JP6828603B2 - Target detection device - Google Patents

Description

The present invention relates to a target detection device that acquires target information such as the width and length of a three-dimensional object (for example, another vehicle) existing around a vehicle (own vehicle).

Conventionally, a lane change support device that supports a steering operation (steering wheel operation) for changing lanes has been known (see, for example, Patent Document 1). The lane change support device detects a three-dimensional object (for example, another vehicle) existing around the vehicle by a plurality of radar sensors (for example, millimeter-wave radar and laser radar) provided in the vehicle, and the three-dimensional object is detected. Obtains information such as "vertical position, horizontal position and relative vehicle speed" for the own vehicle and "width and length" of the three-dimensional object (hereinafter, also referred to as "target information"). Then, the lane change support device monitors whether or not the own vehicle is safe even if the vehicle changes lanes based on the acquired target information, and executes lane change support when it is determined that the vehicle is safe. Such a "target detection device that acquires target information of a three-dimensional object existing around the own vehicle" is adopted not only in the lane change support device but also in other driving support devices.

JP-A-2009-274594

By the way, as is well known, a radar sensor transmits a radar wave around itself, receives a reflected wave generated by the radar wave being reflected by a three-dimensional object, and is based on the received reflected wave. Recognize the target and acquire the target information. For this reason, the detection range, performance, and at least one of the surrounding environment of the radar sensor mounted on the own vehicle may temporarily prevent the target existing around the own vehicle from being detected. It may not be possible to accurately determine the presence of surrounding objects.

Therefore, in the conventional target detection device, after the target is no longer detected by the radar sensor, uninterrupted object detection is performed by performing extrapolation processing unless the target is re-detected. There is. Further, in the conventional target detection device, when the extrapolation process is performed for the maximum extrapolation duration or longer, it is determined that the sensor target for which the extrapolation process has been performed has been lost (no longer exists). ing.

The extrapolation process (sometimes referred to simply as "extrapolation") is the target information of the detected target when it was detected or an estimated value based on the target information. A process of extrapolating a target by estimating the target information of the undetected target based on the target information and replacing the target information of the undetected target with the estimated target information. To say.

Further, in the conventional device, the magnitude of the parameter (hereinafter, also referred to as “existence probability”) that increases at a constant rate of increase in proportion to the number of continuous operations in which the target is continuously detected is large. Indeed, the maximum extrapolation duration is set to be long.

Further, in the conventional device, by grouping a plurality of targets that are likely to be obtained from one three-dimensional object, one target indicated by the plurality of targets corresponding to the one three-dimensional object (“fusion”). It is also called a "target").

In this case, even if one of the plurality of targets corresponding to one three-dimensional object that has been detected is determined to be lost, even one other target corresponding to one three-dimensional object is detected. If so, the fusion target corresponding to the one solid object is continuously recognized. On the other hand, if all of the plurality of targets corresponding to one three-dimensional object are not detected, the fusion target is determined to be lost.

By the way, a plurality of different targets obtained from one three-dimensional object may alternately repeat detection, or one target obtained from one three-dimensional object may repeat detection and loss temporarily for a short period of time. In these cases, since the number of continuous operations for continuously detecting the target is small, the existence probability is small, the maximum extrapolation duration of the target is also short, and the fusion target corresponding to one three-dimensional object is used. Recognition may be interrupted for a short period of time and temporarily.

The present invention has been made to address the above-mentioned problems. That is, one of the objects of the present invention is that when recognizing one fusion target indicated by a plurality of targets corresponding to one three-dimensional object, the recognition of the fusion target is interrupted for a short period of time and temporarily. It is an object of the present invention to provide a target detection device (hereinafter, also referred to as "device of the present invention") capable of reducing the possibility.

Each device of the present invention detects the reflection point of the radar wave transmitted around the own vehicle by a three-dimensional object as a sensor target, and the vertical distance, the horizontal position, and the relative speed of the detected sensor target with respect to the own vehicle. A plurality of radar sensors (16FC, 16FL, 16FR, 16RL and 16RR) for detecting position / velocity information for acquiring sensor target information including
Every time a predetermined time elapses, a fusion target indicating a three-dimensional object existing around the own vehicle is generated based on the sensor target, and the position and speed of the generated fusion target with respect to the own vehicle are determined. Fusion target generation means (10) calculated based on sensor target information, and
Existence probability calculation means (10) for calculating the existence probability (Trst) of the fusion target, and
When the sensor target corresponding to the specific fusion target, which is the fusion target generated before the predetermined time, is no longer detected, the sensor target corresponding to the specific fusion target is not detected. An extrapolation processing means (10) that performs extrapolation processing to generate an extrapolation fusion target corresponding to the specific fusion target based on the position and speed of the specific fusion target with respect to the own vehicle.
When the extrapolation process is performed for the maximum extrapolation duration or longer, the sensor target presence determination means for determining that the sensor target for which the extrapolation process has been performed does not exist, and
It is provided with a maximum extrapolation duration calculation means (10) for calculating the maximum extrapolation duration, which is the maximum value of the time for performing the extrapolation process based on the existence probability and the vertical relative velocity of the specific fusion target.
The existence probability calculation means
The existence probability on the basis of the longitudinal relative velocity of the fusion object to longitudinal relative velocity magnitude is larger the increase rate of the existence probability of the fusion object is increased when the fusion objects is generated The increase rate of the existence probability is obtained, and the increase amount of the existence probability determined based on the value corresponding to the time during which the fusion target is continuously generated and the increase rate of the existence probability is integrated. It is configured to calculate the probability of existence (step 1422).
The maximum extrapolation duration calculation means is
The maximum extrapolation duration is calculated so that the larger the existence probability is, the longer the maximum extrapolation duration is, and the smaller the magnitude of the longitudinal relative velocity is, the longer the maximum extrapolation duration is. It is configured as in step 1430).

According to this, the existence probability, which is a parameter that increases as the time during which the fusion target is continuously generated increases, is set for the fusion target. Thereby, an appropriate existence probability can be set according to the time during which the fusion target continues to be generated, and an appropriate maximum extrapolation duration can be determined according to the existence probability.

As a result, when the sensor target obtained from one three-dimensional object is detected as follows, the recognition of the fusion target corresponding to one three-dimensional object may be interrupted for a short period of time and temporarily. Can be lowered.
-When multiple different sensor targets obtained from one three-dimensional object repeat detection alternately-When one sensor target obtained from one three-dimensional object repeatedly detects and loses temporarily for a short period of time

In the above description, in order to help the understanding of the present invention, the names and / or symbols used in the embodiments are added in parentheses to the configurations of the invention corresponding to the embodiments described later. However, each component of the present invention is not limited to the embodiment defined by the above name and / or reference numeral. Other objects, other features and accompanying advantages of the present invention will be readily understood from the description of embodiments of the present invention described with reference to the following drawings.

FIG. 1 is a schematic configuration diagram of a driving support device according to an embodiment of the present invention. FIG. 2 is a plan view of the own vehicle showing the arrangement position of the peripheral radar sensor shown in FIG. FIG. 3 is a plan view of the own vehicle and the road for explaining the lane keeping control. 4 (A), (B) and (C) are diagrams for explaining a grouping process for integrating sensor targets. 5 (A) and 5 (B) are diagrams for explaining a grouping process for integrating sensor targets. FIG. 6 is a diagram for explaining a grouping process for integrating sensor targets. FIG. 7 is a plan view of the own vehicle, a three-dimensional object, and a road for explaining the extrapolation of the fusion target. FIG. 8 is a graph showing the relationship between the absolute value of the vertical relative velocity and the rate of increase. FIG. 9 is a graph showing the relationship between the absolute value of the vertical relative velocity, the maximum extrapolation time, and the existence probability. FIG. 10 is a time chart for explaining an embodiment. FIG. 11 is a time chart for explaining a reference example. FIG. 12 is a flowchart showing a routine executed by the CPU of the driving support ECU shown in FIG. FIG. 13 is a flowchart showing a routine executed by the CPU of the driving support ECU shown in FIG. FIG. 14 is a flowchart showing a routine executed by the CPU of the driving support ECU shown in FIG. FIG. 15 is a flowchart showing a routine executed by the CPU of the driving support ECU shown in FIG.

Hereinafter, the target information acquisition device according to the embodiment of the present invention will be described with reference to the drawings. This target information acquisition device is incorporated in a lane change support device (hereinafter, also referred to as "this implementation device") which is a part of a driving support control device (vehicle travel control device).

(Constitution)
As shown in FIG. 1, the present implementation device is applied to a vehicle (hereinafter, referred to as "own vehicle" to distinguish it from other vehicles), and is applied to a driving support ECU 10, an engine ECU 30, and a brake ECU 40. , Steering ECU 50, meter ECU 60 and display ECU 70. In the following, the driving support ECU 10 is also simply referred to as a “DS ECU”.

These ECUs are electric control units (Electric Control Units) including a microcomputer as a main part, and are connected to each other so as to be able to transmit and receive information via a CAN (Controller Area Network) (not shown). In the present specification, the microcomputer includes a CPU, ROM, RAM, non-volatile memory, interface I / F, and the like. The CPU realizes various functions by executing instructions (programs, routines) stored in ROM. Some or all of these ECUs may be integrated into one ECU.

The DSECU is connected to the sensors (including switches) listed below and is adapted to receive detection signals or output signals of those sensors. Each sensor may be connected to an ECU other than the DSECU. In that case, the DSECU receives the detection signal or the output signal of the sensor from the ECU to which the sensor is connected via the CAN.

The accelerator pedal operation amount sensor 11 detects the operation amount (accelerator opening degree) of the accelerator pedal 11a of the own vehicle, and outputs a signal representing the accelerator pedal operation amount AP. The brake pedal operation amount sensor 12 detects the operation amount of the brake pedal 12a of the own vehicle and outputs a signal indicating the brake pedal operation amount BP.

The steering angle sensor 13 detects the steering angle of the own vehicle and outputs a signal representing the steering angle θ.
The steering torque sensor 14 detects the steering torque applied to the steering shaft US of the own vehicle by operating the steering handle SW, and outputs a signal representing the steering torque Tra.
The vehicle speed sensor 15 detects the traveling speed (vehicle speed) of the own vehicle and outputs a signal representing the vehicle speed Vsx. That is, the vehicle speed Vsx is the speed (that is, the vertical speed) in the front-rear direction of the vehicle (the direction along the central axis extending in the front-rear direction of the own vehicle).

The peripheral sensor 16 includes a peripheral radar sensor 16a and a camera sensor 16b.

As shown in FIG. 2, the peripheral radar sensor 16a includes a central front peripheral sensor 16FC, a right front peripheral sensor 16FR, a left front peripheral sensor 16FL, a right rear peripheral sensor 16RR, and a left rear peripheral sensor 16RL.

Peripheral sensors 16FC, 16FR, 16FL, 16RR and 16RL are referred to as peripheral radar sensors 16a when it is not necessary to distinguish them individually. The peripheral sensors 16FC, 16FR, 16FL, 16RR and 16RL have substantially the same configuration as each other.

The peripheral radar sensor 16a includes a radar transmission / reception unit and a signal processing unit (not shown). The radar transmitter / receiver emits radar waves (hereinafter referred to as "millimeter waves"), which are radio waves in the millimeter wave band, and further, three-dimensional objects (for example, other vehicles, pedestrians, bicycles) existing within the radiation range. And receive millimeter waves (ie, reflected waves) reflected by buildings, etc.). A point of a three-dimensional object that reflects radar waves is also called a "reflection point".

The signal processing unit is based on the phase difference between the transmitted millimeter wave and the received reflected wave, their frequency difference, the attenuation level of the reflected wave, the time from the transmission of the millimeter wave to the reception of the reflected wave, and the like. Therefore, the reflection point information indicating the distance between the own vehicle SV and the reflection point of the three-dimensional object, the relative speed between the own vehicle SV and the reflection point of the three-dimensional object, and the direction of the reflection point of the three-dimensional object with respect to the own vehicle SV is predetermined. It is acquired every time the time elapses and supplied to the DSECU. The reflection point of this three-dimensional object is regarded as a target and is also called a "sensor target".

The center front peripheral sensor 16FC is provided in the front center portion of the vehicle body and detects a sensor target existing in the front region of the own vehicle SV. The right front peripheral sensor 16FR is provided at the right front corner of the vehicle body, and mainly detects a sensor target existing in the right front region of the own vehicle SV. The left front peripheral sensor 16FL is provided at the left front corner of the vehicle body, and mainly detects a sensor target existing in the left front region of the own vehicle SV. The right rear peripheral sensor 16RR is provided at the right rear corner of the vehicle body, and mainly detects a sensor target existing in the right rear region of the own vehicle SV. The left rear peripheral sensor 16RL is provided at the left rear corner of the vehicle body, and mainly detects a sensor target existing in the left rear region of the own vehicle SV. For example, the peripheral radar sensor 16a detects a sensor target whose distance from the own vehicle SV is within a range of about 100 meters. The peripheral radar sensor 16a may be a radar sensor that uses radio waves (radar waves) in a frequency band other than the millimeter wave band.

As shown in FIG. 2, the DSECU defines the XY coordinates. The X-axis is a coordinate axis that extends along the front-rear direction of the own vehicle SV so as to pass through the center position in the width direction of the front end portion of the own vehicle SV and has the front as a positive value. The Y-axis is a coordinate axis that is orthogonal to the X-axis and has a positive value in the left direction of the own vehicle SV. The origin of the X-axis and the origin of the Y-axis are the center positions in the width direction of the front end portion of the own vehicle SV.

Based on the above-mentioned reflection point information, the peripheral radar sensor 16a transmits the following "information about the sensor target" to the DSPE every time a predetermined time (calculation cycle) elapses. The information about the sensor target is hereinafter referred to as "sensor target information". Each time the calculation cycle elapses, the DSPE directly acquires the above-mentioned reflection point information from the peripheral radar sensor 16a, and calculates and obtains the "sensor target information" based on the reflection point information. Sensor target information may be acquired.

-X coordinate position (Xobj) of the sensor target. That is, the signed distance between the own vehicle SV and the sensor target in the X-axis direction. The X coordinate position Xobj is also referred to as a vertical distance Xobj or a vertical position Xobj.
-Y coordinate position (Yobj) of the sensor target. That is, the signed distance between the own vehicle SV and the sensor target in the Y-axis direction. The Y coordinate position Yobj is also referred to as a horizontal position Yobj.
-Velocity (that is, vertical relative velocity) Vxobj in the X-axis direction with respect to the own vehicle SV of the sensor target. The magnitude of the vertical relative velocity Vxobj (| Vxobj |) may be expressed as the vertical absolute velocity Vxobj. The vertical absolute speed Vaxobj is a value obtained by adding the vehicle speed V of the own vehicle SV to the vertical relative speed Vxobj.
-Velocity in the Y-axis direction (that is, lateral relative velocity) Vyobj of the sensor target with respect to the own vehicle SV. The magnitude of the lateral relative velocity Vyobj (| Vyobj |) may be expressed as the lateral absolute velocity Vyobj. The lateral absolute velocity Vyobj is set to a value equal to the lateral relative velocity Vyobj.
-Information for each sensor target (sensor target ID) for identifying (identifying) the sensor target

By the way, one three-dimensional object may have two or more reflection points. Therefore, each of the peripheral radar sensors 16a may detect a plurality of sensor targets for one three-dimensional object. That is, each of the peripheral radar sensors 16a may acquire a plurality of sets of sensor target information. Further, two or more peripheral radar sensors 16a may acquire a plurality of sets of sensor target information for one three-dimensional object.

Therefore, the DSECU groups (integrates, fuses) a plurality of sensor targets that are likely to detect one three-dimensional object, thereby indicating one target indicated by the plurality of sensor targets (hereinafter, "fusion"). It is called a "target").

Further, the DSECU acquires the "information on the attribute value of the fusion target" as described later, and transmits the information to the DSECU every time a predetermined time (calculation cycle) elapses. The information about the fusion target is referred to as "fusion target information or fusion target attribute value" and includes the information described below.

-X coordinate position (Xf) of the fusion target. That is, the signed distance between the own vehicle SV and the fusion target in the X-axis direction. In this example, the X coordinate position Xf is the X coordinate position of the center point of the fusion target.
-Y coordinate position (Yf) of the fusion target. That is, the signed distance between the own vehicle SV and the fusion target in the Y-axis direction. In this example, the Y coordinate position Yf is the Y coordinate position of the center point of the fusion target.
-Velocity (that is, vertical relative velocity) Vxf in the X-axis direction with respect to the own vehicle SV of the fusion target.
-Velocity (that is, lateral relative velocity) Vyf in the Y-axis direction with respect to the own vehicle SV of the fusion target.
-Length of fusion target Lf (length of fusion target in the X-axis direction).
-Width of fusion target Wf (length of fusion target in the Y-axis direction).
-Information for each fusion target (fusion target ID) for identifying (identifying) the fusion target

The camera sensor 16b includes a camera unit that is a stereo camera, and a lane recognition unit that analyzes image data obtained by photographing by the camera unit and recognizes a white line on a road. The camera sensor 16b (camera unit) captures the scenery in front of the own vehicle SV. The camera sensor 16b (lane recognition unit) analyzes the image data of the image processing area having a predetermined angle range (range extending in front of the own vehicle SV), and a white line (section) formed on the road in front of the own vehicle SV. Line) is recognized (detected). The camera sensor 16b transmits information about the recognized white line to the DSP.

Based on the information supplied from the camera sensor 16b, the DSECU has left and right white lines WL in the lane in which the own vehicle SV is traveling (hereinafter, also referred to as “own lane”), as shown in FIG. The lane center line CL, which is the center position in the width direction, is specified. This lane center line CL is used as a target traveling line in lane keeping support control described later. Further, the DESCU calculates the curvature Cu of the curve of the lane center line CL. The curvature Cu is defined to have a positive value when the lane center line CL is curved to the right and a negative value when the lane center line CL is curved to the left.

In addition, the DESCU calculates the position and orientation of the own vehicle SV in the lane partitioned by the left white line and the right white line. For example, as shown in FIG. 3, the DESCU calculates the signed distance Dy in the road width direction between the reference point P (for example, the position of the center of gravity) of the own vehicle SV and the lane center line CL. The magnitude of the signed distance Dy indicates the distance at which the own vehicle SV deviates from the lane center line CL in the road width direction. The signed distance Dy becomes a positive value when the reference point P of the own vehicle SV shifts to the right side in the road width direction with respect to the lane center line CL, and the reference point P of the own vehicle SV is the lane center line CL. It is defined to be a negative value when shifting to the left side in the road width direction. This signed distance Dy is also referred to hereinafter as "lateral deviation Dy".

The DESCU calculates the angle θy formed by the direction of the lane center line CL and the direction in which the own vehicle SV is facing (the direction of the front-rear axis of the own vehicle SV). This angle θy is also referred to hereinafter as “yaw angle θy”. The yaw angle θy becomes a positive value when the direction in which the own vehicle SV is facing is clockwise with respect to the direction of the lane center line CL, and the direction in which the own vehicle SV is facing is in the direction of the lane center line CL. On the other hand, it is defined to have a negative value when it is on the counterclockwise side. Hereinafter, the information (Cu, Dy, θy) representing the curvature Cu, the lateral deviation Dy, and the yaw angle θy may be referred to as “lane-related vehicle information”.

The camera sensor 16b supplies the DSECU with information about the types of the left white line and the right white line of the own lane (for example, whether they are solid lines or broken lines) and the shape of the white lines. Further, the camera sensor 16b also supplies the DSECU with the types of left and right white lines and the shape of the white lines in the lane adjacent to the own lane. That is, the camera sensor 16b also supplies "information about the white line" to the DSECU. If the white line is a solid line, vehicles are prohibited from changing lanes across the white line. On the other hand, if the white line is a broken line (white line formed intermittently at regular intervals), the vehicle is allowed to change lanes across the white line. Lane-related vehicle information (Cu, Dy, θy) and information on white lines may be referred to as "lane information".

Referring to FIG. 1 again, the operation switch 17 makes a selection as to whether or not to execute each of the “lane change support control, lane keeping control, and follow-up inter-vehicle distance control” described later. It is an operator operated by. Therefore, the operation switch 17 outputs a signal indicating whether or not the execution of each of the above controls is selected according to the operation of the driver. In addition, the operation switch 17 also has a function of causing the driver to input or select a parameter (for example, the inter-vehicle time described later) for reflecting the driver's preference when executing each of the above controls.

The DESCU determines whether or not the execution of the following inter-vehicle distance control is selected based on the signal supplied from the operation switch 17, and if the execution of the following inter-vehicle distance control is not selected, the lane change support control and the lane Maintenance control is not executed. Further, the DSECU determines whether or not the execution of the lane keeping control is selected based on the signal supplied from the operation switch 17, and if the execution of the lane keeping control is not selected, the lane change support control is executed. It is designed not to.

The yaw rate sensor 18 detects the yaw rate YRt of the own vehicle SV and outputs the actual yaw rate YRt. The actual yaw rate YRt is a positive value when the own vehicle SV is turning left while moving forward, and is a negative value when the own vehicle SV is turning right while moving forward.
The front-rear acceleration sensor 19 detects the acceleration Gx in the front-rear direction of the own vehicle SV and outputs the actual front-rear acceleration Gx. The actual front-rear acceleration Gx becomes a positive value when the own vehicle SV is accelerating forward, and becomes a negative value when the vehicle is decelerating.
The lateral acceleration sensor 20 detects the acceleration Gy in the lateral (vehicle width) direction of the own vehicle SV (the direction orthogonal to the central axis of the own vehicle SV) and outputs the actual lateral acceleration Gy. The actual lateral acceleration Gy becomes a positive value when the own vehicle SV is turning left while moving forward (that is, with respect to the acceleration in the right direction of the vehicle), and the own vehicle SV is turning right while moving forward. In some cases (ie, with respect to vehicle leftward acceleration) it will be a negative value.

The engine ECU 30 is connected to the engine actuator 31. The engine actuator 31 is an actuator for changing the operating state of the internal combustion engine 32. In this example, the internal combustion engine 32 is a gasoline fuel injection / spark ignition type / multi-cylinder engine, and includes a throttle valve for adjusting the intake air amount. The engine actuator 31 includes at least a throttle valve actuator that changes the opening degree of the throttle valve. The engine ECU 30 can change the torque generated by the internal combustion engine 32 by driving the engine actuator 31. The torque generated by the internal combustion engine 32 is transmitted to drive wheels (not shown) via a transmission (not shown). Therefore, the engine ECU 30 can control the driving force of the own vehicle SV and change the acceleration state (acceleration) by controlling the engine actuator 31.

The brake ECU 40 is connected to the brake actuator 41. The brake actuator 41 is provided in a hydraulic circuit between a master cylinder (not shown) that pressurizes hydraulic oil by the pedaling force of a brake pedal and a friction brake mechanism 42 provided on the left, right, front, and rear wheels. The friction brake mechanism 42 includes a brake disc 42a fixed to the wheel and a brake caliper 42b fixed to the vehicle body. The brake actuator 41 adjusts the hydraulic pressure supplied to the wheel cylinder built in the brake caliper 42b in response to an instruction from the brake ECU 40, and operates the wheel cylinder by the hydraulic pressure to press the brake pad against the brake disc 42a to cause friction. Generates braking force. Therefore, the brake ECU 40 can control the braking force of the own vehicle SV and change the acceleration state (deceleration) by controlling the brake actuator 41.

The steering ECU 50 is a well-known control device for an electric power steering system, and is connected to a motor driver 51. The motor driver 51 is connected to the steering motor 52. The steering motor 52 is incorporated in the "steering mechanism including the steering wheel, the steering shaft connected to the steering handle, the steering gear mechanism, and the like" of the vehicle. The steering motor 52 generates torque by the electric power supplied from the motor driver 51, and the steering assist torque can be applied or the left and right steering wheels can be steered by this torque. That is, the steering motor 52 can change the steering angle (steering angle of the steering wheel) of the own vehicle SV.

The steering ECU 50 is connected to the turn signal lever switch 53. The blinker lever switch 53 is a detection switch that detects an operating position of the blinker lever operated by the driver to operate (blink) the turn signal lamp 61 described later.

The blinker lever is provided on the steering column. The winker lever has two positions: a first stage position that is rotated by a predetermined angle in the clockwise operation direction from the initial position, and a second stage position that is rotated in a clockwise operation direction by a predetermined rotation angle from the first stage position. It can be operated in one position. The turn signal lever maintains that position as long as it is maintained in the first stage position in the clockwise operation direction by the driver, but when the driver releases the turn signal lever, it automatically returns to the initial position. There is. The blinker lever switch 53 outputs a signal to the steering ECU 50 indicating that the blinker lever is maintained in the first stage position in the clockwise operation direction when the blinker lever is in the first stage position in the clockwise operation direction.

Similarly, the winker lever has a first stage position rotated counterclockwise from the initial position by a predetermined angle and a second stage position rotated counterclockwise by a predetermined rotation angle from the first stage position. It can be operated in two positions,, and. The blinker lever maintains its position as long as it is maintained in the first stage position in the counterclockwise operation direction by the driver, but when the driver releases the blinker lever, it automatically returns to the initial position. There is. The blinker lever switch 53 outputs a signal to the steering ECU 50 indicating that the blinker lever is maintained in the first stage position in the counterclockwise operation direction when the blinker lever is in the first stage position in the counterclockwise operation direction. It should be noted that such a blinker lever is disclosed in, for example, Japanese Patent Application Laid-Open No. 2005-138647.

Based on the signal from the blinker lever switch 53, the DSECU measures the duration of the blinker lever held in the first stage position in the clockwise operation direction. Further, when the DSPE determines that the measured duration is equal to or longer than the preset support request confirmation time (for example, 0.8 seconds), the lane change support is provided for the driver to change lanes to the right lane. It is determined that a request to receive the request (hereinafter, also referred to as a "lane change support request") has been issued.

Further, the DSECU measures the duration of the blinker lever held in the first stage position in the counterclockwise operation direction based on the signal from the blinker lever switch 53. Further, when the DSPE determines that the measured duration is equal to or longer than the preset support request confirmation time, it determines that the driver has issued a lane change support request in order to change the lane to the left lane. It has become like.

The meter ECU 60 is connected to the left and right turn signal lamps 61 (turn signal lamps) and the information display 62.

The meter ECU 60 blinks the left or right turn signal lamp 61 in response to a signal from the blinker lever switch 53, an instruction from the DSP ECU, or the like via a blinker drive circuit (not shown). For example, the meter ECU 60 blinks the left turn signal lamp 61 when the blinker lever switch 53 outputs a signal indicating that the blinker lever is maintained at the first stage position in the counterclockwise operation direction. Further, the meter ECU 60 blinks the right turn signal lamp 61 when the blinker lever switch 53 outputs a signal indicating that the blinker lever is maintained at the first stage position in the clockwise operation direction.

The information display 62 is a multi-information display provided in front of the driver's seat. The information display 62 displays various information in addition to measured values such as vehicle speed and engine speed. For example, when the meter ECU 60 receives a display command from the DSECU according to the driving support state, the meter ECU 60 displays the screen designated by the display command on the information display 62.

The display ECU 70 is connected to the buzzer 71 and the display 72. The display ECU 70 can sound the buzzer 71 to alert the driver in response to an instruction from the DSECU. Further, the display ECU 70 lights a warning mark (for example, a warning lamp) on the display 72, displays an alarm image, displays a warning message, and controls driving support in response to an instruction from the DSECU. It is possible to display the operating status of. Although the display 72 is a head-up display, it may be another type of display.

(Overview of basic driving support control)
As described above, the DSECU is adapted to execute the following inter-vehicle distance control, the lane keeping control, and the lane change support control. The lane keeping control is executed only when the following inter-vehicle distance control is executed. The lane change support control is executed only when the lane keeping control is executed. Details of these controls will be described later, but these controls will be briefly described below.

The following vehicle-to-vehicle distance control follows the own vehicle SV following the preceding vehicle while maintaining the distance between the preceding vehicle (that is, the vehicle to be followed) and the own vehicle SV traveling in front of the own vehicle SV at a predetermined distance. It is a control to make.

The lane keeping control applies steering torque to the steering mechanism so that the position of the own vehicle SV is maintained near the target driving line in the "lane (own lane) in which the own vehicle SV is traveling". It is a control that changes the steering angle of the SV to support the steering operation of the driver.

The lane change support control applies steering torque to the steering mechanism so that the own vehicle SV moves from the own lane (former lane) to the "lane adjacent to the original lane (target adjacent lane) desired by the driver". By doing so, the steering angle of the own vehicle SV is changed, and thus the steering operation (steering operation for changing lanes) of the driver is supported.

<Outline of operation>
By the way, in order to determine whether or not the own vehicle SV can safely change lanes when executing the lane change support control, the DSECU determines the position and relative speed of a three-dimensional object existing around the own vehicle SV. And it is necessary to acquire three-dimensional object information about the size (length, width) and the like. The DESCU recognizes this three-dimensional object by generating the above-mentioned "fusion target", and also obtains the above-mentioned "fusion target information (for example," length, width and coordinate position "of the fusion target)". Acquired as three-dimensional object information.

Every time a predetermined calculation cycle (Δt) elapses, the DSPE groups (integrates) the sensor targets detected by the peripheral radar sensor 16a by a grouping process described later to generate or update the fusion target. Further, the DSPE generates the fusion target information of the generated or updated fusion target based on the sensor target information of the sensor target (that is, the grouped sensor target) belonging to the fusion target.

(Generation / update of fusion target and lost judgment)
Next, a method of generating / updating a fusion target and determining lost by executing the DSPE will be described.

The DSPE acquires sensor target information from the peripheral radar sensor 16a every time the predetermined time (calculation cycle) Δt elapses. Each target detection range of the peripheral radar sensor 16a partially overlaps with the other peripheral radar sensors 16a. Further, each of the peripheral radar sensors 16a may recognize a plurality of sensor targets from the three-dimensional object even when only one three-dimensional object exists around the own vehicle SV. Therefore, there may be a case where a plurality of sensor targets are acquired from one three-dimensional object.

In this case, the DSECU performs a grouping process described later to group (integrate, fuse) "a plurality of sensor targets that are likely to be obtained from one three-dimensional object n", thereby forming the one three-dimensional object. A fusion target FBn corresponding to n is generated. In other words, the DSPE integrates the plurality of sensor targets to generate the fusion target FBn based on the sensor target information of each of the plurality of sensor targets. Then, the DSPE integrates the fusion target information of the fusion target FBn into the fusion target FBn (that is, belongs to the fusion target FBn), and the sensor of the high AGE sensor target among the sensor targets. Generated based on target information. Hereinafter, the "grouping process" will be described in detail with reference to the examples shown in FIGS. 4A and 4B.

Now, as shown in FIG. 4 (A), it is assumed that the sensor targets B0, B1 and B2 are detected. In this example, the sensor target B0 is a sensor target detected by the right front peripheral sensor 16FR, and the sensor target B1 and the sensor target B2 are sensor targets detected by the central front peripheral sensor 16FC. Furthermore, in this example, no fusion target has been generated so far (in other words, at the time of the previous calculation).

As described above, when the fusion target FBn is not generated at the start of the current calculation, the DSPE performs the grouping process for generating the new fusion target FBn as described below. In addition, this grouping process is called "new target generation grouping process".

First, the DSPE selects any one sensor target (for example, sensor target B0) as the grouping reference target Bs from the plurality of sensor targets (for example, sensor targets B0 to B2). Next, the DESCU has "another sensor target Bn (for example, sensor target Bn, n = 1, 2) that is a grouping candidate" with respect to the grouping reference target Bs (for example, sensor target B0). , It is determined whether or not both of the following conditions (condition G1) and (condition G2) are satisfied. When the sensor target Bn of the grouping candidate satisfies both the following conditions (condition G1) and (condition G2), it is determined that the sensor target Bn satisfies the grouping condition.

(Condition G1) Condition using position as a criterion as shown in the figure on the left side of FIG. 4 (B).
The absolute value (= | XBn-XBs |) of the difference between the "X coordinate position Xobj (= XBn) of the sensor target Bn of the grouping candidate" and the "X coordinate position Xobj (= XBs) of the grouping reference target Bs" is The difference between the predetermined threshold vertical distance Xth or less and the "Y coordinate position Yobj (= YBn) of the grouping candidate sensor target Bn" and the "Y coordinate position Yobj (= YBs) of the grouping reference target Bs". The absolute value (= | YBn-YBs |) is equal to or less than the predetermined threshold lateral distance Yth.
Here, the threshold vertical distance Xth is “target length L0 × 0.5 + predetermined value α”. The threshold lateral distance Yth is “target width W0 × 0.5 + predetermined value β”. Arbitrary fixed values suitable for determination are used for the target length L0 and the target width W0. For example, the target length L0 is set to the standard length of the motorcycle, and the target width W0 is set to the standard width of the motorcycle.

(Condition G2) Condition using speed as a criterion as shown in the figure on the right side of Fig. 4 (B).
The absolute value (= | VxBn-VxBs |) of the difference between the "vertical relative velocity Vxobj (= VxBn) of the grouping candidate sensor target Bn" and the "vertical relative velocity Vxobj (= VxBs) of the grouping reference target Bs" is , A predetermined threshold vertical velocity difference Vxth or less, and "horizontal relative velocity Vyobj (= VyBn) of the grouping candidate sensor target Bn" and "horizontal relative velocity Vyobj (= VyBs) of the grouping reference target Bs". The absolute value of the difference (= | VyBn-VyBs |) is equal to or less than the predetermined threshold lateral velocity difference Vyth.

Whether or not the condition G2 is satisfied may be determined using the absolute velocity. That is, the condition G2 may be as follows.
The absolute value of the difference between the "vertical absolute velocity of the grouping candidate sensor target Bn" and the "vertical absolute velocity of the grouping reference target Bs" is equal to or less than the threshold longitudinal velocity difference Vxth, and the "grouping candidate sensor target" The absolute value of the difference between "the lateral absolute velocity of Bn" and "the lateral absolute velocity of the grouping reference target Bs" is equal to or less than the threshold lateral velocity difference Vyth.

When the grouping candidate sensor target Bn satisfies the grouping condition consisting of both (condition G1) and (condition G2) with respect to the grouping reference target Bs, the DSPE determines the sensor target Bn and the grouping standard. Integrate with the marker Bs to generate a new fusion target FBn. Further, the DSECU sets identification information (ID) for distinguishing (identifying) the fusion target FBn from other fusion targets with respect to the new fusion target FBn.

For example, in FIG. 4A, it is assumed that the grouping candidate sensor target B1 satisfies both the conditions (condition G1) and (condition G2) with respect to the grouping reference target B0. In this case, the DSPE integrates the sensor target B1 and the sensor target B0 to newly generate the fusion target FB1. In this case, the identification information of the fusion target FB1 is, for example, "ID1".

Further, in FIG. 4A, when the grouping candidate sensor target B2 also satisfies both the conditions (condition G1) and (condition G2) with respect to the grouping reference target B0, the DESCU determines the sensor target. B2 is also integrated with the sensor target B0. That is, the sensor target B2 is integrated with the fusion target FB1.

On the other hand, when the grouping candidate sensor target Bn does not satisfy at least one of (condition G1) and (condition G2) with respect to the grouping reference target Bs, the DSPE uses another sensor target Bn. Select as the grouping reference target Bs. Then, in the DESCU, the sensor targets that are grouping candidates (that is, the sensor targets that have not been integrated as fusion targets by then) are (condition G1) and (condition G2) with respect to the grouping reference target Bs. It is determined whether or not both of the grouping conditions of are satisfied. The above process is a new target generation grouping process.

On the other hand, when the fusion target FBn is generated in the previous calculation (calculation before the calculation cycle Δt) (that is, when the fusion target FBn is already generated at the start of the current calculation), the DSPE is used. The fusion target FBn is updated as follows. Further, the DESCU calculates the existence probability of the updated fusion target FBn. In the following, as shown in FIG. 5A, an example is used in which two fusion targets FB1 and FB2 (that is, FBn, n = 1, 2) have already been generated when the current calculation is started. The method of updating (generating) the fusion target will be described. Hereinafter, the fusion target generated or updated in the previous calculation is referred to as "previous fusion target", and the target information of the previous fusion target is referred to as "previous fusion target information".

The DESCU estimates the position and relative velocity of the fusion target FBn in this calculation based on the previous fusion target information of the previous fusion target FBn. This estimated fusion target is referred to as the "estimated target FBn'". For example, in the example shown in FIG. 5A, estimated target targets FB1'and FB2'are generated based on the previous fusion targets FB1 and FB2, respectively.

More specifically, in the XY coordinates at the time of the previous calculation (hereinafter, referred to as "previous XY coordinates"), the XY coordinate positions of the previous fusion target FBn are (Xfn, Yfn). Let Vxfn be the vertical relative velocity of the previous fusion target FBn, and Vyfn be the horizontal relative velocity of the previous fusion target FBn. At this time, the DSPE calculates the XY coordinate positions (Xfn', Yfn') of the estimated target FBn'in the previous XY coordinates according to the following formula.

Xfn'= Xfn + Δt · Vxfn
Yfn'= Yfn + Δt · Vyfn

After that, the DSECU sets the obtained "XY coordinate positions (Xfn', Yfn') of the estimated target FBn'in the previous XY coordinates" to the XY coordinates at the time of this calculation (hereinafter, "this time". It is converted (coordinate conversion) to the XY coordinate position in (referred to as "XY coordinates"). Further, the DSECU converts (coordinate conversion) the "relative velocity (Vxfn, Vyfn) of the previous fusion target FBn" in the previous XY coordinates into the relative velocity in the XY coordinates this time, and this time XY. It is set as the relative velocity of the estimated target FBn'in coordinates. The DESCU describes the relationship between the previous XY coordinates and the current XY coordinates as the amount of change per unit time of "vehicle speed V, lateral deviation Dy, and yaw angle θy of own vehicle SV" and time Δt. From this relationship, the above coordinate transformations such as the XY coordinate position and the relative velocity are performed.

Further, the DSPE sets the "target width and target length" of the estimated target FBn'to the same values as the "target width Wf and target length Lf" of the previous fusion target FBn, respectively. As a result, the DSPE generates the estimated target FBn'(that is, FB1'and FB2').

The estimated target FBn'is a target for grouping (integrating) the sensor target newly detected at the time of this calculation (hereinafter, also referred to as "this detection sensor target"). is there. Therefore, the identification information of the estimated target FBn'is set to the same information as the identification information of the previous fusion target FBn. That is, for example, the identification information of the estimated target FB1'is maintained in "ID1" which is the identification information of the previous fusion target FB1. The identification information of the estimated target FB2'is maintained in "ID2" which is the identification information of the previous fusion target FB2.

Next, the DSPE extracts the sensor target this time, which is a grouping candidate for the estimated target FBn'. This extraction is based on the position of the estimated target FBn'. More specifically, the DSPE extracts the "current detection sensor target" in the grouping target area determined based on the position of the estimated target FBn'as the grouping target of the estimated target FBn'.

In the example shown in FIG. 5A, the sensor target BFC1 is the current detection sensor target detected by the central front peripheral sensor 16FC. The sensor targets BFL1, BFL2, and BFL3 are the current detection sensor targets detected by the left front peripheral sensor 16FL this time. The sensor target BRL1 is the current detection sensor target detected this time by the left rear peripheral sensor 16RL. Neither the right front peripheral sensor 16FL nor the right rear peripheral sensor 16RR has detected the detection sensor target this time. The grouping candidates for the estimated target FB1'are "sensor target BFC1, sensor target BFL1, BFL2 and BFL3, and sensor target BRL1" existing in the grouping target area surrounded by the dotted line R1. The grouping candidate for the estimated target FB2'is the "sensor target BRL1" existing in the grouping target area surrounded by the dotted line R2.

The DESCU executes a grouping process (hereinafter, referred to as "first grouping process") for associating the detection sensor target with the previous fusion target FBn based on the estimated target FBn'.

That is, the DSPE first selects the estimated target FBn'as the grouping reference target. Next, the DSPE sets a grouping condition in which the current detection sensor target, which is a grouping candidate, comprises the above-mentioned (condition G1) and (condition G2) with respect to the grouping reference target (that is, the estimated target FBn'). Determine if it is satisfied. In this way, when the grouping reference target is the estimated target FBn', the target information of the grouping reference target is the target information (XY coordinate position, vertical relative velocity, and horizontal relative) of the estimated target FBn'. Speed) is used. In the first grouping process, the threshold vertical distance Xth used in the condition G1 is "target length L1 x 0.5 + predetermined value α", and the threshold lateral distance Yth used in the condition G2 is ". The target width W1 × 0.5 + predetermined value β ”. The "target length and target width" of the estimated target FBn'are used for the target length L1 and the target width W1, respectively.

When the current detection sensor target of the grouping candidate satisfies both the conditions (condition G1) and (condition G2) with respect to the estimated target FBn'selected as the grouping reference target, the DSECU determines the estimated target. The fusion target FBn is updated (generated) by integrating the FBn'and the detection sensor target this time. The DSECU performs this process on all of the grouping candidate current detection sensor targets to update the fusion target FBn. The identification information of the fusion target FBn is maintained in the same information as the identification information of the estimated target FBn'. As a matter of course, the current detection sensor target of the grouping candidate satisfies at least one of (condition G1) and (condition G2) with respect to the estimated target FBn'selected as the grouping reference target. If not, the DSPE does not integrate the estimated target FBn'and its current detection sensor target.

In the example shown in FIG. 5B, among the currently detected sensor targets of the grouping candidates surrounded by the dotted line R1 with respect to the estimated target FB1', the sensor target BFC2 and the sensor target BFL1 are (condition G1). It is assumed that both the conditions (that is, the grouping condition) of (condition G2) are satisfied. In this case, as shown in FIG. 6, the DSPE updates (generates) the fusion target FB1 by integrating the estimated target FB1'and the "sensor target BFC2 and the sensor target BFL1". Therefore, the number of sensor targets (grouping characteristic) determined to be integrated into the estimated target FB1'is "2". Further, the DESCU calculates the existence probability of the updated fusion target FB1. The details of the calculation method of the existence probability will be described later.

Further, in the example shown in FIG. 5B, it is assumed that the sensor target BRL1 which is a grouping candidate does not satisfy the grouping condition with respect to the estimated target FB2'. That is, there is no sensor target that satisfies the grouping condition among the currently detected sensor targets of the grouping candidates surrounded by the dotted line R2 with respect to the estimated target FB2'. In other words, the number of sensor targets (grouping characteristic) determined to be integrated with the estimated target FB2'is "0". In this case, the DESCU extrapolates the fusion target FB2. That is, the DSPE regards the estimated target FB2'as the current fusion target FB2 obtained by extrapolating the fusion target FB2 last time, and the target information of the current fusion target is the estimated target FB2'. Replace with the target information of. This process is referred to as extrapolation or extrapolation of the fusion target. When extrapolated, the existence probability of the fusion target FB2 is maintained unchanged. That is, the DSECU applies the existence probabilities calculated up to that point to the fusion target FB2 as it is.

Further, when there is a detection sensor target (hereinafter, also referred to as “residual sensor target”) that has not been integrated with any of the estimated targets by the first grouping process, the DSPE is used between the residual sensor targets. Try grouping. This process is called a second grouping process.

For example, in the example shown in FIG. 6, the “sensor target BFL2 and BFL3 and the sensor target BRL1” surrounded by the dotted line R3 are residual sensor targets. The DSECU executes the same process as the above-mentioned "new target generation grouping process" for these residual sensor targets as the second grouping process.

That is, the DSPE selects one of the residual sensor targets as the grouping reference target, and extracts the residual sensor target that is a grouping candidate for the selected grouping reference target. Next, the DSPE determines whether or not the extracted residual sensor target, which is a grouping candidate, satisfies the above-mentioned grouping condition with respect to the grouping reference target. Then, the DSPE generates a new fusion target FBn by integrating the grouping reference target and the residual sensor target that satisfies the grouping condition. The DESCU sets identification information (ID) for the new fusion target FBn to distinguish (identify) the fusion target FBn from other fusion targets. The DSECU performs such processing on all of the residual sensor targets.

By the way, like the previous fusion target FB2 described above, there is also one sensor target (this time detection sensor target) that can be integrated into the "estimated target FBn'corresponding to the previous fusion target FBn" in the first grouping process. If this is not done, it can be considered that the three-dimensional object corresponding to the previous fusion target FBn no longer exists around the own vehicle SV. That is, when the number of detection sensor targets (grouping characteristics) that can be integrated with the estimated target FBn'is "0", it can be considered that the fusion target FBn has been lost.

However, in such a situation, as shown in FIG. 7, the three-dimensional object 150 detected by the DSPE as the fusion target FB2 does not emit millimeter waves from any of the peripheral radar sensors 16a, and the blind spot region Rdr or Rdl. It can also occur if you temporarily enter the area. That is, in such a situation, at the current calculation timing, the three-dimensional object 150 corresponding to the fusion target FB2 actually exists around the own vehicle SV, but the sensor target is detected from the three-dimensional object 150. It may be a situation that has not been done.

Therefore, when the grouping characteristic with respect to the estimated target FBn'is "0", if the DSPE immediately determines that "the fusion target FBn has been lost", the determination may be erroneous.

In order to avoid such an erroneous determination, when the grouping characteristic number with respect to the estimated target FBn'is "0", the DSPE extrapolates the fusion target FBn this time based on the estimated target FBn'. When starting the extrapolation of the fusion target, the DESCU calculates the maximum extrapolation duration tg. The calculation method of the maximum extrapolation duration tg will be described later. Then, the extrapolation of the fusion target is continued until the duration (extrapolation duration) from the start of the extrapolation becomes the predetermined maximum extrapolation duration tg or more, and the extrapolation duration is outside the maximum. It ends when the insertion duration tg is reached. However, the extrapolation of the fusion target is a sensor target that is integrated with the estimated target corresponding to the fusion target by the extrapolation before the extrapolation duration reaches the maximum extrapolation duration tg. Appears and ends when the sensor target is integrated into the estimated target.

When a three-dimensional object that temporarily enters the blind spot region Rdr or blind spot region Rdl of the peripheral radar sensor 16a comes out of the blind spot region Rdr or blind spot region Rdl, the estimated target corresponding to the fusion target by extrapolation is used. A sensor target that satisfies the grouping conditions is detected. In this case, the DSPE integrates the sensor target and the estimated target to generate a fusion target, and ends the extrapolation of the fusion target. This makes it possible to reduce the possibility of the above-mentioned erroneous determination occurring.

On the other hand, if a sensor target that satisfies the grouping condition is not detected for the estimated target corresponding to the fusion target by extrapolation until the extrapolation duration reaches the maximum extrapolation duration tg or more, the fusion object is used. It is highly possible that the marker is not temporarily in the blind spot area, but no longer exists around the vehicle SV. Therefore, in this case, the DSECU determines that the fusion target has been lost.

(Calculation of existence probability)
As will be described later, the DSECU calculates the maximum extrapolation duration based on the existence probability of the fusion target. Therefore, when the fusion target is updated, the DESCU calculates the existence probability of the updated fusion target. Specifically, the DSECU adds a predetermined existence probability rup × Cy to the existence probability of the previous fusion target by the following equation (A) to obtain the existence probability Trst of the fusion target in this calculation. calculate.

Trst = Trst _pre + rup x Cy ... (A)
(Trst: Existence probability of fusion target in this calculation, Trst _pr : Existence probability of previous fusion target, rup: Increase rate of existence probability, Cy: Number of calculation cycles)

Therefore, the peripheral radar sensor 16a continuously detects the sensor targets that can be integrated, and the larger the number of calculation cycles that can update the fusion target (the longer the continuous generation time), the higher the existence probability. It is calculated to be large. At this time, as shown in FIG. 8, the DSPE sets the increase rate rup of the existence probability so as to increase as the magnitude (absolute value) of the vertical relative velocity Vxobj of the fusion target increases.

As a result, the existence probability of the fusion target having a small vertical relative velocity Vxf as described below is less likely to increase.
-The magnitude of the vertical relative velocity Vxf generated when a road surface, wall, roadside object, etc. is falsely detected (false positive) is small. The magnitude of the vertical relative velocity Vxf generated when a fusion target or sensor ghost is detected (detected). Fusion target with a small vertical relative speed Vxobj generated when another vehicle with a substantially constant velocity is detected with the own vehicle.

On the other hand, the probability of existence of a fusion target having a large speed and approaching from a distance tends to be large.

(Calculation of maximum extrapolation duration)
The DESCU calculates the maximum extrapolation duration based on the probability of existence of the fusion target. Specifically, as shown in FIG. 9, the maximum extrapolation duration increases as the existence probability increases, and the maximum extrapolation duration increases as the vertical relative velocity Vxf decreases. Calculate the maximum extrapolation duration so that it is longer.

(Effect of this implementation device)
As described above, the present implementation device uses the existence probability, which is a parameter that increases in proportion to the number of operations (number of operation cycles) for which the fusion target is continuously updated (generated), to the fusion target. It is set for. This makes it possible to set an appropriate probability of existence that increases as the period during which the fusion target is generated and continues to be recognized increases, and determines an appropriate maximum extrapolation duration that increases as the probability of existence increases. can do.

As a result, when the sensor target obtained from one three-dimensional object is detected as follows, the recognition of the fusion target corresponding to one three-dimensional object may be interrupted for a short period of time and temporarily. Can be lowered.
-When multiple different sensor targets obtained from one three-dimensional object repeat detection alternately-When one sensor target obtained from one three-dimensional object repeatedly detects and loses temporarily for a short period of time

Hereinafter, the effects of the present implementation device will be described more specifically with reference to FIGS. 10 and 11. In FIGS. 10 and 11, the sensor target Bfr1 and the sensor target Bfc1 obtained from one three-dimensional object h are alternately detected.

Specifically, the sensor target Bfr1 obtained from the three-dimensional object h is detected by the right front peripheral sensor 16FR during the period from the time t0 to immediately before the time t1. The sensor target Bfr1 was not detected during the period from time t1 to immediately before time t4. The sensor object obtained from the three-dimensional object h by the central front peripheral sensor 16FC during the period from time t2 to immediately before time t3 between time t1 and immediately before time t4 when the sensor target Bfr1 is not temporarily detected. Mark Bfc1 has been detected. Then, after time t3, the sensor target Bfr1 is detected again by the right front peripheral sensor 16FR during the period from time t4 to immediately before time t5.

Under such circumstances, an example in which the existence probability is set in the fusion target Fh and the maximum extrapolation duration is determined based on the existence probability (hereinafter, referred to as "Example"). Will be described with reference to FIG. An example in which the existence probability is set in the sensor target without setting the existence probability in the fusion target Fh, and the maximum extrapolation duration of the sensor target is determined based on the existence probability (hereinafter, "reference example"). ”) Will be described with reference to FIG. In the reference example of FIG. 11, the sensor target is extrapolated without extrapolating the fusion target. In this case, the sensor target is extrapolated by replacing the sensor target information estimated based on the sensor target information of the previous sensor target with the target information of the sensor target.

As shown in the embodiment of FIG. 10, the DESCU has a fusion corresponding to the three-dimensional object h by the sensor target Bfr1 obtained from one three-dimensional object h detected by the right front peripheral sensor 16FR at time t0. Generate target Fh. During the period from time t0 to immediately before time t1, the DSPE continues to generate the fusion target Fh by the sensor target Bfr1. Therefore, during this period, the existence probability of the fusion target Fh increases.

At time t1, the sensor target Bfr1 is not detected by the right front peripheral sensor 16FR, and other sensor targets obtained from the three-dimensional object h are not detected by the peripheral radar sensor 16a. Then, the DESCU starts the extrapolation of the fusion target Fh. Further, at this time, the DSPE determines the maximum extrapolation duration tg1 and the maximum extrapolation duration tg1 based on the magnitude of the existence probability Trst1 of the fusion target Fh at time t1.

During the period from time t1 to immediately before time t2, the sensor target Bfr1 is not detected by the right front peripheral sensor 16FR, and other sensor targets obtained from the three-dimensional object h are not detected by the peripheral radar sensor 16a. Therefore, during this period, the DSECU continues to extrapolate the fusion target Fh. During the period of extrapolation of the fusion target, the DSECU maintains the existence probability of the fusion target unchanged.

At time t2 during the extrapolation of the fusion target Fh, the sensor target Bfc1 obtained from the three-dimensional object h is detected by the center front peripheral sensor 16FC. Then, the DSPE finishes the extrapolation of the fusion target Fh and generates the fusion target Fh by the sensor target Bfc1.

During the period from time t2 to immediately before time t3, the DSPE continues to generate the fusion target Fh by the sensor target Bfc1. Therefore, during this period, the existence probability of the fusion target rises again from the maintained existence probability Trist1.

At time t3, the sensor target Bfc1 is not detected by the central front peripheral sensor 16FC, and other sensor targets obtained from the three-dimensional object h are not detected by the peripheral radar sensor 16a. Then, the DESCU starts the extrapolation of the fusion target Fh. Further, the DESCU determines the maximum extrapolation duration tg2 based on the magnitude of the existence probability Trst2 of the fusion target Fh at time t3. At this time, since the existence probability Trst2 of the fusion target Fh at time t3 is larger than the existence probability Trst1 of the fusion target at time t1, the maximum extrapolation duration tg2 is longer than the maximum extrapolation duration tg1. Will be decided.

During the period from time t3 to immediately before time t4, the sensor target Bfc1 is not detected by the central front peripheral sensor 16FC, and other sensor targets obtained from the three-dimensional object h are not detected by the peripheral radar sensor 16a. Therefore, during this period, the DSECU continues to extrapolate the fusion target Fh.

Immediately before time t4 during the extrapolation of the fusion target Fh, the sensor target Bfr1 obtained from the three-dimensional object h is detected again by the right front peripheral sensor 16FR. Then, the DSPE finishes the extrapolation of the fusion target Fh and generates the fusion target Fh by the sensor target Bfr1. Further, the DSPE continues to generate the fusion target Fh by the sensor target Bfr1 which has been continuously detected from the time t4 to just before the time t5.

At time t5, the sensor target Bfr1 is not detected again by the right front peripheral sensor 16FR, and other sensor targets obtained from the three-dimensional object h are not detected by the peripheral radar sensor 16a. Then, the DESCU starts the extrapolation of the fusion target Fh. Further, at this time, the DSPE determines the maximum extrapolation duration tg3 based on the magnitude of the existence probability Trst3 of the fusion target at time t5. At this time, since the existence probability Trst3 of the fusion target Fh at time t5 is larger than the existence probability Trst2 of the fusion target at time t2, the maximum extrapolation duration tg3 is longer than the maximum extrapolation duration tg2. Will be decided.

From the time t5 to the time t6, which is the time when the maximum extrapolation duration tg3 elapses, the sensor target obtained from the three-dimensional object h is not detected by the peripheral radar sensor 16a. Then, at time t6, the DSPE determines that the fusion target is lost.

On the other hand, as shown in the reference example of FIG. 11, the DSECU uses the sensor target Bfr1 obtained from one three-dimensional object h detected by the right front peripheral sensor 16FR at time t0 to form a three-dimensional object. Generate a fusion target Fh corresponding to h.

During the period from time t0 to immediately before time t1, the DSPE continues to generate the fusion target Fh by the sensor target Bfr1 detected by the right front peripheral sensor 16FR. During this period, the sensor target Bfr1 continues to be detected by the right front peripheral sensor 16FR, so that the existence probability of the sensor target Bfr1 increases.

When the sensor target Bfr1 is no longer detected by the right front peripheral sensor 16FR at time t1, the DSPE starts extrapolation of the sensor target Bfr1. Further, at this time, the DSPE determines the maximum extrapolation duration tg1'based on the magnitude of the existence probability Trst1'of the sensor target Bfr1 at time t1.

During the period from time t1 to immediately before time t2, the sensor target Bfr1 is not detected by the right front peripheral sensor 16FR. Therefore, during this period, the DSPE continues to extrapolate the sensor target Bfr1. During the period during which the sensor target Bfr1 is extrapolated, the DSECU maintains the existence probability of the sensor target Bfr1 unchanged. Furthermore, during this period, other sensor targets obtained from the three-dimensional object h are not detected by the peripheral radar sensor 16a. Therefore, the DESCU generates the fusion target Fh by the sensor target Bfr1 during extrapolation.

At time t2 during extrapolation of the sensor target Bfr1, the sensor target Bfc1 obtained from the three-dimensional object h is detected by the center front peripheral sensor 16FC. Then, the DSPE generates the fusion target Fh by the sensor target fr1 and the sensor target Bfc1 during extrapolation.

During the period from the time t2 to the time immediately before the time t2a, the sensor target Bfc1 obtained from the three-dimensional object h is detected by the central front peripheral sensor 16FC, and the sensor target Bfr1 is still undetectable. , Extrapolating. Therefore, the DSPE continues to generate the fusion target Fh by the sensor target Bfr1 being extrapolated and the sensor target Bfc1 detected by the center front peripheral sensor 16FC.

At time t2a, which is the time when the maximum extrapolation duration tg1'elapses from time t1, the sensor target Bfc1 obtained from the three-dimensional object h is detected by the central front peripheral sensor 16FC, and the sensor object being extrapolated. Mark Bfr1 remains undetectable. Therefore, the DSPE determines that the sensor target Bfr1 is lost. At this time, the existence probability of the sensor target Bfr1 is reset (set to an initial value (for example, “0”)). Further, the DSEC generates the fusion target Fh by the sensor target Bfc1 detected by the center front peripheral sensor 16FC.

During the period from the time t2a to the time immediately before the time t3, the DSPE continues to generate the fusion target Fh by the sensor target Bfc1 detected by the center front peripheral sensor 16FC. Since the sensor target Bfc1 is continuously detected by the central front peripheral sensor 16FC during the period from the time t2a to immediately before the time t3, the existence probability of the sensor target Bfc1 increases from the initial value.

When the sensor target Bfc1 is no longer detected by the center front peripheral sensor 16FC at time t3, the DSPE starts extrapolation of the sensor target Bfc1. Further, at this time, the DSPE determines the maximum extrapolation duration tg2'based on the magnitude of the existence probability Trst2'of the sensor target Bfc1 at time t3. Further, at this time, other sensor targets obtained from the three-dimensional object h are not detected by the peripheral radar sensor 16a. Therefore, the DESCU generates the fusion target Fh by the sensor target Bfc1 during extrapolation.

Here, according to the reference example, since the existence probability is set for each sensor target, the existence probability Trst2'is a size corresponding to the period from the time t2 to the time t3 when the sensor target is detected. become. The period from the period t0 recognized as the fusion target Fh to immediately before the time t2 is not reflected in the existence probability Trst2'. Therefore, the existence probability Trst2'is not as large as the case where the period from time t2 to time t3 and the period from time t0 to time t2 are reflected as in the existence probability Trst2 of the embodiment, and the maximum extrapolation duration tg2 'Is also shortened.

Therefore, according to the reference example, the maximum extrapolation duration tg2'elapses at the time t3a at the timing before the time t4 when the next sensor target Bfr1 is detected. Therefore, the DSPE determines that the sensor target Bfc1 is lost at time t3a. Further, the DESCU determines that the fusion target is lost during the period from t3a to immediately before the time t4 because no other sensor target obtained from the three-dimensional object h is detected by the peripheral radar sensor 16a.

On the other hand, according to the embodiment, since the existence probability is set to the fusion target Fh, the existence probability Trst2 has a large value corresponding to the period from the time t0 to the time t3 when the fusion target Fh is recognized. It will be. That is, the period from the period t0 to the time t3, which can be recognized as the fusion target Fh, is accurately reflected in the existence probability of the fusion target Fh. Therefore, the maximum extrapolation duration tg2 is also set to an appropriate length according to the existence probability of the fusion target Fh.

As a result, when a plurality of different sensor targets obtained from one three-dimensional object repeatedly detect each other, it is possible to reduce the possibility that the recognition of the fusion target is interrupted for a short period of time and temporarily. Similarly, when one sensor target obtained from one three-dimensional object repeatedly detects and loses temporarily for a short period of time, it is unlikely that the recognition of the fusion target will be interrupted for a short period of time and temporarily. can do.

(Specific operation)
Next, the specific operation of the present implementation device will be described. The CPU of the DSPE (hereinafter, simply referred to as “CPU”) is a routine shown in FIGS. 12, 13 and 15 every time a predetermined time (predetermined calculation cycle) Δt elapses at a predetermined timing. To execute.

Therefore, at a predetermined timing, the CPU starts the process from step 1200 of the new target generation routine of FIG. 12 and proceeds to step 1205 to determine whether or not the above-mentioned previous fusion target does not exist. In other words, the CPU determines whether or not the fusion target has been generated or updated at the time of the previous calculation (one calculation cycle Δt before). If the fusion target did not exist last time, the CPU determines "Yes" in step 1205, performs the processes of steps 1210 to 1220 described below in order, and then proceeds to step 1295 to temporarily execute this routine. finish.

Step 1210: The CPU acquires the sensor target (that is, the current detection sensor target) detected by the peripheral radar sensor 16a at the time of the current calculation. Although not shown in the figure, if no sensor target is acquired in step 1210, the CPU directly proceeds to step 1295 and temporarily ends this routine.
Step 1215: The CPU performs the above-mentioned new target generation grouping process to generate a new fusion target.

Step 1220: The CPU calculates the fusion target information "target width Wf, target length Lf, coordinate position (Xf, Yf) of the fusion target, vertical relative velocity Vxf, etc.". In this example, the coordinate position (Xf, Yf) of the fusion target is the coordinate (Xfc, Yfc) of the center position of the fusion target. The vertical relative velocity Vxf of the fusion target can be obtained, for example, by calculating the average value of the vertical relative velocity Vxobj of the sensor target belonging to the fusion target. At this time, the CPU adds identification information (ID), which is one of the fusion target information, to the newly generated fusion target information.

If the fusion target exists last time at the time of the process of step 1205, the CPU determines "No" in step 1205, proceeds directly to step 1295, and temporarily ends this routine.

Further, at a predetermined timing, the CPU starts processing from step 1300 of the existing fusion target tracking routine shown in FIG. 13 and proceeds to step 1310 to determine whether or not the previous fusion target exists. .. In other words, the CPU determines whether or not the fusion target has been generated or updated at the time of the previous calculation (one calculation cycle Δt before). If the fusion target does not exist last time, the CPU determines "No" in step 1310, proceeds directly to step 1395, and temporarily ends this routine.

If the fusion target exists last time, the CPU determines "Yes" in step 1310 and proceeds to step 1315, and the sensor target detected by the peripheral radar sensor 16a at the time of this calculation (that is, this time). Detect sensor target) is acquired.

After that, the CPU proceeds to step 1320 and generates an estimated target based on the previous fusion target information according to the method described above. At this time, the identification information of the estimated target is set to be the same as the identification information of the previous fusion target information from which the estimated target was generated.

After that, the CPU proceeds to step 1325 and performs the first grouping process described above based on the estimated target generated in step 1320. That is, the CPU satisfies the above grouping conditions with respect to the estimated target so as to associate (associate) the sensor target acquired in step 1315 with the previous fusion target (this time detection sensor target). By integrating each other, the fusion target will be updated (generated) this time.

After that, the CPU proceeds to step 1330 and determines whether or not there is a sensor target determined to be unable to be integrated with the estimated target among the sensor targets acquired in step 1315.

If there is a sensor target that is determined not to be integrated into the estimated target, the CPU determines "Yes" in step 1330, proceeds to step 1335, and sets the value of the second grouping flag Xg2 to "1". Set. The CPU then proceeds to step 1340. The value of the second grouping flag Xg2 is set to "0" in an initial routine (not shown) executed when the ignition key switch of the own vehicle SV is changed from off to on. ..

On the other hand, if there is no sensor target determined to be unable to be integrated into the estimated target, the CPU determines "No" in step 1330 and proceeds directly to step 1340.

When the CPU proceeds to step 1340, the target information update process and the lost determination process are performed by performing the routine processing shown in FIG. More specifically, when the CPU proceeds to step 1340, it proceeds to step 1405 via step 1400 of FIG. 14 and selects an arbitrary estimated target. Next, the CPU proceeds to step 1410, and in the first grouping process of step 1325, whether or not the number of "sensor targets determined to be integrated into the estimated target" (grouping target) is "1" or more. To judge.

When the grouping characteristic is "1" or more, the CPU determines "Yes" in step 1410. Then, the CPU proceeds to step 1415, and based on the sensor target information of the sensor target integrated so as to form the fusion target, the fusion target information of the fusion target, "target width Wf, target". The length Lf, the coordinates of the center position (Xfc, Yfc), the vertical relative velocity Vxf, etc. ”are calculated. Further, the CPU sets the value of the number of continuous extrapolated frames f included in the fusion target information to “0”. The value of the number of continuous extrapolated frames f will be described later.

When the processing of step 1415 is completed, the CPU proceeds to step 1420 and updates the fusion target information with the fusion target information calculated in step 1415. Next, the CPU uses a look-up table in which the magnitude (absolute value) of the vertical relative velocity Vxf in the fusion target information of the fusion target updated in step 1420 is shown in the block BK1. It is calculated by applying it to M1 (hereinafter, also referred to as "map"). The CPU calculates the current existence probability Trst by applying the previous existence probability Trst _pre , the increase rate rup of the existence probability, and the number of calculation cycles Cy to the equation (A).
Trst = Trst _pre + rup x Cy ... (A)

The CPU then proceeds to step 1425 to determine whether or not all of the estimated targets at the time of this calculation have been selected. If all of the estimated targets are not selected, the CPU determines "No" in step 1425, returns to step 1405, and selects an unselected estimated target. On the other hand, when all of the estimated targets are selected, the CPU determines "Yes" in step 1425, and proceeds to step 1395 in FIG. 13 via step 1495.

On the other hand, when the number of sensor targets (grouping target) determined to be integrated with the estimated target selected in step 1405 at the time when the CPU performs the process of step 1410 is "0". The above-mentioned extrapolation process of the fusion character is performed. That is, in this case, the CPU determines "No" in step 1410, proceeds to step 1425, and "is the source of the estimated target whose number of sensor targets that can be selected and integrated in step 1405 is" 0 ". It is determined whether or not the "previous fusion target that became" is already extrapolated. Actually, the CPU performs this determination by determining whether or not the value of the number of continuous extrapolated frames f with respect to the previous fusion target is "0".

The value of the number of continuous extrapolated frames f is set to "0" in step 1415 described above when the fusion target was not extrapolated last time. Therefore, when the CPU performs the process of step 1425, if the value of the number of continuous extrapolation frames f is "0", it can be determined that the previous fusion target is not "already extrapolated". In this case, the CPU determines "Yes" in step 1425, proceeds to step 1430, and determines the maximum extrapolation duration tg, which is the maximum value of the time for continuing the extrapolation processing of the fusion target. Specifically, the CPU applies the magnitude (absolute value) of the existence probability Trst _pre of the previous fusion target and the vertical relative velocity of the estimated target to the map M2 shown in the block BK2, thereby maximizing extrapolation. The maximum extrapolation duration tg is determined by calculating the duration tg.

After that, the CPU proceeds to step 1435, executes the extrapolation process of the fusion target described above, and uses the fusion target information (this time fusion target information) of the fusion target at the current calculation timing as the estimated target target. Update to information. That is, this time, the fusion target information is replaced with the target information of the estimated target. After that, the CPU proceeds to step 1440 and increments the value of the number of continuous extrapolated frames f of the fusion target by "+1". In the following, a fusion target having a continuous extrapolation frame number f of 1 or more after the processing of step 1435 is referred to as a “fusion target during extrapolation”.

On the other hand, at the time when the CPU performs the process of step 1425, the "previous fusion target which was the source of the estimated target whose number of sensor targets selected and integrated in step 1405 was" 0 "" is ". In the case of "a fusion target during extrapolation" (that is, when the number of continuous extrapolated frames f is "1" or more), the CPU determines "No" in step 1425. Then, the CPU proceeds to step 1445 and calculates the residual extrapolation time tg'by subtracting the extrapolation duration (calculation cycle Δt × number of continuous extrapolation frames f) from the maximum extrapolation duration tg.

After that, the CPU proceeds to step 1450 and determines whether or not the residual extrapolation time tg'calculated in step 1445 is 0 or less.

If the residual extrapolation time tg'is greater than 0, the CPU determines "No" in step 1450 and proceeds to step 1435 to perform extrapolation processing of the fusion target. In this way, the "fusion target during extrapolation" is repeated every time the calculation cycle Δt elapses, unless the number of sensor targets that can be grouped with respect to the estimated target in step 1410 is 1 or more. Will be After that, the CPU updates the value of the number of continuous extrapolated frames f in step 1440, and proceeds to step 1425.

On the other hand, when the extrapolation process of the fusion target is repeatedly performed and the residual extrapolation time tg'is 0 or less, the CPU determines "Yes" in step 1450 and proceeds to step 1455. It is determined that the "fusion target during extrapolation" has been lost. That is, the CPU determines that the fusion target during extrapolation has disappeared. At this time, the CPU sets the value of the number of continuous extrapolated frames f to "0".

By the way, as described above, when there is a sensor target that is determined not to be integrated with the estimated target, the value of the second grouping flag Xg2 is set to "1" in step 1335 of FIG. When the value of the second grouping flag Xg2 is set to "1", the second grouping routine shown in FIG. 15 is executed as described below, and as a result, the sensor object determined to be unable to be integrated into the estimated target. A new fusion target is generated based on the target.

At a predetermined timing, the CPU starts processing from step 1500 of the routine of FIG. 15, proceeds to step 1510, and determines whether or not the value of the second grouping flag Xg2 is "1".

When the value of the second grouping flag Xg2 is "1", the CPU determines "Yes" in step 1510, performs the processes of steps 1515 to 1525 described below in order, and then proceeds to step 1595. Terminate this routine once.

Step 1515: The CPU executes the second grouping process described above.
Step 1520: The CPU determines the fusion target information of the new fusion target generated in step 1515.
Step 1525: Set the value of the second grouping flag Xg2 to "0".

If the value of the second grouping flag Xg2 is "0" at the time of processing in step 1510, the CPU determines "No" in step 1510, proceeds directly to step 1595, and temporarily ends this routine.

<Modification example>
Although the embodiments of the present invention have been specifically described above, the present invention is not limited to the above-described embodiments, and various modifications based on the technical idea of the present invention are possible.

For example, in the above-described embodiment, it is a premise for executing the lane change support control that the following inter-vehicle distance control and the lane keeping control are being executed, but such a premise is not always necessary. do not do.

10 ... Driving support ECU, 15 ... Vehicle speed sensor, 16a ... Peripheral radar sensor, 16FC ... Center front peripheral sensor, 16FR ... Right front peripheral sensor, 16FL ... Left front peripheral sensor, 16RR ... Right rear peripheral sensor, 16RL ... Left rear peripheral Sensor, 16b ... Camera sensor, 17 ... Operation switch, 52 ... Steering motor, 53 ... Winker lever switch

Claims (1)

Each detects the reflection point of the radar wave transmitted around the own vehicle by a three-dimensional object as a sensor target, and the sensor target including the vertical distance, the lateral position, and the relative speed of the detected sensor target with respect to the own vehicle. Multiple radar sensors that detect position and velocity information to acquire information,
Every time a predetermined time elapses, a fusion target indicating a three-dimensional object existing around the own vehicle is generated based on the sensor target, and the position and speed of the generated fusion target with respect to the own vehicle are determined. Fusion target generation means calculated based on sensor target information,
Existence probability calculation means for calculating the existence probability of the fusion target, and
When the sensor target corresponding to the specific fusion target, which is the fusion target generated before the predetermined time, is no longer detected, the sensor target corresponding to the specific fusion target is not detected. An extrapolation processing means that performs extrapolation processing to generate an extrapolation fusion target corresponding to the specific fusion target based on the position and speed of the specific fusion target with respect to the own vehicle.
When the extrapolation process is performed for the maximum extrapolation duration or longer, the sensor target presence determination means for determining that the sensor target for which the extrapolation process has been performed does not exist, and
It is provided with a maximum extrapolation duration calculation means for calculating the maximum extrapolation duration, which is the maximum value of the time for performing the extrapolation process based on the existence probability and the vertical relative velocity of the specific fusion target.
The existence probability calculation means
The existence probability on the basis of the longitudinal relative velocity of the fusion object to longitudinal relative velocity magnitude is larger the increase rate of the existence probability of the fusion object is increased when the fusion objects is generated The increase rate of the existence probability is obtained, and the increase amount of the existence probability determined based on the value corresponding to the time during which the fusion target is continuously generated and the increase rate of the existence probability is integrated. It is configured to calculate the probability of existence,
The maximum extrapolation duration calculation means is
The maximum extrapolation duration is calculated so that the maximum extrapolation duration becomes longer as the existence probability increases and the maximum extrapolation duration becomes longer as the magnitude of the longitudinal relative velocity decreases. Consists of,
Target detection device.

Images