JP5206579B2 - Object detection device - Google Patents

Description

  The present invention relates to an object detection device.

  In recent years, driving support devices such as collision prevention devices and inter-vehicle distance control devices have been developed. In such a driving support device, it is important to detect an object (for example, another vehicle, a pedestrian) around the host vehicle. In the object detection device, reflection points satisfying a predetermined condition among a number of reflection points (detection points) detected by a radar such as a millimeter wave radar are grouped into the same segment as reflection points reflected by the same object, and the grouping is performed. Object information is obtained from the reflection point data of each segment. The apparatus described in Patent Document 1 includes a radar having a detection range on the front left side of the host vehicle and a radar having a detection range on the front right side, and detects an object on the front side of the host vehicle.

JP 2008-152389 A JP 2004-132734 A JP 2003-57339 A JP-A-8-240660

  When information on other vehicles ahead is required as in the inter-vehicle distance control device, the radar is attached so that its center axis coincides with the traveling direction of the own vehicle in order to make the radar detection range in front of the own vehicle. ing. In this case, for example, the radar reflecting surface for the other vehicle traveling in front of the host vehicle becomes the rear surface of the other vehicle, and hardly changes during traveling. Therefore, the multiple reflection points detected by the radar at each cycle time are reflected at substantially the same location on the rear surface of the other vehicle. Therefore, when grouping is always performed, the reflection points in each cycle are the same. The reflection points reflected by the object can be easily grouped as the same segment in time series. As a result, the same object can be detected stably, and physical information (relative distance in the depth direction, relative distance in the horizontal direction, relative velocity, etc.) obtained from the reflection point data of each segment is also stable information.

  When information on other vehicles on the front side is required, such as an accident prevention device, the radar is attached so that its center axis is different from the traveling direction of the own vehicle in order to set the front side of the own vehicle as the radar detection range. It has been. In this case, for example, the radar reflecting surface for the other vehicle traveling on the road intersecting the road on which the host vehicle is traveling is the side surface, front surface or rear surface of the other vehicle, and the angle with respect to each surface changes during traveling. It becomes a complicated shape. For this reason, a plurality of reflection points detected by the radar at each cycle time are reflected at different locations on each surface of the other vehicle. Therefore, when the grouping is always performed, the reflection points in each cycle are the same. It is not easy to group reflection points reflected by an object as the same segment in time series. For example, a plurality of reflection points for the same object at each cycle time may be grouped as the same segment in one cycle, but may be grouped into a plurality of different segments in the next cycle. As a result, the same object cannot be detected stably, and physical information (relative distance in the depth direction, relative distance in the horizontal direction, relative speed, etc.) obtained from the reflection point data of each segment is not stable. As a result, the system accuracy and responsiveness of the latter stage are lowered.

  Therefore, an object of the present invention is to provide an object detection apparatus that can detect an object with high accuracy even when time-series radar reflection data is unstable.

  An object detection apparatus according to the present invention is an object detection apparatus that groups reflection data detected by a radar detection unit into segments and detects an object based on the reflection data for each segment, and reflects the reflection based on a predetermined condition. Based on grouping means for grouping data into segments, segment history specifying means for specifying identification information of segments belonging to the past for each reflection data, and identification information of past segments of reflection data belonging to the segment grouped this time And a segment identifying means for discriminating the identity between the segment grouped this time and the segment grouped in the past.

  The radar detection means performs detection every predetermined cycle time, and outputs reflection data detected in each cycle. In the object detection apparatus, when the reflection data is detected by the radar detection means in each cycle, the reflection data is grouped into segments based on a predetermined condition by the grouping means. In the object detection means, the segment history specifying means specifies the identification information of the segment in which each reflection data belonged to the past (previous cycle, previous cycle, etc.). Furthermore, in the object detection device, the identity of the segment grouped this time and the segment grouped in the past based on the identification information of the past segment of the reflection data belonging to the segment grouped this time by the segment identification means, The identification information of the past segment determined to be the same as the current segment is given. And in an object detection apparatus, an object is detected based on the reflection data for every segment. In this way, in the object detection device, the reflection data for the same object in time series is determined by determining the identity of the segments in time series based on the identification numbers of the segments to which the reflection data belonging to the segments belonged in the past. Can be the same segment having the same identification information, and the object can be detected with high accuracy even when the reflection data from the radar in time series is unstable.

  In the object detection device of the present invention, the grouping unit determines a representative point from the reflection data based on a predetermined condition, and the reflection data located within the segment range set based on the position of the representative point is determined. A configuration may be adopted in which the representative points and members are grouped as the same segment by extracting them as members.

  The grouping means of this object detection apparatus determines a representative point from the reflection data based on a predetermined condition, and sets a segment range based on the position of the representative point. Further, the grouping means searches the reflection data included in the segment range from the reflection data, and groups the reflection data included in the segment range as a member and the same segment as the representative point. As described above, the object detection apparatus can perform grouping into a segment including a representative point and a member (however, a segment including only a representative point may be included).

  In the object detection apparatus of the present invention, the segment history specifying means specifies the state of each reflection data in the past as a representative point or a member, and the segment identification means is the past of each reflection data belonging to the currently grouped segment. It is preferable to determine which segment identification information is to be taken over in accordance with the state of the representative point or member and the continuation state of the segment to which each reflection data belonged in the past.

  In this object detection device, the state of whether each reflection data belongs to the past or the representative point or member is specified by the segment history specifying means. In the object detection device, the state of the representative point / member and the identity of the segment to which each reflection data belonged in the past in the segment to which each reflection data belonging to the segment grouped this time by the segment identification means belonged in the past. In accordance with the continuation state, each piece of reflection data belonging to the currently grouped segment is determined which identification information is to be taken over among the identification information of the segment to which it belongs in the past. As described above, the object detection device determines which identification information is to be taken over based on the past representative point / member state of each reflection data belonging to the segment grouped this time and the continuation state of the segment belonging to the past. By determining, the same identification information can be inherited with high accuracy for the segment to which the reflection data for the same object belongs in time series.

  The object detection apparatus according to the present invention preferably includes a travel environment detection unit that detects a travel environment of a vehicle on which the radar detection unit is mounted, and a segment change unit that changes a segment range according to the travel environment.

  In this object detection device, the traveling environment detection means detects the traveling environment of the vehicle (for example, the environment around the vehicle, the traveling state of the vehicle). In the object detection device, the segment range (shape, size, etc.) is changed by the segment changing means according to the traveling environment. As described above, in the object detection device, by changing the segment range according to the traveling environment of the vehicle on which the radar detection means is mounted, an appropriate segment range can be set according to the situation. High precision grouping is possible.

  In the object detection apparatus according to the present invention, the physical quantity calculation means for calculating the physical quantity of the segment based on the physical quantity of each reflection data belonging to the segment, and depending on the current and past representative points or members of each reflection data It is preferable to include weight determination means for determining the weight of each reflection data in the calculation of the physical quantity of the segment in the physical quantity calculation means.

  In this object detection apparatus, the weight determination means determines the weight for each reflection data belonging to the segment according to the current and past representative point / member states. In the object detection device, the physical quantity calculation means calculates the physical quantity of the segment (for example, the relative distance, lateral position, and speed with respect to the radar detection means) based on the physical quantity and weight of each reflection data belonging to the segment. As described above, in the object detection device, the physical quantity of the detected object can be calculated with high accuracy by setting the weights that contribute to the calculation of the physical quantity as the segment for each reflection data belonging to the segment.

  In the object detection apparatus of the present invention, it is preferable that the weight determination unit changes the weight of the reflection data as a member in accordance with the distance between the radar detection unit and the representative point. As described above, the object detection apparatus calculates the physical quantity of the detected object with higher accuracy by changing the weight of the member (and thus the weight of the representative point) according to the distance between the radar detection means and the representative point. can do.

  According to the present invention, it is possible to identify the same reflection data for the same object in time series by determining the identity of the segments in time series based on the identification numbers of the segments to which the reflection data belonging to the past belonged. The same segment having information can be used, and an object can be detected with high accuracy even when the reflection data by the radar is unstable in time series.

It is a block diagram of the object detection apparatus which concerns on this Embodiment. It is an example of temporary grouping with respect to the reflective point detected with the millimeter wave radar. It is a list of patterns for determining segment identification numbers. It is an example of the segment grouped by the last cycle and this cycle, respectively. It is a map which shows the distance coefficient with respect to the distance of the representative point used for weight ratio calculation. This is another example of segments grouped in the previous cycle and the current cycle.

  Hereinafter, an embodiment of an object detection device according to the present invention will be described with reference to the drawings. In addition, the same code | symbol is attached | subjected about the element which is the same or it corresponds in each figure, and the overlapping description is abbreviate | omitted.

  In the present embodiment, the object detection device according to the present invention is applied to an object detection device mounted on a vehicle. The object detection device according to the present embodiment detects an object (for example, another vehicle or a pedestrian) existing around the host vehicle (in particular, forward or front side), and uses the ACC [Adaptive Cruise Control ] Provided to driving support devices such as traffic jam tracking, obstacle warning (including sensory brakes and seat belt warnings) system, PBA [Pre-crashBreak Assist] system, PSB [Pre-crash SeatBelt] system, intervention brake system, etc. To do.

  With reference to FIGS. 1-6, the object detection apparatus 1 which concerns on this Embodiment is demonstrated.

  The object detection device 1 groups a large number of reflection points detected by the millimeter wave radar into segments based on a predetermined condition. Then, in the object detection device 1, for each grouped segment, the segment identification number is determined based on information on the segment to which each reflection point belonging to the segment previously belongs. Further, the object detection device 1 calculates a physical quantity (distance, lateral position, relative speed, etc.) as a segment (detected object) based on the physical quantity of the reflection point included in the segment for each segment for which the identification number is determined. To do. For this purpose, the object detection apparatus 1 includes a millimeter wave radar 10 and an ECU [Electronic Control Unit] 20, and the ECU 20 includes a signal processing unit 21, a temporary grouping unit 22, a segment identification number determination unit 23, and a physical quantity calculation unit 24. Composed.

  In the present embodiment, the millimeter wave radar 10 corresponds to the radar detection means described in the claims, the temporary grouping section 22 corresponds to the grouping means described in the claims, and the segment identification number determination section. Reference numeral 23 corresponds to the segment history specifying means and segment identification means described in the claims, and the physical quantity calculation unit 24 corresponds to the physical quantity calculation means and weight determination means described in the claims.

  The millimeter wave radar 10 is an FMCW [Frequency Modulated Continuous Wave] type radar using millimeter waves. The millimeter wave radar 10 is provided at the center, right end, and left end of the front end of the host vehicle. The millimeter wave radar 10 provided at the center has a radar center axis (detection center direction) that coincides with the traveling direction of the host vehicle, and has a detection area with a predetermined angle range in front of the host vehicle centering on the center axis. doing. The millimeter wave radar 10 provided at the right end has a center area of the radar different from the traveling direction of the host vehicle, and has a detection area in a predetermined angle range diagonally right forward of the host vehicle with the center axis as a center. The millimeter wave radar 10 provided at the left end is different from the traveling direction of the own vehicle in the center axis of the radar, and has a detection area in a predetermined angle range on the left front side of the own vehicle with the center axis as a center.

  The millimeter wave radar 10 transmits a millimeter wave at every constant angle in the left-right direction and receives a reflected millimeter wave at every constant cycle time. At this time, the millimeter wave radar 10 modulates the frequency of the millimeter wave to be transmitted, and transmits the millimeter wave by increasing / decreasing the frequency of the millimeter wave continuously and linearly. When the transmitted millimeter wave is reflected by the object, the reflected millimeter wave is received with a time delay of 2R, which is twice the distance R between the millimeter wave radar 10 and the object. The millimeter wave radar 10 transmits a beat signal obtained by mixing the transmission wave and the reception wave to the ECU 20.

  The ECU 20 is an electronic control unit including a CPU [Central Processing Unit], a ROM [Read Only Memory], a RAM [Random Access Memory], and the like, and comprehensively controls the object detection device 1. In the ECU 20, the signal processing unit 21, the temporary grouping unit 22, the segment identification number determination unit 23, and the physical quantity calculation unit 24 are configured by loading an application stored in the ROM into the RAM and executing it by the CPU. The ECU 20 receives beat signals from the three millimeter wave radars 10 for each cycle time of the millimeter wave radar 10. Then, the ECU 20 performs processing in each of the units 21, 22, 23, and 24 for each received beat signal, and when an object can be detected, transmits information on the object to the driving support device.

  The signal processing unit 21 performs FFT [Fast Fourier Transform] analysis on the received beat signal for each cycle in the millimeter wave radar 10, and the beat frequency in the frequency increasing section and the beat frequency in the frequency decreasing section by peak detection. To extract. In this peak detection, a normal threshold for detecting an object with high reflectance (for example, a vehicle) and a low threshold for detecting an object with low reflectance (for example, a pedestrian) (<normal threshold) are used.

  The signal processing unit 21 combines (pairs) the beat frequency in the increasing section and the beat frequency in the decreasing section. In this pairing, when the relative speed between the host vehicle (millimeter wave radar 10) and the object to be detected is not 0, the reflected millimeter wave shifts the frequency due to the Doppler effect. This frequency shift is taken into account when combining with the beat frequency in the decreasing section. The signal processing unit 21 calculates the distance and relative speed in the depth direction from the own vehicle (millimeter wave radar 10) based on the beat frequency in the increasing section and the beat frequency in the decreasing section for each pair that can be paired. Calculate. The signal processing unit 21 calculates the distance (lateral position) from the radar center axis for each pair based on the angle with respect to the radar center axis when the millimeter wave is transmitted.

  The signal processing unit 21 performs the above processing for each cycle in the millimeter wave radar 10. Then, the signal processing unit 21 determines whether or not the reflection points of each pair are detected continuously for several cycles, and the pair detected for several cycles is determined as a confirmed pair of reflection points. As for the information on the confirmed pair of reflection points, distance, lateral position, and relative speed are calculated as physical quantities, and normal threshold flag (ON / OFF), low threshold flag (ON / OFF), moving object flag (stationary object, (Approaching moving object, moving object moving away) is set. The normal threshold flag is set to ON for the reflection point detected with the normal threshold in the peak detection. The low threshold flag is set to ON for a reflection point detected with a low threshold in peak detection. Whether the moving object flag is a stationary object, an approaching moving object, or a separated moving object is determined by a change in relative speed with time.

  When the signal processing unit 21 detects the reflection point of the confirmed pair, the temporary grouping unit 22 groups the reflection point of the confirmed pair into each segment based on the grouping condition. The grouping condition includes a condition for determining a representative point and a condition for determining a member.

  There are three conditions a, b, and c for determining the representative point, and the condition determination is performed in order from condition a. Condition a is a reflection point whose distance is close to the millimeter wave radar 10. Condition b is a reflection point whose lateral position is close to the radar central axis. c is a reflection point which is an approaching moving object and has a high relative speed.

  There are three conditions d, e, f, and g for determining members, and a condition determination that satisfies these three conditions d, e, f, and g is performed. The condition d is a reflection point included in the segment range set with the representative point as a reference. The condition e is a reflection point whose relative speed is within a predetermined range of the relative speed of the representative point. The condition f is a reflection point detected with the same threshold as the representative point. The condition g is a reflection point of the same type as the representative point (stationary object, approaching moving object, separation moving object).

  Specifically, first, the temporary grouping unit 22 extracts the reflection point closest to the millimeter wave radar 10 from the reflection points of all the definite pairs, and a plurality of reflection points are located at the closest distance. If there is a reflection point whose lateral position is close to the radar center axis, and if there are multiple reflection points near its lateral position, extract a reflection point with a large relative velocity in the approaching moving object. Let the extracted reflection point be a representative point. Then, the temporary grouping unit 22 sets the segment range with the position of the representative point as the position closest to the millimeter wave radar 10 in the depth direction (distance direction) in the segment range and the center position in the horizontal direction. The segment range is a rectangular shape having a range of several meters from the representative point in the depth direction and ± m from the representative point in the horizontal direction, and is set in advance based on the size of the vehicle. Further, the temporary grouping unit 22 is a reflection point within the segment range from the reflection points excluding the representative point from the reflection points of all the definite pairs, and the relative speed is ± several than the relative speed of the representative point. It is a reflection point within km / h and is a reflection point detected with the same threshold as the representative point based on the normal threshold flag and the low threshold flag, and the same type as the representative point based on the moving object flag (stationary object, approaching) The reflection points of the moving object and the moving object are extracted as members. Then, the temporary grouping unit 22 sets the segment including the extracted representative points and members.

  Each time a segment is grouped, the provisional grouping unit 22 extracts a reflection point that is a representative point in the same processing as described above from the reflection points of a definite pair that is not grouped. Then, the temporary grouping unit 22 sets a segment range by the same process as described above with reference to the position of the representative point. Further, the temporary grouping unit 22 extracts a reflection point to be a member from the reflection points obtained by removing the representative points from the reflection points of the fixed pair that are not grouped by the same processing as described above. Then, the temporary grouping unit 22 sets the segment including the extracted representative points and members.

  A temporary identification number is assigned as the segment identification number. In addition, since members may not be extracted, in this case, the segment is composed only of representative points.

  FIG. 2 shows the reflection of the confirmed pair detected in the detection area (LS to RS) diagonally to the left of the host vehicle MV around the radar central axis CA of the millimeter wave radar 10 provided at the front left end of the host vehicle MV. An example of grouping for points P1, P2,..., P9 is shown. In this example, first, the reflection point P1 for the roadside stationary object SO1 is extracted as a representative point, the segment range A1 is set with the reflection point P1 as a reference, and there is no other reflection point that falls within this segment range A1. A segment composed of reflection points P1 as representative points is grouped. Next, the reflection point P2 with respect to the other vehicle OV1 is extracted as a representative point, the segment range A2 is set on the basis of the reflection point P2, and there are reflection points P3 and P4 with respect to the other vehicle OV1 that fall within this segment range A2. Then, the segment composed of the reflection point P2 as the representative point and the reflection points P3 and P4 as the members is grouped. Next, the reflection point P5 with respect to the stationary object SO2 is extracted as a representative point, the segment range A3 is set on the basis of the reflection point P5, and there is a reflection point P6 with respect to the stationary object SO3 that falls within this segment range A3. A segment composed of a reflection point P5 as a point and a reflection point P6 as a member is grouped. Next, the reflection point P7 for the other vehicle OV2 is extracted as a representative point, the segment range A4 is set based on the reflection point P7, and there is a reflection point P8 for the other vehicle OV2 that falls within this segment range A4. A segment composed of a reflection point P7 as a point and a reflection point P8 as a member is grouped. Finally, the reflection point P9 for the roadside stationary object SO4 is extracted as a representative point, and the segment range A5 is set based on the reflection point P9, and there is no other reflection point that falls within this segment range A5. As a result, the segment consisting of the reflection point P9 is grouped.

  In the segment identification number determination unit 23, when the segments are grouped by the temporary grouping unit 22 in order to determine the identity of the segments in the time series for each cycle in the millimeter wave radar 10, the identification number is determined for each segment. Determine the identification number based on the conditions. There are five conditions A, B, C, D, and E as identification number determination conditions, and the condition determination is performed in order from condition A.

  Condition A basically identifies the segment to which the reflection point that was the representative point in the previous cycle belonged in the previous cycle with respect to the segment consisting of the reflection point detected at the normal threshold in the current cycle. Take over the number. In particular, when there are multiple reflection points that were representative points in the previous cycle in the segment, the segment counters set for each reflection point are compared, and the reflection point with the largest segment counter value is the previous one. Inherit the identification number of the segment that belonged to the cycle. Further, when there are a plurality of reflection points having the largest counter value, the condition determination is performed in the order of the three conditions X, Y, and Z. The condition X is a reflection point whose distance is close to the millimeter wave radar 10. Condition Y is a reflection point whose lateral position is close to the radar central axis. Condition Z is a reflection point that is an approaching moving object and has a large relative speed. In addition, when there are a plurality of reflection points where these three conditions X, Y, and Z completely match, the smaller one of the identification numbers of the segments to which each reflection point belongs in the previous cycle is inherited. .

  The segment counter is a counter that is set for each reflection point of the definite pair and counts the number of cycles that continue to exist in the segments having the same identification number. The segment counter is not reset even if the member takes over the identification number of the segment, and the counter value is reset when the identification number is not taken over.

  Condition B is a segment consisting of reflection points detected at the normal threshold in this cycle, and a segment whose identification number has not been inherited by condition A is a member in the previous cycle and a representative point in this cycle. A reflection point that inherits the identification number of a segment to which a reflection point belonged in the previous cycle, or a reflection point that is a new reflection point and is not a new member in this cycle and is a member that is not a new member Takes over the identification number of the segment it belonged to in the previous cycle. In particular, when there are a plurality of non-new members and representative points whose identification numbers are not inherited by the condition A in the segment, the segment counter set at each reflection point is compared, and the counter value of the segment counter is the largest Takes over the identification number of the segment to which the reflection point belonged in the previous cycle. Further, when there are a plurality of reflection points having the largest counter value, the condition determination is performed in the order of the three conditions X, Y, and Z.

  The condition C is a condition for a segment composed of reflection points detected at a low threshold value in this cycle (that is, a segment whose identification number has not yet been inherited by the conditions A and B). Except for these conditions, the conditions are the same as the conditions A.

  Condition D is a segment consisting of reflection points detected at a low threshold in the current cycle, in which the identification number is not inherited by condition C (that is, the identification number is inherited by conditions A, B, and C). This is the same condition as the condition D except for the low threshold value or the normal threshold value.

  Condition E is a condition for a new segment. For a segment whose identification number is not inherited by conditions A, B, C, and D, or a segment composed of a representative point of a new reflection point and a member of the new reflection point. Give a new identification number.

  As a specific process, the segment identification number determination unit 23 first performs a determination under the condition A on the segment with the normal threshold among the segments grouped by the temporary grouping unit 22, and a representative that satisfies the condition A If a point or member exists, the identification number of the segment to which the representative point or member belonged in the previous cycle is taken over, and the segment counter is incremented from the counter value of the representative point or member.

  Next, in the segment identification number determination unit 23, the segment in the condition B is excluded from the segments grouped by the temporary grouping unit 22 except the segment whose identification number is inherited by the condition A from the segment at the normal threshold. If there is a representative point or member that satisfies the condition B, the identification number of the segment to which the representative point or member belonged in the previous cycle is taken over, and the segment counter is counted from the representative point or member counter value. Increment.

  Next, the segment identification number determination unit 23 performs the determination under the condition C for the segment with the low threshold among the segments grouped by the temporary grouping unit 22, and there is a representative point or member that satisfies the condition C. In this case, the identification number of the segment to which the representative point or member belongs in the previous cycle is taken over, and the segment counter is incremented from the counter value of the representative point or member.

  Next, in the segment identification number determination unit 23, the segment in the condition D is excluded from the segments grouped by the temporary grouping unit 22 except the segment whose identification number is inherited by the condition C from the segment at the low threshold. When there is a representative point or member satisfying the condition D, the identification number of the segment to which the representative point or member belongs in the previous cycle is taken over, and the segment counter is determined from the counter value of the representative point or member. Increment.

  Finally, the segment identification number determination unit 23 assigns a new identification number to the segment that has not been inherited among the segments grouped by the temporary grouping unit 22 and sets the segment counter to 1. .

  With reference to FIG. 3, main patterns for determining segment identification numbers based on conditions A to E will be described.

  In pattern 1, the state in the previous cycle of the representative point in the current cycle is the representative point, and the members in the current cycle do not include those in which the state in the previous cycle was the representative point In the case of a segment, the same identification number as the previous time (that is, the identification number of the segment to which the representative point in the current cycle belonged last time) is taken over. This pattern 1 is condition A when the segment whose normal threshold flag is ON, and condition C when the segment whose low threshold flag is ON.

  In pattern 2, the state of the representative point in this cycle in the previous cycle is the representative point, and the members in this cycle include those in which the state in the previous cycle was the representative point, When the segment counter of the representative point is compared with the segment counter of the member that was the previous representative point, if the segment has a large representative point counter value, the same identification number as the previous one (that is, the representative point in this cycle belongs to the previous time). Inherited the identification number of the segment that had been saved). This pattern 2 is condition A when the normal threshold flag is an ON segment, and condition C when the low threshold flag is ON.

  In pattern 3, the state in the previous cycle of the representative point in the current cycle is the representative point, and the members in the current cycle include those in which the state in the previous cycle was the representative point, When the segment counter of the representative point is compared with the segment counter of the member that was the previous representative point, if the segment has a large counter value, the identification number of the segment to which the member belonged last time is inherited. This pattern 3 is condition A when the normal threshold flag is an ON segment, and condition C when the low threshold flag is an ON segment.

  In pattern 4, the state of the representative point in the current cycle is the member in the previous cycle, and one member in the current cycle includes one in which the state in the previous cycle was the representative point. In the case of a segment, the identification number of the segment to which the member that was the last representative point belonged is taken over. This pattern 4 is condition A when the segment whose normal threshold flag is ON, and condition C when the segment whose low threshold flag is ON.

  Pattern 5 is a member in which the state in the previous cycle of the representative point in the current cycle is a member, and the members in the current cycle include a plurality of members in which the state in the previous cycle was the representative point In the case of the above, the identification number of the segment to which the member with the largest counter value of the segment counter among the plurality of members that were the representative points last time belongs is inherited. This pattern 5 is condition A when the normal threshold flag is an ON segment, and condition C when the low threshold flag is ON.

  In pattern 6, the state in the previous cycle of the representative point in the current cycle is a member, and the member in the current cycle does not include the member in which the state in the previous cycle was the representative point. The member is a segment that is not a new reflection point, and if the identification number of the representative point of this time and the segment that previously belonged to the non-new member has already been inherited in the above pattern, there is no identification number to be inherited. An identification number is assigned. This pattern 6 is condition E.

  In pattern 7, the state in the previous cycle of the representative point in the current cycle is a member, and the member in the current cycle does not include the member in which the state in the previous cycle was the representative point. The member is a segment that is not a new reflection point, and since the identification number of the current representative point and the segment to which the non-new member last belonged has not yet been inherited in the above pattern, the representative point segment counter and the non-new member When the segment counter has a large representative point counter value, the identification number of the segment to which the representative point belongs last time is inherited. This pattern 7 is a condition B when the segment whose normal threshold flag is ON, and a condition D when the segment whose low threshold flag is ON.

  In pattern 8, the state of the representative point in the current cycle is the member in the previous cycle, and the member in the current cycle does not include the member in which the state in the previous cycle was the representative point. The member is a segment that is not a new reflection point, and since the identification number of the current representative point and the segment to which the non-new member last belonged has not yet been inherited in the above pattern, the representative point segment counter and the non-new member In the case of a segment whose member counter value is large when compared with the segment counter, the identification number of the segment to which the member previously belonged is taken over. This pattern 8 is condition B when the segment whose normal threshold flag is ON, and condition D when the segment whose low threshold flag is ON.

  In pattern 9, the state of the representative point in the current cycle is the member in the previous cycle, and the member in the current cycle does not include the member in which the state in the previous cycle was the representative point. The member is a segment that is a new reflection point, and if the identification number of the segment to which the representative point belonged last time has already been inherited in the above pattern, there is no identification number to inherit, so the identification number of the new segment Is granted. This pattern 9 is condition E.

  In pattern 10, the state in the previous cycle of the representative point in the current cycle is a member, and the member in the current cycle does not include the member in which the state in the previous cycle is the representative point. The member is a segment which is a new reflection point, and the identification number of the segment to which the representative point last belonged has not yet been inherited in the above pattern, so the identification of the cement to which the representative point belonged last time Take over the number. This pattern 10 is condition B when the segment whose normal threshold flag is ON, and condition D when the segment whose low threshold flag is ON.

  In the pattern 11, the representative point in the current cycle is a new reflection point, and in the case of a segment including a member in which the state in the previous cycle is a representative point among the members in the current cycle, the member Takes over the identification number of the previous segment. This pattern 11 is condition A when the segment whose normal threshold flag is ON, and condition C when the segment whose low threshold flag is ON.

  In Pattern 12, the representative point in this cycle is a new reflection point, and the members in this cycle are segments that do not include those in which the state in the previous cycle was the representative point. If the identification number of the segment to which the previous member belongs has already been inherited in the above pattern, since there is no identification number to be inherited, the identification number of the new segment is assigned. This pattern 12 is condition E.

  In pattern 13, the representative point in the current cycle is a new reflection point, and the members in this cycle do not include the segment whose state in the previous cycle was the representative point. Since the identification number of the segment to which the previous member belongs has not yet been inherited in the above pattern, the member having the largest segment counter value among the members takes over the identification number of the cement to which the previous member belonged. This pattern 13 is condition B when the segment whose normal threshold flag is ON, and condition D when the segment whose low threshold flag is ON.

  In the pattern 14, the representative point in the current cycle is a new reflection point, and in the case of a segment in which the member in the current cycle is a new reflection point, there is no identification number to be inherited. Give. This pattern 14 is condition E.

  FIG. 4 shows an example of taking over the segment identification number. In this example, the segment TS11 and the segment TS12 exist in the previous cycle, and the segment TS2 is grouped in the current cycle. The reflection point P1 as a representative point belongs to the segment TS11, and the counter value of the segment counter at the reflection point P1 is 10. Further, the reflection point P2 as a representative point belongs to the segment TS12, and the counter value of the segment counter at the reflection point P2 is 50. A reflection point P1 as a representative point and a reflection point P2 as a member belong to the segment TS2. In this case, it corresponds to the pattern 3 described above, the state of the representative point P1 in the current cycle in the previous cycle is the representative point, and the state of the member P2 in the current cycle in the previous cycle is the representative point. Yes, when the segment counter of the representative point P1 and the segment counter of the member P2 are compared, the counter value of the member P2 is large. Therefore, in the segment TS2, the identification number of the segment TS12 to which the member P2 previously belonged is taken over, and the segment counter is set to 50. Incremented to 51.

  In the physical quantity calculation unit 24, for each segment whose identification number is determined by the segment identification number determination unit 23, a physical quantity (distance, horizontal) as a segment (object) based on the physical quantity of the representative point belonging to the segment and the physical quantity of each member. Position, relative speed, etc.). At that time, the physical quantity calculation unit 24 obtains the weight ratio based on the representative counter of the reflection point that is the representative point belonging to the segment and the representative counter of the reflection point that is the member, and the physical quantity and member of the representative point according to the weight ratio. The physical quantity is obtained based on the physical quantity of

  The representative counter is a counter that is set for each reflection point of the definite pair and counts the number of times it has become a representative point. Even if the representative counter becomes a member, the counter value is not reset, and the counter value is reset when the reflection point disappears.

  A method for calculating the weight ratio will be specifically described. Here, in order to make the explanation easy to understand, a case where the segment TS2 to which the representative point P1 and one member P2 belong as shown in FIG. Here, the counter value of the representative counter of the reflection point P1, which is the representative point when the state of the segment TS2 is first (first cycle), is 10, and the counter value of the representative counter of the reflection point P2, which is the member, is Suppose that it was 50. Each time the cycle is updated, the representative counter of the reflection point P1 that is the representative point is incremented, but the representative counter of the reflection point P2 that is the member is not incremented.

  In the above example, the weight ratio at the first cycle (weight value of representative point P1: weight value of member P2) = counter value of the representative counter at the first cycle of representative point P1: (first cycle of member P2) Counter value x distance coefficient α) at 2nd cycle, weight ratio at 2nd cycle = counter value of representative counter at 2nd cycle of representative point P1: (weight value at 1st cycle of member P2) (Distance coefficient α), and in the nth cycle, the weight ratio = the counter value of the representative counter at the nth cycle of the representative point P1: (weight value of the member P2 at the (n−1) th cycle × distance coefficient α) ). When the reflection point, which is a representative point, has changed from the previous cycle, the calculation of the weight ratio is performed again from the first cycle.

  The distance coefficient α is a coefficient that decreases the influence degree of the member (increases the influence degree of the representative point) on the physical quantity as the segment (object) as the representative point is closer to the millimeter wave radar 10. Therefore, as the cycle is updated, the influence of the members gradually decreases. FIG. 5 shows an example of a distance coefficient map corresponding to the distance of the representative point with respect to the millimeter wave radar 10.

  The case where the weight ratio in the above example is obtained as a specific numerical value using the distance coefficient α shown in FIG. 5 is shown. However, the distance coefficient α corresponding to the distance between the representative points is 0.8 for the first cycle, 0.7 for the second cycle, 0.6 for the third cycle, 0.6 for the fourth cycle, and 0.6 for the fifth cycle. Assume that the 0.4th and 6th cycles were 0.4 and the 7th cycle was 0.3. Weight ratio of the first cycle (weight value of representative point P1: weight value of member P2) = 10: (50 × 0.8) = 10: 40 and weight ratio of second cycle = 11: (40 × 0 7) = 11: 28, the third cycle weight ratio = 12: (28 × 0.6) = 12: 16, and the fourth cycle weight ratio = 13: (16 × 0.6) = 13: 9, fifth cycle weight ratio = 14: (9 × 0.4) = 14: 3, sixth cycle weight ratio = 15: (3 × 0.4) = 15: 1 Yes, the weight ratio of the seventh cycle = 16: (1 × 0.3) = 16: 0. Here, each weight value is rounded down.

  Furthermore, the case where a distance is calculated | required as a physical quantity of a segment in said example is shown. Assuming that the distance of the reflection point P1 as the representative point in the first cycle is Z11 and the distance of the reflection point P2 as the member is Z12, the distance Z1S = (Z11 × (10 / (10 + 40)) + as a segment (Z12 × (40 / (10 + 40)) Further, in the second cycle, if the distance of the reflection point P1, which is the representative point, is Z21, and the distance of the reflection point P2, which is the member, is Z22, the segment is The distance Z2S = (Z21 × (11 / (11 + 28)) + (Z22 × (28 / (11 + 28))).

  Incidentally, in the above example, for example, when the reflection point P2 that was a member in the 10th cycle is a segment different from the reflection point P1, the weight ratio calculation in the segment to which the reflection point P2 belongs in the 10th cycle When the point P2 is a representative point, the calculation is performed using the count value 51 of the representative counter, and when the reflection point P2 is a member, the calculation is performed using the count value 50 of the representative counter.

  Further, the method for calculating the weight ratio will be described specifically with reference to the example shown in FIG. Here, until the second cycle, the weight ratios for the segment TS21 to which the representative point P1 and one member P2 belong and the segment TS22 to which the representative point P3 and one member P4 belong are calculated as in the above example. The method of calculating the weight ratio when the segment TS21 and the segment TS22 in the second cycle become one segment TS3 in the third cycle and become the segment TS3 to which the representative point P1 and the three members P2, P3, P4 belong will be described. To do. In this case, since the reflection point P2 as a member has the same representative point as the previous cycle, the weight value of the second cycle is used for the weight ratio calculation. Since the reflection point P3, which is a member, is a member that was the representative point in the previous cycle, it becomes the first cycle in the weight ratio calculation, and the counter value of the representative counter is used in the weight ratio calculation. The reflection point P4, which is a member, has different representative points in the previous cycle and the current cycle. However, since the reflection point P3 that was the representative point in the previous cycle is included in the same segment, the weight value in the second cycle is included in the weight ratio calculation. Is used. Therefore, the weight ratio in this third cycle (weight value of representative point P1: weight value of member P2: weight value of member P3: weight value of member P4) = counter value of representative counter of representative point P1: (member P2 (Weight value × distance coefficient α) of the second cycle :) (counter value of representative counter of member P3 × distance coefficient α): (weight value of second cycle of member P4 × distance coefficient α). However, when the reflection point P3 does not belong to this segment TS3, the reflection point P4 as a member is in the first cycle in the weight ratio calculation, and the counter value of the representative counter is used in the weight ratio calculation.

  With reference to FIG. 1, the operation | movement in the object detection apparatus 1 is demonstrated. Each millimeter wave radar 10 transmits a millimeter wave at a certain angle in the left-right direction and receives a reflected millimeter wave for each cycle time, and transmits a beat signal obtained by mixing the transmitted wave and the received wave to the ECU 20.

  Each time the beat signal of each millimeter wave radar 10 is received, the ECU 20 performs FFT analysis on the beat signal, and beat detection in the frequency increase interval and beat decrease in the frequency decrease interval by peak detection (normal threshold and low threshold). Extract the frequency. Then, the ECU 20 pairs the beat frequency in the increasing section and the beat frequency in the decreasing section, and calculates a relative distance, a lateral position, a relative speed, etc. with respect to the millimeter wave radar 10 as a physical quantity for each pair. Further, the EUU 20 determines whether or not several cycles are continuously detected for each pair, determines a pair detected for several cycles continuously as a reflection point, and sets a physical quantity, Information on threshold flag, low threshold flag, and moving object flag is set.

  When the reflection points of the confirmed pair are detected, the ECU 20 sequentially determines the representative points for the reflection points of all the confirmed pairs according to the conditions a, b, and c, and from among the reflection points of all the confirmed pairs. Extract representative points. Then, the ECU 20 sets a segment range based on the position of the representative point, and is within the segment range from the reflection points excluding the representative point among the reflection points of all defined pairs, and the relative speed is the representative point. The relative speed is within ± several km / h and is detected with the same threshold as the representative point based on the normal threshold flag and the low threshold flag, and the same type as the representative point based on the moving object flag (stationary object, approaching) A reflection point which is a moving object or a moving object is extracted as a member. Then, the ECU 20 groups the segment composed of the extracted representative points and members. Every time one segment is grouped, the ECU 20 sequentially determines the representative points with respect to the reflection points of the definite pair that is not grouped by the conditions a, b, and c, and the reflection of the definite pair that is not grouped. A representative point is extracted from the points. Then, the ECU 20 sets the segment range with reference to the position of the representative point, and is within the segment range from the reflection points excluding the representative point among the reflection points of the definite pair that are not grouped, and the relative speed is Reflection that is within ± several km / h with respect to the relative speed of the representative point, is detected at the same threshold as the representative point based on the normal threshold flag and the low threshold flag, and is of the same type as the representative point based on the moving object flag Extract points as members. Then, the ECU 20 groups the segment composed of the extracted representative points and members. The ECU 20 performs the above grouping until all the reflection points of the confirmed pair are grouped.

  When the grouping for the reflection point of the confirmed pair is completed, the ECU 20 makes a determination with the identification number determination condition A for each segment at the normal threshold among the grouped segments, and there is a representative point or member that satisfies the condition A. If so, the identification number of the segment to which the representative point or member belonged in the previous cycle is taken over. Next, the ECU 20 performs the determination under the identification number determination condition B for each segment excluding the segment whose identification number is determined under the identification number determination condition A out of the grouped segments with the normal threshold. If there is a representative point or member that satisfies this condition B, the identification number of the segment to which the representative point or member belonged in the previous cycle is taken over. Next, the ECU 20 performs a determination under the identification number determination condition C for each segment with a low threshold among the grouped segments. If there is a representative point or a member that satisfies the condition C, the representative point is determined. Alternatively, the identification number of the segment to which the member belonged in the previous cycle is taken over. Next, the ECU 20 performs the determination under the identification number determination condition D for each segment excluding the segment whose identification number is determined under the identification number determination condition C out of the grouped segments at the low threshold, If there is a representative point or member that satisfies this condition D, the identification number of the segment to which the representative point or member belonged in the previous cycle is taken over. Finally, the ECU 20 assigns a new identification number to each segment other than the segments whose identification numbers are determined by the identification number determination conditions A, B, C, and D among the grouped segments.

  When the identification numbers of all the grouped segments are determined, the ECU 20 determines, for each segment for which the identification numbers are determined, a reflection point representative counter that is a representative point belonging to the segment, a reflection point representative counter that is a member, and a millimeter wave radar. The weight ratio is calculated based on the distance coefficient corresponding to the distance of the representative point from 10, and the physical quantity (distance, lateral position, Relative speed).

  When the physical quantity of the segment is calculated, the ECU 20 transmits information (physical quantity and the like) about each segment (object) to the driving support device.

  According to this object detection device 1, the reflection point for the same object in time series is determined by determining the time series identity of the segments based on the information in the segments to which the reflection points belonging to the segments belonged in the previous cycle. The point can be the same segment with the same identification number, and the object can be detected with high accuracy even when the reflection point in time series is unstable. In particular, when the radar center axis is different from the traveling direction of the own vehicle as in the millimeter wave radar 10 provided on the front side of the own vehicle, the radar reflecting surface for other vehicles has a complicated shape. Can be determined in segment units as reflection points reflected by the same object, and the same object can be stably detected in time series.

  Furthermore, according to the object detection apparatus 1, by using a segment composed of a representative point and a member, processing can be performed easily and with high accuracy using the representative point and member in the identification number determination process and the physical quantity calculation process. it can.

  In particular, in the object detection apparatus 1, any identification number is determined based on the state of the representative point / member in the previous cycle for each reflection point belonging to the segment in the current cycle and the continuation state of the segment belonging to the past by the segment counter. By determining whether to take over, the same identification number can be taken over with high accuracy for the segment to which the reflection point for the same object in each cycle belongs. At this time, by focusing on the fact that it was a representative point in the previous cycle, the time-series identity of segments can be determined with high accuracy. In addition, it is possible to determine the identity of a segment in time series with high accuracy by making a determination with emphasis on the fact that the number of continuous segments is large.

  In the object detection device 1, the physical quantity of the segment (object) can be obtained with high accuracy by obtaining the weight ratio between the representative point belonging to the segment and the member in the physical quantity calculation processing. Furthermore, in the object detection device 1, the weight ratio can be obtained with high accuracy by obtaining the weight ratio by placing importance on the number of times the representative point has become a representative point in the past. Further, in the object detection device 1, the weight ratio is increased by obtaining the weight ratio by decreasing the influence degree of the member (and consequently increasing the influence degree of the representative point) as the distance of the representative point to the millimeter wave radar 10 is shorter. The accuracy can be obtained.

  As mentioned above, although embodiment which concerns on this invention was described, this invention is implemented in various forms, without being limited to the said embodiment.

  For example, although the present embodiment is applied to an object detection device mounted on a vehicle, an object detection function incorporated in a driving support device such as an ACC system, a PBA system, a PSB system, or an intervention brake system may be used. Further, the present invention may be applied to an object detection device mounted on a moving body other than a vehicle, or may be applied to an object detection device that is not mounted on a moving body such as a vehicle.

  In this embodiment, the FMCW millimeter wave radar is applied as the radar detection means. However, the present invention can be applied to other radars such as a laser radar, or a detection method other than the FMCW method may be used.

  In the present embodiment, the reflection point is detected using the normal threshold and the low threshold. However, the reflection point may be detected using only one threshold, or three or more thresholds may be used. A reflection point may be detected.

  In the present embodiment, the detection direction is set to the front and front sides, but other detection directions such as side, rear, and rear sides can also be applied.

  Further, in this embodiment, the representative point determination condition and the member determination condition are shown, but the representative point and the member may be determined under other conditions.

  Moreover, although the identification number confirmation condition is shown in the present embodiment, the identification number may be confirmed under other conditions.

  Further, in the present embodiment, a grouping method has been described in which representative points are determined from a number of reflection points based on conditions, and the reflection points included in the segment range are grouped as members based on the positions of the representative points. Other grouping methods may be used.

  In this embodiment, the weight ratio between the representative point and the member belonging to the segment is obtained, and the physical quantity is obtained as the segment according to the weight ratio. However, the physical quantity of the reflection point belonging to the segment is used as a segment by another method. A physical quantity may be obtained.

  Further, in this embodiment, the segment range is a rectangular fixed size range, but the segment range is L-shaped, concave shape, etc. in order to detect other vehicles on the side and front or rear of the laser reflecting surface. Other shapes may be used, or the surrounding environment (guardrail, etc.) of the own vehicle is recognized using means for recognizing the surrounding environment of the own vehicle, and the size of the segment range is changed according to the surrounding environment. Alternatively, the traveling state (vehicle speed, turning state, etc.) of the host vehicle may be detected using means for detecting the traveling state of the host vehicle, and the size of the segment range may be changed according to the traveling state.

  DESCRIPTION OF SYMBOLS 1 ... Object detection apparatus, 10 ... Millimeter wave radar, 20 ... ECU, 21 ... Signal processing part, 22 ... Temporary grouping part, 23 ... Segment identification number determination part, 24 ... Physical quantity calculation part

Claims (6)

An object detection device that groups reflection data detected by a radar detection means into segments and detects an object based on the reflection data for each segment,
Grouping means for grouping reflection data into segments based on a predetermined condition;
A segment history specifying means for specifying identification information of a segment belonging to the past for each reflection data;
Segment identification means for determining the identity of a segment grouped this time and a segment grouped in the past based on identification information of past segments of reflection data belonging to the segment grouped this time Object detection device.   The grouping means determines a representative point from the reflection data based on a predetermined condition, extracts reflection data located within a segment range set based on the position of the representative point as a member, and the representative point The object detection apparatus according to claim 1, wherein the members and the members are grouped as the same segment. The segment history specifying means specifies a state in which each reflection data is a representative point or a member in the past,
The segment identification means identifies any segment according to the state of the past representative point or member of each reflection data belonging to the currently grouped segment and the continuation state of the segment to which each reflection data belonged in the past. The object detection apparatus according to claim 2, wherein it is determined whether to take over the information. Traveling environment detection means for detecting the traveling environment of the vehicle on which the radar detection means is mounted;
The object detection apparatus according to claim 2, further comprising: a segment changing unit that changes a segment range according to a traveling environment. Physical quantity calculating means for calculating the physical quantity of the segment based on the physical quantity of each reflection data belonging to the segment;
Weight determination means for determining the weight of each reflection data in the calculation of the physical quantity of the segment in the physical quantity calculation means according to the state of each reflection data at the present and past representative points or members. The object detection apparatus of any one of Claims 2-4.   6. The object detection apparatus according to claim 5, wherein the weight determination unit changes a weight of reflection data as a member according to a distance between the radar detection unit and the representative point.

Images