JP2022142252A - Human body detector and brake control system - Google Patents

Abstract

To detect a human body based solely on data obtained by an ultrasonic sensor.SOLUTION: A human body detector according to an embodiment obtains detection information from an ultrasonic receiving a reflection wave of ultrasonic transmitted to a prescribed area. The waveform of the reflection wave according to the detection information is normalized based on the attenuation characteristic information according to the distance to a reflection object for the ultrasonic and the distance regarding the ultrasonic. If the peak of the normalized waveform of the reflection wave is larger than or equal to the prescribed threshold, it is determined that an object is detected in the prescribed area. If an object detection unit detects an object, the waveform of the reflection wave in a prescribed range including the peak is extracted from the normalized waveform of the reflection wave. For the waveform of the reflection wave in the prescribed range, a prescribed frequency analysis that will not lose time information is conducted to generate multiple parameters. From the multiple parameters, a parameter for determining whether the object is a human or not is extracted. It is determined whether the object is a human or not based on the extracted parameter.SELECTED DRAWING: Figure 9

Description

The present invention relates to a human body detection device and a brake control system.

2. Description of the Related Art Conventionally, for example, in a vehicle control system, there is a technique for detecting a human body (pedestrian, etc.) existing around a vehicle. This technology is realized, for example, based on the result of object detection by a radar device, an ultrasonic sensor, or the like, and an image captured by a camera.

Japanese Patent No. 6643166 Japanese Patent No. 6657934 JP 2006-250927 A JP 2012-189441 A

However, the conventional technology described above has room for improvement in terms of cost and the like.

Therefore, one of the objects of the present invention is to provide a human body detection device and a brake control system that can detect a human body based only on data obtained by an ultrasonic sensor.

The human body detection device of the embodiment includes, for example, an acquisition unit that acquires detection information from an ultrasonic sensor that receives a reflected wave of an ultrasonic wave that has been transmitted to a predetermined area, and a reflected wave waveform that corresponds to the detection information. Based on the distance to the reflecting object of the ultrasonic wave and the attenuation characteristic information related to the ultrasonic wave according to the distance, a normalizing unit that normalizes the peak in the normalized reflected wave waveform is a predetermined threshold or more an object detection unit that determines that an object has been detected in the predetermined area when the object is detected by the object detection unit, and a reflection in a predetermined range including a peak from the normalized reflected wave waveform A waveform extraction unit that extracts a wave waveform; an analysis unit that performs a predetermined frequency analysis without losing time information on the reflected wave waveform in the predetermined range to generate a plurality of parameters; a parameter extraction unit for extracting parameters for determining whether the object is a human body, a human body detection unit for determining whether the object is a human body and detecting the human body based on the extracted parameters; Prepare.
With such a configuration, normalization of the reflected wave waveform of ultrasonic waves, object detection based on peaks in the normalized reflected wave waveform, parameter generation by predetermined frequency analysis without losing time information on the reflected wave waveform, A human body can be detected based only on the data obtained by the ultrasonic sensor by extracting the parameters for human body discrimination and detecting the human body based on the extracted parameters.

Further, the human body detection apparatus further includes a calculation unit that calculates the position and velocity of the object based on the normalized reflected wave waveform when the object is detected by the object detection unit. , the human body detection unit detects a human body by determining whether the object is a human body based on the extracted parameters and the position and speed of the object.
With such a configuration, the human body detection accuracy can be further improved by using the position and speed of the object in addition to the extracted parameters.

Further, in the human body detection device, for example, the analysis unit performs wavelet transform as the predetermined frequency analysis without losing time information on the reflected wave waveform in the predetermined range to generate a plurality of parameters.
With such a configuration, wavelet transform can be specifically used as a predetermined frequency analysis that does not lose time information.

Further, in the human body detection device, for example, the parameter extraction unit determines whether or not the object is a human body among the plurality of parameters based on machine learning data related to parameter extraction stored in a storage unit in advance. Extract the parameters for
With such a configuration, it is possible to extract a parameter for person discrimination from among a plurality of parameters based on machine learning data relating to parameter extraction.

Further, a brake control system can be realized by a human body detection device mounted on a vehicle and a braking control section that controls the brakes of the vehicle based on the human body detection result of the human body detection device.

FIG. 1 is a plan view of a vehicle equipped with a vehicle control system according to an embodiment. FIG. 2 is a block diagram showing the overall configuration of the vehicle control system of the embodiment. FIG. 3 is a block diagram showing the functional configuration of the vehicle control device of the embodiment. FIG. 4 is an explanatory diagram of normalization of reflected wave waveforms of ultrasonic waves in the embodiment. FIG. 5 is an explanatory diagram of extraction of the vicinity of the peak of the reflected wave waveform in the embodiment. FIG. 6 is a graph showing an example of target data for wavelet transform in the embodiment. FIG. 7 is a graph showing a Mexican hat function as an example of basis functions for wavelet transform in the embodiment. FIG. 8 is a flowchart showing overall processing by the vehicle control system of the embodiment. FIG. 9 is a flow chart showing the details of the processing in step S5 of FIG. FIG. 10 is a flow chart showing the details of the processing in step S506 of FIG.

Illustrative embodiments of the invention are disclosed below. The configurations of the embodiments shown below and the actions and effects brought about by the configurations are examples. The present invention can be realized by configurations other than those disclosed in the following embodiments. Moreover, according to the present invention, at least one of various effects (including derivative effects) obtained by the configuration can be obtained.

FIG. 1 is a plan view of a vehicle 10 equipped with a vehicle control system 20 (FIG. 2) of the embodiment. The vehicle 10 may be, for example, an automobile (internal combustion engine automobile) using an internal combustion engine (engine, not shown) as a drive source, or an automobile (electric automobile) using an electric motor (motor, not shown) as a drive source. , a fuel cell vehicle, etc.), or a vehicle (hybrid vehicle) using both of them as a driving source. In addition, the vehicle 10 can be equipped with various transmission devices, and can be equipped with various devices (systems, parts, etc.) necessary for driving the internal combustion engine and the electric motor. Moreover, the system, number, layout, etc. of the devices related to the driving of the wheels 13 in the vehicle 10 can be set variously.

As shown in FIG. 1, the vehicle 10 includes a vehicle body 12, four wheels 13, and twelve ultrasonic sensors 16a-16l. Hereinafter, the ultrasonic sensors 16a to 16l are referred to as the ultrasonic sensor 16 when there is no need to distinguish between them.

The vehicle body 12 constitutes a vehicle compartment in which passengers ride. The vehicle body 12 accommodates or holds wheels 13, ultrasonic sensors 16, and the like.

The four wheels 13 are provided two on the front side of the vehicle body 12 and two on the rear side. For example, the front two wheels 13 function as steered wheels, and the rear two wheels 13 function as driving wheels.

The ultrasonic sensor 16 transmits ultrasonic waves to a predetermined area and receives reflected waves. That is, the ultrasonic sensor 16 is provided, for example, on the outer periphery of the vehicle 10, transmits ultrasonic waves as detection waves, and detects reflections of objects such as human bodies (pedestrians, etc.) existing around the vehicle 10. It is a sonar that catches waves. The ultrasonic sensor 16 outputs detection information (amplitude of reflected waves, time required for transmission and reception, etc.) to the vehicle control device 34 . In addition, when the ultrasonic sensor 16 receives a plurality of detection waves reflected by a plurality of parts of the object for one detection wave transmission, the ultrasonic sensor 16 uses only the required time of the earliest received detection wave as detection information. may be included.

The ultrasonic sensors 16 a , 16 b , 16 c , 16 d are also called side sonars and are provided on the left and right sides of the vehicle 10 . The ultrasonic sensors 16a, 16b, 16c, and 16d detect objects on the sides of the vehicle 10 and output detection information. The ultrasonic sensors 16e and 16f are also called corner sonars, and are provided at the rear of the vehicle 10 (for example, near corners of the vehicle 10) relative to the ultrasonic sensors 16a, 16b, 16c, and 16d. 16c, 16d are oriented rearward (eg, rearwardly outward).

The ultrasonic sensors 16e and 16f detect objects obliquely behind the vehicle 10 and output detection information. The ultrasonic sensors 16g and 16h are also called corner sonars, and are provided in front of the vehicle 10 (for example, near corners of the vehicle 10) relative to the ultrasonic sensors 16a, 16b, 16c, and 16d. , 16c, 16d (for example, to the outside of the front). The ultrasonic sensors 16g and 16h detect objects obliquely ahead of the vehicle 10 and output detection information. The ultrasonic sensors 16 i and 16 j are also called rear sonar and are provided at the rear end of the vehicle 10 . The ultrasonic sensors 16i and 16j detect objects behind the vehicle 10 and output detection information. The ultrasonic sensors 16 k and 16 l are also called front sonar and are provided at the front end of the vehicle 10 . The ultrasonic sensors 16k and 16l detect objects in front of the vehicle 10 and output detection information.

FIG. 2 is a block diagram showing the overall configuration of the vehicle control system 20 of the embodiment. The vehicle control system 20 is mounted on the vehicle 10 and supports driving of the vehicle 10 by automatic driving (partially including automatic driving) according to objects around the vehicle 10 .

As shown in FIG. 2, the vehicle control system 20 includes an ultrasonic sensor 16, a braking system 22, an acceleration system 24, a steering system 26, a transmission system 28, a vehicle speed sensor 30, a monitor device 32, and a vehicle. A controller 34 and an in-vehicle network 36 are provided.

Braking system 22 controls braking of vehicle 10 . Braking system 22 includes a brake section 40 , a brake control section 42 and a brake section sensor 44 .

The braking unit 40 is a device for decelerating the vehicle 10 including, for example, a brake, a brake pedal, and the like.

The braking control unit 42 is, for example, a computer such as a microcomputer having a hardware processor such as a CPU (Central Processing Unit). The braking control unit 42 controls braking of the vehicle 10 by controlling the braking unit 40 based on an instruction (stop control signal) from the vehicle control device 34 .

The braking portion sensor 44 is, for example, a position sensor, and detects the position of the braking portion 40 when the braking portion 40 is a brake pedal. The brake sensor 44 outputs the detected position of the brake 40 to the in-vehicle network 36 .

Acceleration system 24 controls acceleration of vehicle 10 . The acceleration system 24 has an acceleration section 46 , an acceleration control section 48 and an acceleration section sensor 50 .

The acceleration unit 46 is a device for accelerating the vehicle 10 including, for example, an accelerator pedal.

The acceleration control unit 48 is, for example, a computer such as a microcomputer having a hardware processor such as a CPU. The acceleration control unit 48 controls acceleration of the vehicle 10 by controlling the acceleration unit 46 based on an instruction (acceleration control signal) from the vehicle control device 34 .

The acceleration section sensor 50 is, for example, a position sensor, and detects the position of the acceleration section 46 when the acceleration section 46 is an accelerator pedal. The acceleration unit sensor 50 outputs the detected position of the acceleration unit 46 to the in-vehicle network 36 .

The steering system 26 controls the direction of travel of the vehicle 10 . The steering system 26 includes a steering section 52 , a steering control section 54 and a steering section sensor 56 .

The steering unit 52 is a device that includes, for example, a steering wheel, a steering wheel, or the like, and steers the traveling direction of the vehicle 10 by turning the steered wheels of the vehicle 10 .

The steering control unit 54 is, for example, a computer such as a microcomputer having a hardware processor such as a CPU. The steering control unit 54 controls the steering unit 52 based on an instruction (steering control signal) from the vehicle control device 34 to control the traveling direction of the vehicle 10 .

The steering section sensor 56 is, for example, an angle sensor including a hall element or the like, and detects a steering angle, which is the rotation angle of the steering section 52 . The steering unit sensor 56 outputs the detected steering angle of the steering unit 52 to the in-vehicle network 36 .

Transmission system 28 controls the transmission ratio of vehicle 10 . Transmission system 28 includes a transmission section 58 , a transmission control section 60 and a transmission section sensor 62 .

The transmission unit 58 is a device that includes, for example, a shift lever and the like, and changes the gear ratio of the vehicle 10 .

The shift control unit 60 is, for example, a computer such as a microcomputer having a hardware processor such as a CPU. The shift control unit 60 controls the shift unit 58 based on an instruction (shift control signal) from the vehicle control device 34 to control the gear ratio of the vehicle 10 .

The shift section sensor 62 is, for example, a position sensor that detects the position of the shift section 58 when the shift section 58 is a shift lever. The transmission sensor 62 outputs the detected position of the transmission 58 to the in-vehicle network 36 .

The vehicle speed sensor 30 is, for example, a sensor that has a Hall element provided near the wheels 13 of the vehicle 10 and detects the amount of rotation of the wheels 13 or the number of revolutions per unit time. The vehicle speed sensor 30 outputs the wheel speed pulse number indicating the detected rotation amount or number of rotations to the in-vehicle network 36 as a sensor value for calculating the speed of the vehicle 10 (vehicle speed). The vehicle control device 34 can calculate the speed (vehicle speed), movement amount, etc. of the vehicle 10 based on the sensor value acquired from the vehicle speed sensor 30 .

The monitor device 32 is provided on a dashboard or the like inside the vehicle 10 . The monitor device 32 has a display section 64 , an audio output section 66 and an operation input section 68 .

The display unit 64 displays an image based on the image data transmitted by the vehicle control device 34 . The display unit 64 is, for example, a display device such as a liquid crystal display (LCD) or an organic EL display (OELD: Organic Electroluminescent Display). The display unit 64 displays, for example, an image for accepting an operation instruction for switching between automatic operation and manual operation.

The audio output unit 66 outputs audio based on the audio data transmitted by the vehicle control device 34 . The audio output unit 66 is, for example, a speaker. The audio output unit 66 outputs, for example, audio regarding an operation instruction for switching between automatic operation and manual operation.

The operation input unit 68 receives input from the passenger. The operation input unit 68 is, for example, a touch panel. The operation input section 68 is provided on the display screen of the display section 64 . The operation input unit 68 is configured to allow the image displayed by the display unit 64 to pass therethrough. Thereby, the operation input unit 68 allows the passenger to visually recognize the image displayed on the display screen of the display unit 64 . The operation input unit 68 receives an instruction input by the occupant by touching a position corresponding to the image displayed on the display screen of the display unit 64 and transmits the instruction to the vehicle control device 34 . Note that the operation input unit 68 is not limited to a touch panel, and may be a hard switch such as a push button type.

The vehicle control device 34 is a computer including a microcomputer such as an ECU (Electronic Control Unit), and performs control such as automatic driving of the vehicle 10 and parking assistance.

The vehicle control device 34 includes a CPU 34a, a ROM (Read Only Memory) 34b, a RAM (Random Access Memory) 34c, a display control section 34d, an audio control section 34e, and an SSD (Solid State Drive) 34f. . The CPU 34a, ROM 34b and RAM 34c may be integrated in the same package.

The CPU 34a is an example of a hardware processor, reads programs stored in a non-volatile storage device such as the ROM 34b, and executes various arithmetic processes and controls according to the programs. The CPU 34a executes, for example, automatic driving of the vehicle 10 and parking assistance.

The ROM 34b stores each program and parameters necessary for executing the program. The RAM 34c temporarily stores various data used in calculations by the CPU 34a. The display control unit 34 d mainly performs data conversion of images for display to be displayed on the display unit 64 among the arithmetic processing in the vehicle control device 34 . The voice control unit 34e mainly executes voice processing to be output to the voice output unit 66 among the arithmetic processing in the vehicle control device 34. FIG. The SSD 34f is a rewritable non-volatile storage device that retains data even when the vehicle control device 34 is powered off.

The in-vehicle network 36 includes, for example, CAN (Controller Area Network), LIN (Local Interconnect Network), and the like. The in-vehicle network 36 includes the acceleration system 24, the braking system 22, the steering system 26, the transmission system 28, the ultrasonic sensor 16, the vehicle speed sensor 30, the operation input unit 68 of the monitor device 32, the vehicle control device 34 and are connected to each other so that information can be sent and received.

FIG. 3 is a block diagram showing the functional configuration of the vehicle control device 34 of the embodiment. As shown in FIG. 3, the vehicle control device 34 includes, as a functional configuration, an acquisition unit 341, a normalization unit 342, an object detection unit 343, a calculation unit 344, a waveform extraction unit 345, an analysis unit 346, A parameter extraction unit 347 , a human body detection unit 348 , and a control unit 349 are provided. These functional configurations are realized by the CPU 34a reading a program stored in a storage device such as the ROM 34b and executing it. Part or all of each functional configuration may be configured by hardware such as a circuit including an ASIC (Application Specific Integrated Circuit).

Note that each unit 341 to 348 constitutes a human body detection device 3400 . Further, the vehicle control device 34 including the human body detection device 3400 and the braking control section 42 configure an AEB (Advanced Emergency Braking) system (collision damage mitigation braking system) (an example of a braking control system).

The acquisition unit 341 acquires various data from each configuration. For example, the acquisition unit 341 acquires detection information from the ultrasonic sensor 16 that has received a reflected wave of an ultrasonic wave transmitted to a predetermined area.

The normalization unit 342 converts the reflected wave waveform according to the detection information into a distance to an object reflecting the ultrasonic wave (for example, a road surface, a human body (pedestrian, etc.), another vehicle, etc.) and a distance related to the ultrasonic wave. Normalization is performed based on the attenuation characteristic information (for example, attenuation function).

Here, FIG. 4 is an explanatory diagram of normalization of the reflected wave waveform of ultrasonic waves in the embodiment. Ultrasonic waves emitted from the ultrasonic sensor 16 have attenuation characteristics. From this attenuation characteristic, since the amplitude value of the reflected wave waveform decreases as the distance from the reflecting object increases, normalization is performed using an attenuation function in order to match the scale. This attenuation function can be defined by, for example, fitting a reflected wave waveform acquired in a state where there is nothing but the road surface as a reflecting object, but the method of deriving the attenuation function is not limited to this. The amplitude near the object (other than the road surface), which was minute in the reflected wave waveform before normalization in FIG. It can be confirmed that there is a large rise in the waveform.

Returning to FIG. 3, the object detection unit 343 detects an object (an object other than the road surface) in a predetermined area when the peak of the reflected wave waveform normalized by the normalization unit 342 is equal to or greater than a predetermined threshold. It is determined that Peak detection is not limited to such threshold determination, and other methods such as using a peak detection algorithm (for example, CFAR (Constant False Alarm Rate)) may be used.

When the object detection unit 343 detects an object, the calculation unit 344 calculates the position and speed of the object based on the normalized reflected wave waveform. Specifically, the position of the object is calculated in time series based on the normalized reflected wave waveform, and the velocity is calculated based on the position in time series.

When the object detection unit 343 detects an object, the waveform extraction unit 345 extracts a reflected wave waveform in a predetermined range including a peak from the normalized reflected wave waveform (hereinafter also referred to as near-peak extraction). Here, FIG. 5 is an explanatory diagram of extraction of the vicinity of the peak of the reflected wave waveform in the embodiment. The peak of the reflected wave waveform refers to the point where the amplitude value is the largest in the portion of the waveform corresponding to the object. Experiments have shown that if the accuracy of human body discrimination is investigated for each distance without extracting the vicinity of the peak (waveform near the peak coordinates), unnatural results may occur at long distances (5 to 7 m). As a cause of this, as shown in FIG. 5(a), in the raw data, it is considered that there is a deviation in the peak value time coordinate for each object. As a countermeasure, 200 steps before and after the peak value of the acquired reflected wave waveform are extracted, and the peak value time coordinates are aligned. FIG. 5(b) is an example of the waveform after the process of aligning the peak value time coordinates. Also, from the knowledge so far, it is known that the generalization error can be reduced by using the vicinity of the peak in the subsequent processing.

Returning to FIG. 3, the analysis unit 346 performs predetermined frequency analysis without loss of time information on the reflected wave waveform in the predetermined range extracted by the waveform extraction unit 345 to generate a plurality of parameters. A predetermined frequency analysis that does not lose time information is, for example, a wavelet transform.

Here, FIG. 6 is a graph showing an example of target data for wavelet transform in the embodiment. Wavelet transform is one of the methods of frequency analysis, and has the feature that Fourier transform can retain time information that is lost when mapping to frequency space. This makes it possible to grasp the amount of frequency components at each time point of the target waveform.

Specifically, the Fourier transform can retain information in the time domain by using a window function (STFT (Short-Time Fourier Transform)), but the window width must be fixed according to the frequency. is required, so it is not suitable for analysis of a wide frequency range.

On the other hand, the relational expression of continuous wavelet transform is as shown in the following equation (1). Note that t is time. x(t) is the original signal. W _Ψ is the wavelet transform based on the basis functions Ψ. a is the scale parameter. b is a shift parameter. * is the complex conjugate. Note that a and b are continuous real numbers.

Also, an example of the basis function of the continuous wavelet transform is the Mexican hat function shown in Equation (2) below.

Here, FIG. 7 is a graph showing a Mexican hat function as an example of basis functions for wavelet transform in the embodiment. Then, the prepared basis functions (Mexican hat functions) are scaled to create a group of basis functions having various wave widths. If the width of the wave is wide, it is a low-frequency wave, and if it is narrow, it is a high-frequency wave. Here, the inner product is taken while translating each of the prepared basis function groups having various frequency components in the time axis direction with respect to the target waveform. This yields multiple parameters.

Returning to FIG. 3, the parameter extraction unit 347 extracts parameters for determining whether or not the object is a human body from among the plurality of parameters generated by the analysis unit 346. FIG. For example, the parameter extraction unit 347 extracts a parameter for determining whether an object is a human body or not from among a plurality of parameters based on machine learning data related to parameter extraction stored in advance in a storage unit (such as the SSD 34f). . That is, machine learning is performed in advance to identify, among a plurality of parameters, parameters that greatly contribute to determining whether an object is a human body or not. However, the method of parameter extraction is not limited to this, and may be determined in advance on a rule basis. Also, the number of parameters to be extracted is not particularly limited and may be any number.

The human body detection unit 348 detects a human body by determining whether the object is a human body based on the parameters extracted by the parameter extraction unit 347 . For example, the human body detection unit 348 detects a human body by determining whether the object is a human body based on the extracted parameters and the position and speed of the object.

The control unit 349 performs various controls. The control unit 349 controls the running of the vehicle 10, parking assistance, etc. by controlling all or part of the systems 22, 24, 26, and 28, for example, so that the vehicle 10 automatically operates. Also, the control unit 349 transmits a braking instruction to the braking control unit 42, for example. The braking control unit 42 controls the braking of the vehicle based on the human body detection result by the human body detection device 3400 . For example, when a human body (pedestrian, etc.) is detected around the vehicle 10 by the human body detection device 3400, the braking control unit 42 controls the braking unit 40 based on an instruction (stop control signal) from the control unit 349. Then, braking of the vehicle 10 is controlled so that the vehicle 10 stops earlier than when an object other than a human body (for example, an object falling on the road) is detected.

Next, FIG. 8 is a flowchart showing overall processing by the vehicle control system 20 of the embodiment. In step S1, the control unit 349 determines whether or not the vehicle 10 is traveling in a low speed region. If Yes, the process proceeds to step S2, and if No, the process returns to step S1. That is, the AEB system is activated when the vehicle is traveling in a low speed region. In addition, both forward and backward movement are included here.

In step S2, the ultrasonic sensor 16 generates ultrasonic waves. Next, in step S<b>3 , the ultrasonic sensor 16 outputs ultrasonic waves from the vehicle 10 .

Next, in step S4, the ultrasonic sensor 16 receives the reflected ultrasonic waves. Next, in step S5, the human body detection device 3400 performs human body discrimination processing. Here, FIG. 9 is a flow chart showing the details of the process of step S5 in FIG.

In step S501, the human body detection device 3400 is initialized. Next, in step S502, the normalization unit 342 reads the reflected wave waveform of the ultrasonic wave.

Next, in step S503, the normalization unit 342 normalizes the reflected wave waveform ( amplitude scale normalization).

Next, in step S504, the object detection unit 343 determines whether an object (an object other than the road surface) is detected based on whether the peak of the reflected wave waveform normalized in step S503 is equal to or greater than a predetermined threshold. determine whether

Next, in step S505, when the object detection unit 343 detects an object (Yes), the process proceeds to step S506, and when the object is not detected (No), the process of step S5 ends.

In step S506, the calculation unit 344 performs state estimation processing of the object. Here, FIG. 10 is a flowchart showing the details of the processing in step S506 of FIG.

In step S521, the calculator 344 reads the normalized reflected wave waveform. Next, in step S522, the coordinates (position) of the object are estimated (calculated).

Next, in step S523, the calculation unit 344 determines whether or not the coordinate estimation has ended normally. If Yes, the process proceeds to step S524, and if No, the process of step S506 ends.

In step S524, the calculation unit 344 estimates (calculates) the velocity based on the time-series coordinates of the object, and ends the process of step S506.

Returning to FIG. 9, after step S506, in step S507, the waveform extraction unit 345 performs peak vicinity extraction of the normalized reflected wave waveform.

Next, in step S508, the analysis unit 346 performs wavelet transform on the reflected wave waveform extracted in step S507 to generate a plurality of parameters.

Next, in step S509, the parameter extracting unit 347 extracts, from among the plurality of parameters generated in step S508, parameters that contribute highly to determining whether the object is a human body.

Next, in step S510, the human body detection unit 348 classifies the object based on the parameters extracted in step S509. That is, it determines whether the object is a human body. In that case, for example, whether or not the object is a human body is determined based on the position and speed of the object in addition to the extracted parameters.

Next, in step S511, the human body detection unit 348 outputs the classification result, and ends the processing of step S5.

Returning to FIG. 8, after step S5, in step S6, if the human body detection device 3400 detects a human body (Yes), the process proceeds to step S7.If no human body is detected (No), the process returns to step S4. .

In step S7, the control section 349 transmits a stop control signal to the braking control section 42 according to the classification result. Next, in step S<b>8 , the braking control section 42 receives a stop control signal from the control section 349 .

Next, in step S9, the braking control section 42 controls braking of the vehicle 10 by controlling the braking section 40 based on the stop control signal. For example, when a human body (pedestrian or the like) is detected around the vehicle 10, the braking control unit 42 controls the braking unit 40 based on an instruction (stop control signal) from the control unit 349 to To control braking of a vehicle 10 so that the vehicle 10 stops earlier than when an object (for example, a falling object on the road) is detected.

Next, in step S10, the braking control unit 42 determines whether or not the vehicle 10 has stopped. If Yes, the process proceeds to step S11, and if No, the process returns to step S9.

In step S11, the control unit 349 cancels the AEB control and terminates the process.

Thus, according to the vehicle control system 20 of the present embodiment, normalization of the reflected wave waveform of ultrasonic waves, object detection based on the peaks in the normalized reflected wave waveform, and loss of time information with respect to the reflected wave waveform are performed. A human body can be detected based only on the data obtained by the ultrasonic sensor 16 by generating parameters by a predetermined frequency analysis, extracting parameters for human body discrimination, and detecting a human body based on the extracted parameters.

In other words, the human body can be determined with high accuracy based only on the information detected by the ultrasonic sensor 16, which originally has a low resolution. In addition, since it is possible to detect obstacles and classify them (classify whether they are human or not) only with the reflected wave waveform obtained from the ultrasonic sensor 16, the conventionally used sensors and chips can be reduced in the same process. It becomes possible. In addition, even when used in combination with other sensors, it is possible to construct a robust safety system from the standpoint of system redundancy.

In addition to the extracted parameters, based on the position and speed of the object, the human body detection accuracy can be further improved. This can be realized, for example, by a predetermined algorithm based on the position and velocity patterns of the human body and the position and velocity patterns of the non-human body.

Further, by specifically using wavelet transform as a predetermined frequency analysis that does not lose time information, a more accurate human body detection result can be obtained. This is because, for example, compared to the case of Fourier transform, it is possible to grasp the amount of frequency components at each time point of the target waveform.

Further, based on machine learning data related to parameter extraction, a parameter for person discrimination can be extracted with high accuracy from among the plurality of parameters.

A more effective AEB system is realized by the human body detection device 3400 mounted on the vehicle 10 and the braking control unit 42 that controls the brakes of the vehicle 10 based on the human body detection result of the human body detection device 3400. be able to.

A comparison with the prior art will be described below. For example, in the prior art (Japanese Patent No. 6643166), based on a radar target detected from a plurality of reflection points using a radar device and an image target detected from an image captured by an imaging device, It is a technology related to an object recognition device that recognizes an object existing in a space. It is determined that an object has been detected when a target that could not be detected by the radar device is included in the image single target. Also, when the image single target is another vehicle, the vehicle width of the other vehicle is calculated from the distance information acquired from the radar device and the azimuth angle information acquired from the photographing device.

This prior art is an object recognition system by fusion using a plurality of sensors, and when an object that could not be detected by radar is detected by a camera image, it is determined that the object exists. However, object recognition using camera images is greatly affected by weather such as rain and fog, so there may be cases where oversights occur depending on the conditions. The use of lidar, which is expected to become mainstream in autonomous driving in the future, cannot completely solve the conditions in which radar and camera oversights occur, and it is not effective because the system is superimposed.

On the other hand, according to the technology of the present embodiment, since the camera image is not used, it is relatively less susceptible to weather such as rain and fog. In addition, since highly accurate human body detection is made possible only with the information detected by the ultrasonic sensor 16, a simple system is sufficient.

Further, for example, another prior art (Japanese Patent No. 6657934) relates to a person detection device that outputs object candidates based on a plurality of measurement points measured by a laser sensor. First, each measurement point is classified as a road surface or an object on the road surface based on the position in the three-dimensional space and the reflection intensity of the point measured by the laser sensor. After that, clustering is performed for the point cloud classified as an object on the road surface, and each object is detected and recognized.

One of the things that can be done with the data that can be obtained from a laser sensor is that if it is installed in a vehicle, it will be possible to recognize vehicles, pedestrians, distances and shapes to buildings, and their positional relationships as 3D information. be done. This uses the three-dimensional coordinates (point cloud) of each point where the emitted laser pulse hits the object and is reflected. As can be seen from this, there is a problem that the amount of data to be handled becomes enormous because there are so many objects in the actual environment.

On the other hand, according to the technology of this embodiment, only the information detected by the ultrasonic sensor 16 is used, so the amount of data to be handled can be reduced.

Further, for example, still another prior art (Japanese Patent Application Laid-Open No. 2006-250927) discriminates the type of object outside the vehicle (person, wall, vehicle) from the reflected wave characteristics of the received ultrasonic wave, and determines the collision object. It is a technique for judging whether or not to restrain the occupant according to the situation. The reflected wave characteristic here indicates the integrated value, harmonics, and maximum amplitude of the waveform of the reflected wave, and at least two of them are used.

In this prior art, the maximum amplitude value/integrated value of the reflected waveform is used to discriminate the type of the object outside the vehicle. However, when the object is at a short distance, the maximum amplitude value is affected by the reflected wave that bounces off the floor surface. Also, looking at the system flow, the preprocessing for received waves is only envelope processing. In the case of only this processing, it is conceivable that the difference in the integrated value for each object becomes small due to the attenuation characteristics of ultrasonic waves, and high discrimination accuracy cannot be expected for long-distance objects. From the above, there is a high possibility that this prior art cannot achieve the human body detection accuracy required for the AEB system.

On the other hand, according to the technology of the present embodiment, as described above, by normalizing the reflected wave waveform based on the attenuation characteristic information (for example, attenuation function) according to the distance regarding the ultrasonic wave, Therefore, it is possible to achieve the human body detection accuracy required for the AEB system.

Further, for example, still another prior art (Japanese Patent Application Laid-Open No. 2012-189441) discloses a human detection device capable of distinguishing between a stationary person and other stationary objects by using characteristics characteristic of humans. It is a technology related to Whether or not a stationary obstacle is a person is determined based on the amount of temporal variation in the time-series pattern of ultrasonic waves transmitted and received each time, and the degree of similarity between the current object and the person calculated by the pattern identification method. determine. Here, the amount of variation represents the standard deviation of the inner product of the 0th received waveform and subsequent received waveforms, and the degree of similarity is a value representing the human-likeness of the obstacle calculated by the pattern identification method.

Although this technique limits objects to stationary objects, it is desirable that the human body identification system mounted on a vehicle does not depend on the state of the object. In addition, regarding the amount of variation used to identify an object, if the object is stationary, the surface on which the sound wave hits hardly changes, so the variation in the energy amount of the reflected wave becomes small, and the difference between objects is easy to see. However, when the object is moving, for example, in the case of a human being, the surface on which the sound wave hits is different for each waveform, so the variation becomes large and it cannot be handled as a feature amount. From the above, it is considered that this technique cannot be used when the object is moving.

On the other hand, according to the technology of the present embodiment, due to the characteristics of the technology in which the above-described processing is performed using the reflected waveform of the ultrasonic wave, it can be sufficiently used even when the object is moving.

The program for executing the above-described processing executed by the vehicle control device 34 is stored as a file in an installable format or an executable format on a CD-ROM, a CD-R, a memory card, a DVD (Digital Versatile Disk), It may be stored in a computer-readable storage medium such as a flexible disk (FD) and provided as a computer program product. Alternatively, the program may be stored on a computer connected to a network such as the Internet, and provided by being downloaded via the network. Also, the program may be provided or distributed via a network such as the Internet.

Although the embodiment of the present invention has been described above, the above embodiment is presented as an example and is not intended to limit the scope of the invention. This novel embodiment can be embodied in various other forms, and various omissions, replacements, and modifications can be made without departing from the scope of the invention. This embodiment and its modifications are included in the scope and gist of the invention, and are included in the scope of the invention described in the claims and its equivalents.

For example, the wavelet transform basis function is not limited to the Mexican hat function, and other basis functions may be used.

Also, the number of ultrasonic sensors 16 is not limited to twelve, and may be another number.

DESCRIPTION OF SYMBOLS 10... Vehicle, 16... Ultrasonic sensor, 20... Vehicle control system, 34... Vehicle control apparatus, 42... Braking control part, 341... Acquisition part, 342... Normalization part, 343... Object detection part, 344... Calculation part, 345...Waveform extractor, 346...Analyser, 347...Parameter extractor, 348...Human body detector, 349...Controller, 3400...Human body detector

Claims (5)

an acquisition unit that acquires detection information from an ultrasonic sensor that receives a reflected wave of an ultrasonic wave that has been transmitted to a predetermined area;
a normalization unit that normalizes the reflected wave waveform according to the detection information based on the distance to the reflecting object of the ultrasonic wave and the attenuation characteristic information according to the distance related to the ultrasonic wave;
an object detection unit that determines that an object has been detected in the predetermined area when a peak in the normalized reflected wave waveform is equal to or greater than a predetermined threshold;
a waveform extraction unit for extracting a reflected wave waveform in a predetermined range including a peak from the normalized reflected wave waveform when the object is detected by the object detection unit;
an analysis unit that performs a predetermined frequency analysis that does not lose time information on the reflected wave waveform in the predetermined range to generate a plurality of parameters;
a parameter extraction unit for extracting a parameter for determining whether the object is a human body from among the plurality of parameters;
a human body detection unit that detects a human body by determining whether the object is a human body based on the extracted parameters;
A human body detection device comprising: further comprising a calculation unit that calculates the position and speed of the object based on the normalized reflected wave waveform when the object is detected by the object detection unit,
2. The human body detection apparatus according to claim 1, wherein said human body detection unit detects a human body by determining whether said object is a human body based on said extracted parameters and the position and speed of said object. 2. The human body detecting apparatus according to claim 1, wherein said analysis unit performs wavelet transform as said predetermined frequency analysis without loss of time information on said reflected wave waveform in said predetermined range to generate a plurality of parameters. 2. The parameter extraction unit extracts, from among the plurality of parameters, a parameter for determining whether the object is a human body or not, based on machine learning data relating to parameter extraction stored in advance in a storage unit. The human body detection device according to . a human body detection device according to any one of claims 1 to 4, which is mounted on a vehicle;
A brake control system comprising: a braking control unit that controls a brake of the vehicle based on a human body detection result by the human body detection device.

Images