JP5556317B2 - Object recognition device - Google Patents

Description

  The present invention relates to a technique for irradiating a transmission wave over a predetermined range in a vehicle width direction (horizontal direction) and recognizing an object in front of a vehicle based on the reflected wave.

  Conventionally, for example, an object recognition device that recognizes an object in front of a vehicle by irradiating a transmission wave such as a laser beam or a millimeter wave and detecting a reflected wave thereof has been considered. This type of device is applied to, for example, a device that detects an obstacle such as a preceding vehicle and generates an alarm, or a device that controls the vehicle speed so as to maintain a predetermined distance from the preceding vehicle. It is used for object recognition such as a preceding vehicle as a control target.

  At the time of such object recognition, a method is conceived in which a roadside object (for example, planting in the vicinity of the median strip) is not mistakenly recognized as a vehicle existing ahead (see, for example, Patent Document 1). ). In this method, a vehicle / non-vehicle determination level based on the reflection intensity of an object is set for each detection target space, and a preceding vehicle and a non-vehicle are distinguished based on the reflection intensity from the object.

  For example, a corresponding area of distance measurement data is determined using a non-vehicle determination map as shown in FIG. 9, and it is determined whether the distance measurement data is an area corresponding to a vehicle or an area corresponding to a non-vehicle. This non-vehicle determination map corresponds to the existence area of the reflecting object when the vehicle width direction, the vehicle height direction, and the vehicle front direction (vehicle traveling direction) are the X axis, the Y axis, and the Z axis, respectively. It is a three-dimensional map in which a range of received light intensity for distinguishing vehicles is set. Specifically, the XY direction is divided into an area near the center, an area around it, and a lowermost area, and the correspondence between the Z direction position and the received light intensity corresponds to each of these areas. (A) to (c) are set. The region near the center in the XY direction corresponds to the correspondence relationship (b), the surrounding region corresponds to the correspondence relationship (a), and the lowermost region corresponds to the correspondence relationship (c).

JP 2002-131433 A

  However, an object present on the roadside is not necessarily a low reflection object. For example, a delineator or a reflection guide plate attached to a guard rail is a highly reflective object, and there are cases where it is not possible to sufficiently cope with discrimination based on received light intensity. In addition, it does not have high reflectivity like a delineator or a reflection guide plate, and even a guardrail or a wall has a certain reflection strength. May be difficult to attach. Thus, if the final vehicle / non-vehicle distinction depends on the reflection intensity, there is a possibility of erroneous determination.

  In addition, not only vehicle / non-vehicle determination but also a case where the vehicles are distinguished from each other even if the vehicles are distinguished from each other even if they are distinguished from each other. is there.

  In order to solve the above-described problems, an object of the present invention is to provide an object recognition technique for accurately recognizing an object.

  The object recognition apparatus of the present invention made to achieve the above object is a result of performing scanning using a plurality of transmission wave beams mounted on a vehicle and set to be adjacent at least along the vehicle width direction. Based on this, the object that reflected the transmitted wave is recognized.

  Then, the detection means detects the received intensity and the position of the object for each transmission wave beam based on the received signal of the reflected wave from the object obtained by scanning, and the target means detects the detection means. Based on the result, a plurality of detection results are integrated as belonging to the same target if the position of another detection result is also included in the search region set based on the position of a certain detection result.

  The length in the vehicle width direction of the search region used when the target unit integrates a plurality of detection results is set to a length that exactly includes the reflected waves from the two adjacent transmission wave beams, The length of the search area in the vehicle traveling direction is set with reference to the distance traveled by the vehicle during the transmission wave beam scanning period, and the object is determined based on the target generated by the targetization means. recognize.

  By integrating using the search area set in this way, targets that extend long in the direction along the road (vehicle traveling direction) or are aligned in the direction are regarded as the same object. Therefore, a roadside object such as a guardrail and a preceding vehicle can be correctly distinguished. Further, if the object itself is separated in the vehicle width direction, the detection results of these other objects are not integrated as belonging to the same target. That is, not only the preceding vehicle and the roadside object, but also the preceding vehicles can be correctly distinguished and the object can be recognized with high accuracy.

Thus, according to the object recognition device of the present invention, an object can be recognized with high accuracy.
In the case where the traveling path of the vehicle is curved, for example, it is conceivable to set a search area as shown in claim 2. In other words, it is configured to be able to acquire the direction of the travel path at an arbitrary point on the travel path of the vehicle, and the search area of the target means is regarded as the vehicle travel direction in the direction of the travel path at the position of the detection result. Set it.

  In this way, when the traveling road ahead of the vehicle is curved, the search area is set in an appropriate state according to the curve. Therefore, as described above, it is possible to correctly distinguish between the preceding vehicle and the roadside object and between the preceding vehicles, and to recognize the object with high accuracy.

1 is an overall configuration diagram of an object recognition device. The flowchart which shows the content of the object recognition process which a control part performs. The flowchart which shows the content of the integration process in an object recognition process. Explanatory drawing of the positional relationship of an own vehicle and a preceding vehicle and a guardrail, and integration. Explanatory drawing which shows a search area | region. Explanatory drawing which shows the positional relationship of the own vehicle and the preceding vehicle and guardrail in the road where the front is curving. Explanatory drawing which shows the search area | region in the case of a curve road. Explanatory drawing which shows the search area | region in the case of a curve road. Explanatory drawing which shows a prior art.

Embodiments of the present invention will be described below with reference to the drawings.
FIG. 1 is a block diagram showing an overall configuration of an object recognition apparatus 1 to which the present invention is applied.
The object recognition apparatus 1 is an apparatus that is mounted on a vehicle, recognizes an object existing in front of the vehicle, and generates object data including information (position, size, etc.) on the object. The object data generated by the object recognition device 1 is supplied to a vehicle control device (not shown), and generates an alarm when, for example, the object candidate is an obstacle existing in a preset alarm area. When the object candidate is a preceding vehicle, it is used for alarm control or so-called inter-vehicle control for controlling the vehicle speed by operating a brake, a throttle, an automatic transmission or the like according to the situation of the preceding vehicle.

  The vehicle control device obtains the steering angle based on the signal from the steering sensor, calculates the yaw rate based on the signal from the yaw rate sensor, and calculates the curve radius (curvature radius) based on the vehicle speed, the steering angle, and the yaw rate. It is comprised so that can be calculated. Such a configuration is well known and is described in, for example, Japanese Patent Application Laid-Open No. 2002-131433 shown as Patent Document 1.

<Overall configuration>
As shown in FIG. 1, the object recognition apparatus 1 includes a light emitting unit 10 that scans the front of the vehicle with laser light, and a light receiving unit 20 that receives reflected light from an object that reflects the laser light emitted from the light emitting unit 10. A detection circuit 30 for measuring the time from when the light emitting unit 10 irradiates the laser beam to when the reflected light is received by the light receiving unit 20 and outputting distance measurement data; and for driving the light emitting unit 10 A control unit 40 that outputs a signal, inputs distance measurement data from the detection circuit 30, and generates object data related to an object reflecting the laser beam based on these input / output signals.

  Among these, the light emitting unit 10 includes a laser diode (LD) 11 that generates laser light, and an LD drive circuit 12 that generates pulsed laser light serving as a transmission wave in the LD 11 in accordance with an LD drive signal from the control unit 40. The light emitting lens 13 for narrowing the beam width of the laser light generated by the LD 11, the polygon mirror that is rotatably supported and reflects the laser light supplied through the light emitting lens 13, and the polygon mirror is driven to rotate. In accordance with the motor drive signal from the scanner 14 constituted by the motor and the controller 40, the motor constituting the scanner 14 is driven to rotate the polygon mirror, thereby changing the irradiation direction of the laser light to set in advance. And a motor drive circuit 15 that realizes scanning of the laser beam within the set angle range.

  In the scanning with the beam in the vehicle width direction (horizontal beam), specifically, a predetermined angle range in the vehicle width direction centering on the front direction of the vehicle is used as the scanning area, and the scanning area is defined as the prescribed beam width of the horizontal beam. Is set to irradiate the laser beam a predetermined number of times at equal intervals. The horizontal beams are scanned from left to right, and each horizontal beam is distinguished by a beam number.

  On the other hand, the light receiving unit 20 receives the reflected light via the light receiving lens 21 that collects the reflected light from the object that reflects the laser light (horizontal beam), and the voltage value corresponding to the intensity. A light receiving element 22 that generates a light receiving signal, and an amplifier 23 that amplifies the light receiving signal from the light receiving element 22.

  The detection circuit 30 detects the round-trip time of the laser beam based on the LD drive signal from the control unit 40 and the output signal from the amplifier 23, and measures the distance along with the corresponding scanning angle θ (what number LD drive signal). The data is output to the control unit 40 as data.

  The control unit 40 is composed of a known microcomputer configured with a CPU, ROM, RAM, and the like. The ROM stores a program for processing (to be described later) executed by the CPU.

<Processing in the control unit>
The control unit 40 drives the light emitting unit 10 with the LD drive signal and the motor drive signal, performs a scan execution process for executing a scan in the scan area, and the laser beam based on the distance measurement data obtained by the scan. At least an object recognition process for generating object data relating to the reflected object is executed.

  Among these, the scanning execution process is started every preset scanning period (100 ms), outputs an LD drive signal a predetermined number of times at regular intervals, and in synchronization with this, the irradiation direction of the radar light is a predetermined angle. A process of outputting a motor drive signal for operating the scanner 14 so as to be shifted one by one is executed.

On the other hand, the object recognition process is started every time the scanning execution process ends, and executes the process shown in the flowchart of FIG.
That is, in the object recognition process, first, ranging data for one scan is read from the detection circuit 30 in S110.

  In S120, the position represented by polar coordinates is converted into the position represented by orthogonal coordinates by the distance data R obtained from the read distance measurement data and the scanning angle θ. The orthogonal coordinates are XY orthogonal coordinates in which the center of the object recognition apparatus 1 is the origin (0, 0), the vehicle width direction is the X axis, and the vehicle forward direction is the Y axis.

  In S130, integration processing is performed on the distance measurement data converted into the Cartesian coordinates to generate target data. This integration process will be described later with reference to FIG. 3. However, when a predetermined condition based on the distance ΔX in the X-axis direction and the distance ΔY in the Y direction between the distance measurement data is satisfied, the point set is integrated. To generate target data.

  The target is a rectangular area set to a size including a set of points integrated by integration, and the target data includes the center coordinates (X, Y) of the area, It is assumed that at least two side data (W (width), D (depth)) representing the size of the region and the left and right beam numbers of the region are included.

  In subsequent S140, object data is generated. Based on the target data obtained in S130, the center position (X, Y) and size (W, D) of the object are obtained, and the own vehicle is determined based on the temporal change of the center position (X, Y). The relative speed (Vx, Vy) of a moving object such as a preceding vehicle with respect to the position is obtained.

  The object data generated by the object recognition device 1 in this manner is supplied to the vehicle control device (not shown) as described above, and an alarm that generates an alarm when the object is an obstacle existing in the alarm region. It is used for control and so-called inter-vehicle distance control for controlling the vehicle speed of the host vehicle according to the situation of the preceding vehicle when the object is the preceding vehicle.

Details of the integration process in S130 of FIG. 2 will be described with reference to FIG.
First, distance measurement data for starting the search is set (S131). Then, a search area is set (S132), and the distance measurement data included in the same search area are combined (S133). This process is integrated.

  In S134, it is determined whether or not the integration is completed. If the integration is not completed (S134: NO), the next distance measurement data is set (S135), and the process returns to S132 to set the search area. Ranging data is combined (S133). When all the integrations are completed (S134: YES), this process ends.

  For example, as shown in FIG. 4A, a situation is assumed in which a preceding vehicle is present in front of a road (traveling road) on which the host vehicle is traveling and a guardrail is present on the left side of the road. By the integration process, as shown in FIG. 4B, target data corresponding to the preceding vehicle and target data corresponding to the guardrail are generated. FIG. 4 is an explanatory view of the positional relationship and integration between the host vehicle and the preceding vehicle / guard rail. In FIG. 4B, the detection position of the distance measurement data is indicated by *, and the target data including the point set in which the distance measurement data is integrated by the integration process is indicated by a rectangle.

<Integration of search area setting and ranging data>
Here, the search area setting and the integration of distance measurement data in S132 and S133 of FIG. 3 will be described in more detail with reference to FIGS.

  FIG. 5 is an explanatory diagram showing search areas to be integrated, and an elliptical area indicated by a broken line in the figure is a search area used when integrating distance measurement data. The equation showing the ellipse in this region is the following equation (1).

d ² = (ΔX / αX) ² + (ΔY / αY) ² (1)
At this time,
ΔX: Distance between data in the X-axis direction ΔY: Distance between data in the Y-axis direction αX: k1 * Distance from the object recognition device 1 to the distance measurement data * Resolution of the object recognition device 1 αY ... k2 * Vehicle speed * Sampling period of the object recognition apparatus 1 (the sampling period coincides with the scan period (100 ms) of the object recognition apparatus 1 and k1 and k2 are coefficients).
When the value on the right side of the equation (1) is d ² or less, the distance measurement data is integrated.

In setting this search area, the following points are taken into consideration.
By relatively increasing the value of αY, as shown in FIG. 5, the search area has an elliptical shape that is long in the Y direction (traveling direction). This will be described more specifically.

  The length of the search region in the X direction (vehicle width direction) is set to a length corresponding to one or two adjacent horizontal beams. The intention is to make the length so that two adjacent distance measurement data of the same object are included exactly, and conversely to make the length not include two distance measurement data from another object. .

  That is, for example, even if a plurality of preceding vehicles are in parallel, the vehicle travels at a certain distance. Moreover, even if the preceding vehicle travels near a roadside object such as a guard rail, the vehicle travels at a certain distance. Accordingly, the length of the search area in the X direction (vehicle width direction) is set so that distance measurement data from such another object is not included. However, the length is set such that two adjacent distance measurement data of the same object are included.

  By integrating the distance measurement data using the search area set in this way, if the object itself is separated in the X direction (vehicle width direction) of the search area, the detection data of these other objects are integrated. It will not be. That is, not only the preceding vehicle and the roadside object but also the preceding vehicles can be correctly distinguished without being mistakenly integrated.

  The Y direction of the search area is set based on the distance traveled by the vehicle during the sampling period. If the coefficient k2 is slightly larger than 1, the distance + α that the vehicle moves during the sampling period becomes the traveling direction length of the search region. By combining the distance measurement data using the search area set in this way, a target corresponding to a roadside object extending in the direction along the road (traveling direction) or aligned in the direction is the same object. To be considered.

  As such a roadside object, for example, there are things that are continuously installed like a guard rail or a wall, and there are things that are installed at certain intervals (so-called discontinuous) like a standing tree or a delineator. In any case, the roadside objects installed along the road are easily recognized as the same object by setting the search area as described above, and thus can be easily distinguished from the preceding vehicle.

Next, the setting of the search area to be integrated when the road ahead is curved will be described.
As shown in FIG. 6, when the road ahead is curved, the road shape (curve radius) is estimated by a known method. For example, various methods such as obtaining based on the steering angle, taking an image of the front of the vehicle, obtaining based on the shape of a running division line (so-called white line) existing in the image, and obtaining the road shape from map data Has been proposed.

  Then, the direction of the road (traveling road) at an arbitrary point is calculated based on the estimated road shape (curve radius) described above, and as shown in FIG. The search area is set by regarding the road direction at the position as the vehicle traveling direction. Thereafter, the same applies as in the case of the straight road described with reference to FIGS.

  The setting of the search area will be described more specifically with reference to FIG. The Y axis in FIG. 8 indicates the traveling direction based on the current vehicle position, and the Y ′ axis indicates the road direction = vehicle traveling direction at the (arbitrary) distance measurement data position.

  Based on the inclination θ of the Y ′ axis with respect to the Y axis, coordinate conversion from the XY orthogonal coordinates to the X′Y ′ orthogonal coordinates is performed, and a similar search region is set in the X′Y ′ orthogonal coordinates after the coordinate conversion. The equation showing the ellipse of this search area is the following equation (2).

(d ′) ² = (ΔX ′ / αX) ² + (ΔY ′ / αY) ² (2)
At this time,
ΔX ′: Distance between data in the X-axis direction ΔY ′: Distance between data in the Y-axis direction αX: k1 * Distance from the object recognition device 1 to distance measurement data * Resolution of the object recognition device 1 αY: k2 * Self-vehicle speed * Sampling cycle of the object recognition device 1 (the sampling cycle coincides with the scanning cycle (100 ms) of the object recognition device 1 and k1 and k2 are coefficients).
When the value on the right side of equation (2) is (d ′) ² or less, the distance measurement data is integrated.

<Effect>
As described above, the object recognition device 1 generates target data by integrating distance measurement data included in the same search area, and performs object recognition using the generated target as a reference. The direction is set based on the distance traveled by the vehicle during the sampling period. By integrating distance measurement data using the search area set in this way, targets that extend long in the direction along the road (traveling direction) or are aligned in that direction are regarded as the same object. To do. Therefore, a roadside object such as a guardrail and a preceding vehicle can be correctly distinguished.

  Further, the length in the X direction (vehicle width direction) is set to a length corresponding to one or two adjacent horizontal beams. By integrating distance measurement data using the search area set in this way, if the object itself is separated in the X direction (vehicle width direction) of the search area, the distance measurement data of these other objects are integrated. It will not be done. That is, not only the preceding vehicle and the roadside object but also the preceding vehicles can be correctly distinguished without being mistakenly integrated.

Even when the road is curved, the search area can be set in an appropriate state according to the curve.
Thus, the object recognition apparatus 1 that recognizes an object with high accuracy can be realized.

[Other Embodiments]
Although one embodiment of the present invention has been described above, the present invention is not limited to the above embodiment, and can be implemented in various modes without departing from the gist of the present invention.

For example, although laser light is used as the transmission wave in the above embodiment, radio waves such as millimeter waves may be used.
In the above embodiment, an elliptical area as shown in FIGS. 5, 7, and 8 is set as the search area, but the search area is not limited to such a shape. In the above embodiment, the length of the search area in the X direction (vehicle width direction) is set to a length corresponding to one or two adjacent horizontal beams, and the Y direction (traveling direction) of the search area is The distance traveled by the vehicle during the sampling period is set as a reference. This intention is as follows.

  In other words, the length in the X direction is set so that two adjacent distance measurement data of the same object are just included, and conversely, two distance data from another object are not included. In addition, with respect to the Y direction, it is intended that the targets corresponding to roadside objects that extend long in the direction along the road (traveling direction) or are aligned in the direction are regarded as the same object. . Therefore, the search area may be a shape that can realize such an intention.

  DESCRIPTION OF SYMBOLS 1 ... Object recognition apparatus 10 ... Light emission part 11 ... Laser diode (LD) 12 ... LD drive circuit 13 ... Light emission lens 14 ... Scanner 15 ... Motor drive circuit 20 ... Light reception part 21 ... Light reception lens 22 ... Light reception element 23 ... Amplifier 30 ... Detection circuit 40 ... control unit

Claims (2)

An object recognition device that is mounted on a vehicle and recognizes an object that reflects a transmission wave based on a result of scanning using a plurality of transmission wave beams set to be adjacent to each other at least along the vehicle width direction. ,
Detection means for detecting a reception intensity and a position of the object for each transmission wave beam based on a reception signal of a reflected wave from the object obtained by the scanning;
Based on the detection result of the detection means, if the position of another detection result is also included in the search region set based on the position of a certain detection result, a plurality of the detection results are integrated as belonging to the same target. A targetization means that
With
The length in the vehicle width direction of the search area of the target means is set to such a length that the reflected waves from two adjacent transmission wave beams are just included, and the length of the search area in the vehicle traveling direction is set. Is set with reference to the distance traveled by the vehicle during the scanning wave beam scanning period,
An object recognition apparatus for recognizing an object with reference to a target generated by the target converting means. It is configured to be able to obtain the direction of the road at any point on the road of the vehicle,
The object recognition apparatus according to claim 1, wherein the search area of the target marking unit is set by regarding the direction of the traveling path at the position of the detection result as the vehicle traveling direction.