JP7742802B2 - Inter-vehicle distance estimation device - Google Patents

Description

The present invention relates to an inter-vehicle distance estimation device that uses an on-board camera to estimate the inter-vehicle distance from a vehicle to a preceding vehicle.

In recent years, there has been an increase in efforts to provide access to sustainable transportation systems that take into consideration vulnerable transport users. To achieve this, research and development is being conducted into autonomous driving technology and driver assistance technologies to further improve traffic safety and convenience.

Autonomous driving technology that follows a preceding vehicle and driving assistance technology that prevents a rear-end collision with a preceding vehicle require the vehicle to know the distance between itself and the preceding vehicle. It is well known that radar, such as millimeter-wave radar, is used to determine the distance between vehicles. However, there is a demand for improving the robustness of inter-vehicle distance measurement methods, i.e., making it possible to determine the distance between vehicles even in the event of radar failure, and for cost reduction, making it possible to determine the distance between vehicles using devices that are cheaper than millimeter-wave radar or devices that can be used for multiple purposes. From this perspective, various methods for measuring inter-vehicle distance using cameras have been proposed.

For example, Patent Document 1 describes detecting the distance between vehicles using cameras installed on the road. Patent Document 2 describes detecting the distance between vehicles using a pair of on-board cameras based on the principle of triangulation. Patent Document 3 describes calculating the position of the license plate of the leading vehicle, the ratio of the vehicle's width to its height, and the ratio of the vehicle area of the leading vehicle to the area of its license plate based on images captured by on-board cameras, and estimating the distance between vehicles from these.

Japanese Utility Model Application Laid-Open Publication No. 05-020198 Japanese Patent Application Publication No. 08-334326 Japanese Patent Application Laid-Open No. 2021-196789

However, the method described in Patent Document 1 required a camera installed on the road rather than on the vehicle. The method described in Patent Document 2 required two cameras, which increased costs. The method described in Patent Document 3 sometimes prevented the on-board camera from obtaining the information necessary to estimate the inter-vehicle distance due to dirt on the license plate of the leading vehicle, etc.

In light of the above background, the present invention aims to provide an inter-vehicle distance estimation device that can estimate the inter-vehicle distance from a vehicle to a preceding vehicle with a certain degree of accuracy, at a relatively low cost, and with few cases where estimation is not possible, thereby contributing to the development of sustainable transportation systems.

In order to solve the above-mentioned problems, one aspect of the present invention is a vehicle-to-vehicle distance estimation device (1) that estimates a vehicle-to-vehicle distance (X) from a host vehicle (6) to a preceding vehicle (7), the device comprising: an on-board camera (3) including an image sensor (2) that captures an _image ahead of the host vehicle; and a control device (5) that acquires an estimated vehicle width (W _OBJ ) of the preceding vehicle based on information acquired by the on-board camera, and estimates the vehicle-to-vehicle distance (X) by multiplying the value obtained by dividing the estimated vehicle width (W _{OBJ ) by an imaging plane vehicle width (w obj} ), which is the width of the preceding vehicle on an imaging plane (8) of the image sensor, by a camera focal length (f). The control device (i) calculates the vehicle-to-vehicle distance (X) by dividing the height (H) from the road surface to the on-board camera by the height (v) from the bottom end of the preceding vehicle on the imaging plane to a vanishing point (vp) of the preceding vehicle _. (ii) a second individual estimated vehicle width which is a value obtained by selecting the type vehicle width (W _CLASS ) corresponding to the vehicle type of the preceding vehicle determined from the imaging plane from the type vehicle widths (W _CLASS ) for each vehicle type stored in the control device; (iii) a third individual estimated vehicle width which is a value obtained by dividing the imaging plane vehicle width (w _obj ) by the imaging plane lane width (w _lane ), which is the lane width at the position of the preceding vehicle on the imaging plane, and multiplying the result by an estimated lane width (W _LANE ), which is an estimate of the lane width in which the host vehicle is traveling, obtained from information from a lane width recognition device (4) provided on the host vehicle; and (iv) a third individual estimated vehicle width which is obtained by dividing the imaging plane vehicle width (w _obj ) by the imaging plane license plate width (w _np ) is multiplied by the prescribed license plate width (W _NP ) which is the prescribed width of the license plate stored in the control device to obtain individual estimated vehicle widths including at least two of the fourth individual estimated vehicle widths, and the estimated vehicle width is calculated as a weighted average value obtained by multiplying each of the individual estimated vehicle widths by a weighting coefficient (C _i ) whose sum is 1 and adding them up.

According to this aspect, the inter-vehicle distance (X) can be estimated using a monocular camera, allowing the device to be constructed relatively inexpensively. Furthermore, since the inter-vehicle distance (X) can be calculated if any of multiple individual estimated vehicle widths can be calculated, the inter-vehicle distance (X) is prevented from becoming inescapable of being estimated, and the inter-vehicle distance (X) can be estimated with a predetermined accuracy or higher.

In the above aspect, the individual estimated vehicle width may include the first to fourth individual estimated vehicle widths.

According to this aspect, since four individual estimated vehicle widths are used, it is possible to prevent the inter-vehicle distance (X) from becoming impossible to estimate, and since the estimated vehicle width (W _OBJ ) is calculated as a weighted average of these, the accuracy of the estimated inter-vehicle distance (X) is improved.

In the above aspect, if any of the first individual estimated vehicle width, the third individual estimated vehicle width, and the fourth individual estimated vehicle width cannot be calculated, the control device (5) may exclude the incalculable first individual estimated vehicle width, the third individual estimated vehicle width, and/or the fourth individual estimated vehicle width from the targets for calculating the estimated vehicle width, and set the corresponding weighting coefficients (C1, C3, C4) to 0.

According to this aspect, since it is often possible to calculate the second individual estimated vehicle width based on the vehicle type, it is possible to estimate the inter-vehicle distance (X) with a certain degree of accuracy without having to calculate other individual estimated vehicle widths.

In the above aspect, the control device (5) calculates a lane width reliability that quantifies the accuracy of the calculated estimated lane width (W _LANE ), and if the lane width reliability is equal to or greater than a first threshold, the weighting coefficient (C ₃ ) of the third individual estimated vehicle width may be set to a value greater than the weighting coefficients (C ₁ , C ₂ , C ₄ ) of the first individual estimated vehicle width, the second individual estimated vehicle width, and the fourth individual estimated vehicle width, and if the lane width reliability is less than the first threshold, the weighting coefficient (C ₃ ) of the third individual estimated vehicle width may be set to a value smaller than at least one of the weighting coefficients (C ₁ , C ₂ , C ₄ ) of the first individual estimated vehicle width, the second individual estimated vehicle width, and the fourth individual estimated vehicle width.

When the reliability of the estimated lane width (W _LANE ) is high, it can be estimated that the accuracy of the third individual estimated vehicle width is high. According to this aspect, if the lane width reliability is equal to or greater than a predetermined threshold, the weighting of the third individual estimated vehicle width is increased, thereby improving the accuracy of the estimated vehicle width (W _OBJ ) and the inter-vehicle distance (X).

In the above aspect, when the lane width reliability is lower than the first threshold value, or when the height (v) from the bottom end of the preceding vehicle on the imaging surface (8) to the vanishing point (vp) cannot be measured, the control device (5) calculates an imaging surface license plate width reliability that quantifies the accuracy of the calculated imaging surface license plate width ( _wnp ), and when the imaging surface license plate width reliability is equal to or greater than a second threshold value, the weighting coefficient ( _C4 ) of the fourth individual estimated vehicle width may be set to a value greater than the weighting coefficients ( _C2 , _C3 ) of the second individual estimated vehicle width and the third individual estimated vehicle width, and when the imaging surface license plate width reliability is less than the second threshold value, the weighting coefficient ( _C4 ) of the fourth individual estimated vehicle width may be set to a value smaller than at least one of the weighting coefficients ( _C1 , _C2 ) of the first individual estimated vehicle width and the second individual estimated vehicle width.

If the elements serving as the basis for the calculation can be obtained satisfactorily, the fourth individual estimated vehicle width calculated based on the width of the license plate is less accurate than the third individual estimated vehicle width calculated based on the lane width, but is more accurate than the second estimated vehicle width based on the vehicle type. Therefore, according to this aspect, when the third individual estimated vehicle width cannot be calculated or is unreliable, the weighting of the fourth individual estimated vehicle width is increased, thereby improving the accuracy of the estimated vehicle width (W _OBJ ) and the inter-vehicle distance (X).

In the above aspect, the estimated vehicle width (W _OBJ ) is preferably calculated using Equation 1 described later. Also, the control device (5) is preferably configured to set the weighting coefficients (C _i ) based on Table 1 described later.

These aspects further improve the accuracy of the estimated inter-vehicle distance (X).

The above aspects make it possible to provide a vehicle-to-vehicle distance estimation device that can estimate the vehicle-to-vehicle distance from the subject vehicle to a preceding vehicle with a predetermined degree of accuracy, with few cases where estimation is impossible, and at a relatively low cost.

1 is a diagram illustrating a configuration of an inter-vehicle distance estimation device according to an embodiment; FIG. 1 is a diagram showing an actual vehicle for explaining a method for estimating a distance between vehicles according to an embodiment; FIG. 1 is a diagram showing an imaging plane for explaining a method for estimating a distance between vehicles according to an embodiment; FIG. 1 is a diagram showing an imaging plane for explaining a method for estimating a distance between vehicles according to an embodiment; Flow diagram of a method for estimating a distance between vehicles according to an embodiment

The following describes the inter-vehicle distance estimation device 1 according to the embodiment, with reference to the drawings.

As shown in Figure 1, inter-vehicle distance estimation device 1 comprises an on-board camera 3 including an image sensor 2, a lane width recognition device 4, and a control device 5 that receives information from the on-board camera 3 and the lane width recognition device 4 and performs calculations. As shown in Figure 2, inter-vehicle distance estimation device 1 is mounted on host vehicle 6 and estimates inter-vehicle distance X from host vehicle 6 to a preceding vehicle 7. Hereinafter, "inter-vehicle distance X" refers to the distance from the on-board camera 3 of host vehicle 6 to the rear end of preceding vehicle 7. To find the distance from the front end of host vehicle 6 to the rear end of preceding vehicle 7, simply calculate the difference between "inter-vehicle distance X" and the distance from the on-board camera 3 to the front end of host vehicle 6.

As shown in Figures 1 and 2, the vehicle-mounted camera 3 is installed inside the cabin of the vehicle 6, captures images of the scenery ahead of the vehicle 6, and transmits the captured image data on the imaging plane 8 of the imaging element 2 to the control device 5. The imaging plane 8 shown in Figure 2 is a virtual plane. The inter-vehicle distance estimation device 1 is configured to be able to estimate the inter-vehicle distance X even if the vehicle-mounted camera 3 is a monocular camera, but multiple vehicle-mounted cameras 3 may also be installed.

The lane width recognition device 4 is configured, for example, by an on-board camera 3 or a navigation device 9 including a GPS receiver 10 and a map information storage unit 11, and transmits the acquired information to the control device 5. The navigation device 9 acquires the vehicle's current position and provides route guidance to the destination. The GPS receiver 10 identifies the vehicle's position (latitude and longitude) based on signals received from artificial satellites (positioning satellites). The map information storage unit 11 is configured by a well-known storage device such as flash memory or a hard disk, and stores map information. The map information includes information on lane width or road standards from which lane width can be derived.

The control device 5 is an electronic control unit (ECU) equipped with a CPU 12 and a storage unit 13 including non-volatile memory (ROM) and volatile memory (RAM). The control device 5 performs various processes by having the CPU 12 execute calculations in accordance with programs stored in the storage unit 13. The control device 5 may be configured as a single piece of hardware, or as a unit consisting of multiple pieces of hardware. Furthermore, at least some of the functional units of the control device 5 may be realized by hardware such as an LSI, ASIC, or FPGA, or may be realized by a combination of software and hardware. The control device 5 can be wirelessly connected to an external communication network.

The memory unit 13 of the control device 5 stores the focal length f of the on-board camera 3, the height H from the road surface to the on-board camera 3, the width of each general vehicle type (vehicle type width W _CLASS ), and the width of the license plate specified in the country or region of use (specified license plate width W _NP ). The vehicle type width W _CLASS is set to, for example, 1.8 m for passenger cars and 2.5 m for trucks.

3 and 4 show the leading vehicle 7 on the imaging plane 8. As shown in FIGS. 1, 3, and 4, the control device 5 acquires, based on image information received from the on-board camera 3, the height v on the imaging plane 8 from the bottom edge of the leading vehicle 7 to the vanishing point vp, the vehicle width of the leading vehicle 7 (imaging plane vehicle width w _obj ), the lane width at the position of the leading vehicle 7 (imaging plane lane width w _lane ), and the width of the license plate attached to the rear of the leading vehicle 7 (imaging plane license plate width w _np ). The vanishing point vp can be understood, for example, as the point where the extensions of the approximate straight lines of the two dividing lines 14 that define the lane on the imaging plane 8 intersect. Note that with regard to the symbols "W" and "w" indicating widths in FIGS. 2 to 4, the capital letters indicate the actual width, and the lowercase letters indicate the width on the imaging plane.

See Figures 1 and 2. The control device 5 determines the vehicle type of the preceding vehicle 7 from the image features captured by the on-board camera 3 using classification using image recognition by a DNN (deep neural network) that applies deep learning. Vehicle types are classified into, for example, passenger cars, trucks, motorcycles, etc.

The control device 5 also obtains an estimated lane width W _LANE , which is an estimate of the lane width in which the vehicle 6 is traveling, based on information received from the lane width recognition device 4. When the on-board camera 3 also serves as the lane width recognition device 4, the control device 5 estimates the lane width W LANE based on the portion of the image data received from the on-board camera 3 closest to the vehicle 6, i.e., the width between two lane markings 14 defining the _lane at the bottom of the imaging plane 8, and information about the on-board camera 3, such as the height H, orientation, and focal length f of the on-board camera 3. Note that instead of the on-board camera 3 capturing images in front of the vehicle 6, other on-board cameras (not shown) capturing images of the sides or rear of the vehicle 6 may be provided and used as the lane width recognition device 4. When a navigation device 9 is used as the lane width recognition device 4, the control device 5 obtains the lane width W _LANE by comparing the vehicle position obtained from the GPS receiver 10 with map information stored in the map information storage unit 11.

The unit of length of the actual object is the legal or customary unit of length in the country or region of use, for example, meters. The unit of length on the imaging plane may be any common unit, for example, it may be the same as the unit of length of the actual object, or it may be the number of pixels on the imaging plane 8.

5 shows the flow of how the control device 5 estimates the inter-vehicle distance X. As shown in FIGS. 1 to 5, the control device 5 calculates the estimated vehicle width W _OBJ of the leading vehicle 7 using Equation 1 based on the information received from the on-board camera 3 and the lane width recognition device 4, and estimates the inter-vehicle distance X from the estimated vehicle width W _OBJ using Equation 2 .
In equation 1 , W _OBJ is the estimated vehicle width, C ₁ to C ₄ are weighting coefficients, the sum of C ₁ to C ₄ is 1, w _obj is the vehicle width at the imaging plane, H is the height from the road surface to the onboard camera 3, v is the height from the bottom of the preceding vehicle 7 on the imaging plane 8 to the vanishing point vp, W _CLASS is the vehicle width by type, W _LANE is the estimated lane width, w _lane is the lane width at the imaging plane, W _NP is the specified license plate width, and w _np is the license plate width at the imaging plane.
In Equation 2 , X is the inter-vehicle distance, f is the camera focal length, W _obj is the estimated vehicle width, and w _obj is the vehicle width at the imaging plane.

The method for estimating inter-vehicle distance X will be explained below, following the steps shown in Figure 5.

The control device 5 calculates first to fourth individual estimated vehicle widths using mutually different elements (ST1). In the terms on the right side of Equation 1 , the items multiplied by C1 _{to C4} correspond to the first to fourth individual estimated vehicle widths. Hereinafter, _C1 to _C4 will be referred to as the first to fourth weighting coefficients.

2 and 3, the first individual estimated vehicle width is obtained by utilizing the fact that the ratio of the vehicle width of the leading vehicle 7 to the imaging plane vehicle width w _obj is equal to the ratio of the height H from the road surface to the on-board camera 3 to the height v from the bottom of the leading vehicle 7 on the imaging plane to the vanishing point vp. The first individual estimated vehicle width is the value obtained by dividing the height H from the road surface to the on-board camera 3 by the height v from the bottom of the leading vehicle 7 on the imaging plane to the vanishing point vp, and multiplying this value by the imaging plane vehicle width w _obj .

1 to 5, the second individual estimated vehicle width is a value selected from the vehicle type W _CLASS for each vehicle type stored in the memory unit 13 of the control device 5, and is a value corresponding to the vehicle type of the preceding vehicle 7 determined by the control device ₅ from the imaging plane 8. The second individual estimated vehicle width estimates a representative value for each vehicle type as the vehicle width of the preceding vehicle 7 belonging to that vehicle type, and is considered to have a larger error than the first, third, and fourth individual estimated vehicle widths.

The third individual estimated vehicle width is obtained by utilizing the fact that the ratio of the vehicle width of the leading vehicle 7 to the imaging plane vehicle width w _obj is equal to the ratio of the estimated lane width W _LANE to the imaging plane lane width w _lane . The third individual estimated vehicle width is the value obtained by dividing the imaging plane vehicle width w _obj by the imaging plane lane width w _lane and multiplying the result by the estimated lane width W _LANE .

The fourth individual estimated vehicle width is obtained by utilizing the fact that the ratio of the vehicle width of the leading vehicle 7 to the imaging plane vehicle width w _obj is equal to the ratio of the estimated lane width W _LANE to the imaging plane lane width w _lane . The fourth individual estimated vehicle width is the value obtained by dividing the imaging plane vehicle width w _obj by the imaging plane license plate width w _np and multiplying the result by the specified license plate width W _NP stored in the memory unit 13 of the control device 5.

If there is an element that cannot be acquired, the control device 5 will exclude from the calculation the individual estimated vehicle width that uses that element as the basis for calculation.

Next, the control device 5 sets the first to fourth weighting coefficients _C1 to _C4 (ST2). Table 1 shows the setting rules for the weighting coefficient _C1 . In Table 1, the second to sixth columns show the judgment conditions (first judgment, second judgment), and the seventh column shows the setting rules for the first to fourth weighting coefficients _C1 to _C4 when the judgment condition in that row is met. Table 1 shows the cases where the second individual estimated vehicle width based on the vehicle type can be calculated.

Assuming that all elements for calculating the first to fourth individual estimated vehicle widths can be obtained under favorable conditions, the second individual estimated vehicle width is not a value based on an individual vehicle but a representative value for each vehicle type, and therefore may contain larger errors than the other individual estimated vehicle widths. Furthermore, the vanishing point vp used to calculate the first individual estimated vehicle width may have errors in its setting position. On the other hand, the elements on the imaging plane 8 used to calculate the third individual estimated vehicle width are shown directly on the image. Furthermore, because the license plate width for the fourth individual estimated vehicle width is smaller than the lane width for the third individual estimated vehicle width, the ratio of error to the true value is likely to be larger for the measured value of the license plate width than for the measured value of the lane width. Therefore, the first and fourth individual estimated vehicle widths may contain larger errors than the third individual estimated vehicle width. Therefore, the basic weighting rule is, in descending order of value, the third weighting coefficient _C3 , the first and fourth weighting coefficients _C1 and _C4 , and the second weighting coefficient _C2 (Case 1 in Table 1). However, in reality, there may be factors that cannot be measured or that are measurable but unreliable, so the control device 5 changes the basic rule as necessary and sets the weighting coefficient _Ci each time it estimates the inter-vehicle distance X.

As shown in Table 1, the control device 5 first makes a first determination based on whether or not each element has been measured. If these elements cannot be measured, the individual estimated vehicle width corresponding to that element cannot be calculated, and the value of the corresponding weighting coefficients _C1 to _C4 becomes 0.

As shown in the second and seventh columns of Table 1, if the imaging plane lane width w _lane cannot be measured, the third individual estimated vehicle width based on the lane width cannot be calculated, and therefore the third weighting coefficient _C3 becomes 0. An example of a case in which the imaging plane lane width w _lane cannot be measured is when the dividing line 14 is thin and the difference in brightness between the dividing line 14 and the surrounding road surface is small.

Furthermore, as shown in the third and seventh columns of Table 1, if the imaging plane license plate width _wnp cannot be measured, the fourth individual estimated vehicle width based on the license plate width cannot be calculated, and the fourth weighting coefficient _C4 will be 0. An example of a case in which the imaging plane license plate width _wnp cannot be measured is when the leading vehicle 7 is located too far away for its license plate to be recognized.

Furthermore, as shown in the fourth and seventh columns of Table 1, if the height v from the bottom end of the leading vehicle 7 on the imaging plane 8 to the vanishing point vp cannot be measured, the first individual estimated vehicle width based on the height v cannot be calculated, and therefore the first weighting coefficient _C1 becomes 0. An example of a case in which the height v cannot be measured is when the leading vehicle 7 is located close to the host vehicle 6 and the bottom end of the leading vehicle 7 does not move onto the imaging plane 8.

Next, the control device 5 performs a second determination based on the reliability and adjusts the weighting coefficient C _i . If the imaging plane lane width w _lane can be measured, the control device 5 calculates a lane width reliability that quantifies the accuracy of the estimated lane width W _LANE and determines whether the calculated value is equal to or greater than a predetermined first threshold (see column 5 of Table 1). The lane width reliability is calculated so that the higher the accuracy, the larger the calculated value. The lane width reliability is high, for example, when the estimated lane width W _LANE is acquired many times within a predetermined time, when there is a large difference in brightness between the dividing line 14 and the surrounding road surface, or when there is little variation in the estimated lane width W _LANE values acquired multiple times. If the lane width reliability is equal to or greater than the first threshold, the control device 5 prioritizes the third individual estimated vehicle width, which is considered to have a small error, and sets the third weighting coefficient C ₃ to a value greater than the other weighting coefficients C ₁ , C ₂ , and C ₄ (see cases 1 and 6 of Table 1). On the other hand, if the lane width reliability is less than the first threshold, the control device 5 sets the third weighting factor _C3 to a value smaller than at least some of the other weighting factors _C1 , _C2 , and _C4 (Cases 2, 3, and 7 in Table 1). Note that Table 1 assumes that the on-board camera 3 is used as the lane width recognition device 4, and that when the leading vehicle 7 is too close to the vehicle 6 to measure the height v, the dividing line 14 is not reflected below the imaging surface 8, and the estimated lane width _WLANE cannot be measured (Cases 4, 5, and 8).

If the lane width reliability is less than a predetermined value or if the height v cannot be measured, the control device 5 calculates the license plate width reliability, which quantifies the accuracy of the license plate width _wnp on the imaging plane, and determines whether the calculated value is equal to or greater than a predetermined second threshold (column 6 of Table 1). The license plate width reliability is calculated so that the higher the accuracy, the larger the value. For example, the larger the size of the license plate on the imaging plane 8, the higher the license plate width reliability. If the license plate width reliability is equal to or greater than the second threshold, the control device 5 sets the fourth weighting factor _C4 to a value greater than at least the second and third weighting factors C2 and _C3 (cases ₂ , 4, and 9 of Table 1). If the license plate width reliability is less than the second threshold, the control device 5 sets the fourth weighting factor _C4 to a value smaller than at least one of the first and second weighting factors _C1 and _C2 (cases 3, 5, and 10 of Table 1).

Next, the control device 5 multiplies the corresponding individual estimated vehicle widths by the set weighting coefficients _Ci , adds up the resulting values, and calculates the estimated vehicle width W _OBJ as the weighted average of the individual estimated vehicle widths (ST3, Equation 1 ).

Next, the control device 5 estimates the inter-vehicle distance X from the estimated vehicle width W _OBJ (ST4). As shown in Fig. 2, the ratio of the inter-vehicle distance X to the camera focal length f is equal to the ratio of the estimated vehicle width W _OBJ to the vehicle width at the imaging plane w _OBJ . Therefore, the inter-vehicle distance X is calculated by dividing the estimated vehicle width W _OBJ by the vehicle width at the imaging plane w _OBJ and multiplying the result by the camera focal length f ( Equation 2 ).

The following describes the effects of the inter-vehicle distance estimation device 1 according to this embodiment.

The inter-vehicle distance estimation device 1 can estimate the inter-vehicle distance X using a monocular camera.

Furthermore, since the estimated vehicle width W _OBJ is calculated as a weighted average of a plurality of individual estimated vehicle widths and the inter-vehicle distance X is estimated based on this estimated vehicle width W _OBJ , the inter-vehicle distance X can be estimated even if some individual estimated vehicle widths cannot be calculated. In particular, it can be said that the second individual estimated vehicle width based on the vehicle type can be calculated in many cases, which prevents the inter-vehicle distance X from being unable to be estimated. Furthermore, the inter-vehicle distance X can be calculated with a certain degree of accuracy without having to use an individual estimated vehicle width with low reliability.

The weighting coefficient C _i increases the influence of the individual estimated vehicle widths that are estimated to be highly accurate on the estimated vehicle width W _OBJ , thereby improving the accuracy of the estimated vehicle width W _OBJ .

This concludes the explanation of specific embodiments, but the present invention is not limited to the above embodiments and variations, and can be implemented in a wide variety of ways. The individual estimated vehicle widths are not limited to the four mentioned above, and may include those calculated based on other factors. Also, it is not necessary to include all four of the above, as long as there are two or more.

1: Inter-vehicle distance estimation device 2: Image pickup element 3: On-board camera 4: Lane width recognition device 5: Control device 6: Vehicle 7: Leading vehicle 8: Imaging surface 9: Navigation device 14: Marking line W _OBJ : Estimated vehicle width C ₁ to C ₄ : Weighting coefficients w _obj : Vehicle width at imaging surface H: Height from road surface to on-board camera v: Height from bottom edge of leading vehicle on imaging surface to vanishing point W _CLASS : Vehicle width by type W _LANE : Estimated lane width w _lane : Lane width at imaging surface W _NP : Specified license plate width w _np : License plate width at imaging surface X: Inter-vehicle distance f: Camera focal length

Claims (7)

A vehicle-to-vehicle distance estimation device that estimates a vehicle-to-vehicle distance (X) from a host vehicle to a preceding vehicle,
an in-vehicle camera including an imaging element for capturing an image of a scene ahead of the vehicle;
a control device that acquires an estimated vehicle width (W _OBJ ) of the preceding vehicle based on information acquired by the on-board camera, and estimates the following distance (X) as a value obtained by dividing the estimated vehicle width (W _OBJ ) by an imaging plane vehicle width (w _obj ), which is the width of the preceding vehicle on the imaging plane of the image sensor, and multiplying the result by a camera focal length (f);
The control device
(i) a first individual estimated vehicle width, which is a value obtained by dividing the height (H) from the road surface to the on-board camera by the height (v) from the bottom edge of the leading vehicle on the imaging plane to the vanishing point, and multiplying the result by the vehicle width (w _obj ) on the imaging plane;
(ii) a second individual estimated vehicle width, which is a value selected from the vehicle type widths (W _CLASS ) for each vehicle type stored in the control device, and which corresponds to the vehicle type of the preceding vehicle determined from the imaging _plane ;
(iii) a third individual estimated vehicle width, which is a value obtained by dividing the image plane vehicle width (w _obj ) by the image plane lane width (w _lane ), which is the lane width at the position of the preceding vehicle on the image plane, and multiplying the result by an estimated lane width (W _LANE ), which is an estimate of the lane width in which the host vehicle is traveling, obtained from information from a lane width recognition device provided on the host vehicle;
(iv) obtaining individual estimated vehicle widths including at least two of the fourth individual estimated vehicle widths, which are values obtained by dividing the image plane vehicle width (w _obj ) by the image plane license plate width (w _np ), which is the width of the license plate of the preceding vehicle on the image plane, and multiplying the result by the prescribed license plate width (W _NP ), which is the prescribed width of the license plate stored in the control device;
The inter-vehicle distance estimation device is configured to calculate the estimated vehicle width as a weighted average value obtained by multiplying each of the individual estimated vehicle widths by a weighting coefficient (C _i ) whose sum is 1 and adding up the results. The inter-vehicle distance estimation device of claim 1, wherein the individual estimated vehicle widths include the first to fourth individual estimated vehicle widths. 3. The inter-vehicle distance estimation device according to claim 2, wherein, when any of the first individual estimated vehicle width, the third individual estimated vehicle width, and the fourth individual estimated vehicle width cannot be calculated, the control device excludes the first individual estimated vehicle width, the third individual estimated vehicle width, and/or the fourth individual estimated vehicle width that cannot be calculated from the targets for calculating the estimated vehicle width, and sets the corresponding weighting coefficients ( _C1 , _C3 , _C4 ) to 0. 4. The inter-vehicle distance estimation device according to claim 3, wherein the control device calculates a lane width reliability that quantifies the accuracy of the calculated estimated lane width (W _LANE ), and when the lane width reliability is equal to or greater than a first threshold, sets the weighting coefficient (C ₃ ) of the third individual estimated vehicle width to a value greater than the weighting coefficients (C ₁ , C ₂ , C ₄ ) of the first individual estimated vehicle width, the second individual estimated vehicle width, and the fourth individual estimated vehicle width, and when the lane width reliability is less than the first threshold, sets the weighting coefficient (C ₃ ) of the third individual estimated vehicle width to a value smaller than at least one of the weighting coefficients (C ₁ , C ₂ , C ₄ ) of the first individual estimated vehicle width, the second individual estimated vehicle width, and the fourth individual estimated vehicle width. 5. The inter-vehicle distance estimation device of claim 4, wherein the control device is configured to calculate an imaging plane license plate width reliability that quantifies the accuracy of the calculated imaging plane license plate width (w _np ) when the lane width reliability is lower than the first threshold or when the height (v) from the bottom end of the preceding vehicle on the imaging plane to the vanishing point cannot be measured, and when the imaging plane license plate width reliability is equal to or greater than a second threshold, set the weighting coefficient ( _C4 ) of the fourth individual estimated vehicle width to a value greater than the weighting coefficients ( _C2 , _C3 ) of the second individual estimated vehicle width and the third individual estimated vehicle width, and when the imaging plane license plate width reliability is less than the second threshold, set the weighting coefficient ( _C4 ) of the fourth individual estimated vehicle width to a value smaller than at least one of the weighting coefficients ( _C1 , _C2 ) of the first individual estimated vehicle width and the second individual estimated vehicle width. 6. The inter-vehicle distance estimation device according to claim 2, wherein the estimated vehicle width is calculated based on the following equation 1 :
In the above formula 1 ,
W _OBJ is the estimated vehicle width,
_C1 to _C4 are the weighting coefficients, the sum of _C1 to _C4 is 1, and in the terms on the right side, the items multiplied by _C1 to _C4 correspond to the first to fourth individual estimated vehicle widths,
w _obj is the width of the imaging plane,
H is the height from the road surface to the vehicle-mounted camera,
v is the height from the bottom edge of the leading vehicle on the imaging plane to the vanishing point,
W _CLASS is the vehicle width according to the type,
W _LANE is the estimated lane width;
w _lane is the imaging plane lane width,
W _NP is the prescribed license plate width,
w _np is the image plane license plate width. The inter-vehicle distance estimation device according to claim 6, wherein the control device is configured to set the weighting coefficients ( _C1 , _C2 , _C3 , _C4 ) so as to satisfy the condition shown in the seventh column of Table 1 below when the judgment conditions in the second to sixth columns of Table 1 below are met.