponding to the minimum dispersion frequency F_std_min is the frequency component FC corresponding to the body among the two frequency components FC_6 and FC_7 included in the synthesized power spectrum PS2_3. do. Further, the second body discriminating unit 36 determines that the echo E_6 corresponding to the frequency component FC_6 is the echo E corresponding to the body among the two echoes E_6 and E_7 detected at time t3 (see FIG. 31). reference).

Here, the discrimination result by the second body discrimination section 36 usually corresponds to the discrimination result by the first body discrimination section 32 . However, depending on the pedestrian's walking habits, etc., exceptionally, the determination result by the second body determination unit 36 may not correspond to the determination result by the first body determination unit 32 . In such a case, the determination result by the second body determination section 36 may be preferentially used over the determination result by the first body determination section 32 .

Hereinafter, in the eighth embodiment, the processes executed by the mode determination section 35 and the second body determination section 36 are collectively referred to as "mode determination processing". Such mode determination processing will be described with reference to the flowchart of FIG.

First, the processes of steps ST51 to ST54 are executed. Since the processing contents of steps ST51 to ST54 are the same as the processing contents of steps ST21 to ST24 shown in FIG. 21, description thereof is omitted.

Next, the second body discrimination section 36 executes a second body discrimination process (step ST55). The details of the second body determination process have already been described with reference to FIG. 31, and therefore will not be described again.

As described above, in the obstacle detection device 100, the search wave is transmitted and received multiple times by the distance measuring sensor 2, and the frequency distribution is calculated multiple times by the frequency analysis unit 13, whereby a plurality of frequency distributions are calculated. In this case, a mode determination unit 35 is provided for determining the mode of the object O by executing statistical processing on a plurality of frequency distributions. Thereby, as described in the first embodiment, the mode of the object O can be determined.

In addition, the obstacle detection device 100 includes a second body discrimination section 36 that discriminates the frequency component FC corresponding to the pedestrian's body in each frequency distribution. A frequency component FC having a smaller variation than other frequency components FC is determined to be the frequency component FC corresponding to the trunk. Thereby, when the object O is a pedestrian, the relative velocity RV of the pedestrian with respect to the vehicle 1 can be calculated accurately. Also, at this time, the position coordinates PC of the pedestrian can be calculated accurately.

In addition, within the scope of the present invention, it is possible to freely combine each embodiment, modify any component of each embodiment, or omit any component in each embodiment. .

The obstacle detection device of the present invention can be used, for example, in AEB (Autonomous Emergency Braking).

1 vehicle, 2 ranging sensor, 3 ranging sensor group, 4 control device, 5 sensors, 6 vehicle control device, 11 transmission/reception control unit, 12 echo detection unit, 13 frequency analysis unit, 14 obstacle determination unit, 15 vehicle Information acquisition unit 16 Collision determination unit 17 Warning signal output unit 21 Transmission control unit 22 Reception control unit 31 Object determination unit 32 First body determination unit 33 Speed calculation unit 34 Position calculation unit 35 Mode determination Section 36 Second Body Discrimination Section 41 Processor 42 Memory 43 Processing Circuit 100 Obstacle Detection Device 200 Obstacle Detection System.

Claims (10)

an echo detection unit that detects an echo group within a window having a predetermined time width or distance width corresponding to a pedestrian , using signals received by a ranging sensor provided in the vehicle;
a frequency analysis unit that calculates a frequency distribution in the echo group by performing frequency analysis on the echo group;
an obstacle discrimination unit that discriminates whether or not an object corresponding to the group of echoes is a pedestrian based on the frequency distribution, and outputs a result of the discrimination ;
The obstacle discrimination unit discriminates whether the object is a pedestrian based on the number of echoes in the echo group and the spectrum width in the frequency distribution ,
The obstacle determination unit determines that the object is a bicycle when the number of echoes is equal to or greater than a predetermined number when a condition for determining that the object is a pedestrian is satisfied .
Obstacle detection device. an echo detection unit that detects an echo group within a window having a predetermined time width or distance width corresponding to a pedestrian, using signals received by a ranging sensor provided in the vehicle;
a frequency analysis unit that calculates a frequency distribution in the echo group by performing frequency analysis on the echo group;
an obstacle discrimination unit that discriminates whether or not an object corresponding to the group of echoes is a pedestrian based on the frequency distribution and outputs a result of the discrimination;
When the search wave is transmitted and received a plurality of times by the ranging sensor, and the frequency distribution is calculated a plurality of times by the frequency analysis unit, a plurality of the frequency distributions are calculated. a mode determination unit that determines the mode of the object by executing a process ;
A second body discrimination unit that discriminates frequency components corresponding to the pedestrian's body in each of the frequency distributions,
The second body determination unit determines that a frequency component with a smaller variation than other frequency components in the plurality of frequency distributions is a frequency component corresponding to the body ,
Obstacle detection device. Further comprising a speed calculation unit that calculates the relative speed of the pedestrian with respect to the vehicle based on the Doppler shift amount of the frequency component corresponding to the torso,
The obstacle detection device according to claim 2 . an echo detection unit that detects an echo group within a window having a predetermined time width or distance width corresponding to a pedestrian, using signals received by a ranging sensor provided in the vehicle;
a frequency analysis unit that calculates a frequency distribution in the echo group by performing frequency analysis on the echo group;
an obstacle discrimination unit that discriminates whether or not an object corresponding to the group of echoes is a pedestrian based on the frequency distribution and outputs a result of the discrimination;
When the search wave is transmitted and received a plurality of times by the ranging sensor, and the frequency distribution is calculated a plurality of times by the frequency analysis unit, a plurality of the frequency distributions are calculated. A mode determination unit that determines the mode of the object by executing processing,
The mode determination unit determines whether the object is a pedestrian or a stationary object based on clustering results of the variance of the number of echoes in the echo group and the variance of the spectrum width in the frequency distribution, and determines whether the object is a pedestrian or a stationary object. determine the direction of movement of a person ,
Obstacle detection device. The mode determination unit determines that the object is the stationary object when the variance value of the spectrum width is less than a first threshold .
The obstacle detection device according to claim 4. The mode determination unit determines that the object is a pedestrian and the movement direction is determined when the spectral width variance value is equal to or greater than a first threshold value and the spectral width variance value is less than a second threshold value. determines that is transverse ,
The obstacle detection device according to claim 4. The mode determination unit determines that the object is a pedestrian and the movement direction is determined when the variance value of the spectrum width is equal to or greater than a second threshold value and the variance value of the spectrum width is less than a third threshold value. is oblique ,
The obstacle detection device according to claim 4. The mode determination unit determines that the object is a pedestrian and that the movement direction is the front-back direction when the variance value of the spectrum width is equal to or greater than a third threshold .
The obstacle detection device according to claim 4. The second threshold and the third threshold are set to values that gradually decrease as the variance of the number of echoes increases .
The obstacle detection device according to claim 7. When the obstacle discrimination unit determines that the plurality of objects are a plurality of pedestrians, a pedestrian among the plurality of pedestrians who reaches the collision determination area of the vehicle within a predetermined time is specified. provided with a collision determination unit that
The obstacle detection device according to claim 4.

Images