JP3520326B2 - Running vehicle detection method using millimeter wave radar - Google Patents

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to the fact that a vehicle is traveling on a road, the vehicle is temporarily stopped on the road, or the vehicle is stopped for a long time due to an accident or the like. Detects various states of vehicles on the road, such as obstacles to traffic, and manages the road use state,
Further, the present invention relates to a traveling vehicle detection method using a millimeter-wave radar for notifying the following vehicle or the like of a road condition by using a bulletin board or the like and preventing an accident from occurring.

[0002]

2. Description of the Related Art Naturally, roads are used to drive vehicles, but they are used at low speeds, temporarily stopped on the roads, or stopped on the roads for a long time due to an accident or the like. It may hinder traffic. Therefore, conventionally, it has been proposed to install an image sensor on the road to monitor the traffic condition. However, when using visible light, there are drawbacks such as not being able to accurately detect the vehicle in the case of insufficient lighting such as at night, heavy rain, heavy fog, etc. In the case of, the vehicle cannot be detected accurately in the case of heavy fog, so that there is a problem that the vehicle detection does not function effectively under all weather conditions. Therefore, for example, "Japanese Patent Laid-Open No. 2000-172980"
In Japanese Patent Laid-Open Publication ", a millimeter wave radar (30 GHz to 300 GHz) excellent in response to bad weather conditions is used to detect" traffic obstacles "such as low-speed running vehicles, stopped vehicles, and parked vehicles. ing.

[0003]

However, in the case of the above-mentioned prior art, although it was possible to detect an obstacle on a road over a wide area by a millimeter wave radar, it is possible to track and detect a traveling vehicle. There wasn't. In addition, it is not possible to prevent a rear-end collision with another vehicle at an intersection or an intersection with a poor visibility, so that there is a problem in safety.

SUMMARY OF THE INVENTION Therefore, an object of the present invention is to eliminate the above-mentioned problems of the conventional obstacle detection device using the millimeter wave radar, and to reliably detect the traveling vehicle by the millimeter wave radar.
A method of detecting a traveling vehicle by a millimeter wave radar is provided.

[0005]

In order to achieve the above object, the invention according to claim 1 uses a millimeter wave radar installed on a road, and when a vehicle is tracked for each scanning, the first time is used.
Using tracking data obtained by scanning in front of the eye,
Predict the speed, and use the position / speed data obtained from this
Position / velocity data obtained by the second scan below
Segmentation processing, attribute generation processing, and the result
With the obtained position and velocity data of each segment
Divide the distance by the first and second measurement points
The smaller value compared to each of the 2nd candidates
It is characterized in that a matching process for selecting is performed . The invention according to claim 2 provides the matching according to claim 1,
The number of measurement points belonging to the same segment and the
Focusing on the tracking data obtained from the second scan
It is characterized by performing .

The invention described in claim 3 is the same as in claim 1.
Thus, when there is no match between the tracking data obtained by the first scanning and the tracking data obtained by the second scanning, a segment with a small number of measurement points obtained by the second scanning is noisy. It is characterized in that a segment having a large number of measurement points is newly added to the tracking data.

The invention according to claim 4 is the same as in claim 1.
If there is no match for the tracking data obtained by the first scan due to the effect of shadowing that cannot be observed by the millimeter-wave radar, the predicted position / velocity data in the tracking data and the number of shadows It is characterized by accumulating and holding the tracking data.
The invention according to claim 5 is characterized in that, in claim 4 , the observation noise of the tracking data caused by shadowing or the like is separated in the traveling direction of the vehicle and the vehicle width direction, and then filtered.

[0008]

DETAILED DESCRIPTION OF THE INVENTION Embodiments of the present invention will be described in detail below with reference to the drawings. 1 and 2 are configuration diagrams showing an entire apparatus according to a traveling vehicle detection method by a millimeter wave radar of the present invention (FM-CW method). That is, the millimeter wave radar 1 is installed at a high place such as a pillar provided on the road, and the millimeter wave radar 1 scans a vehicle traveling on the road with a fan-shaped (elliptical) beam. (Scanning). In this case, the scanning of the fan beam emitted from the millimeter wave radar 1 is performed at a predetermined angle (azimuth angle). The scanning method by the millimeter wave radar 1 may be a mechanical method or an electric method. As shown in FIG. 2, the millimeter wave radar 1 is a millimeter wave sensor 2 that can obtain the position and speed of a vehicle (object) by transmitting and receiving millimeter waves and processing signals, and vehicle data from the millimeter wave sensor 2. A millimeter-wave radar 1 configured as described above, and a plurality of millimeter-wave radars 1 are provided in order to cover the entire area of the road. L for communication
They are connected to each other by AN or the like. 4 is L
A host computer connected to the AN performs overall control by the host computer 4 and management of data obtained by each millimeter wave radar 1. Where LA
The type of N may be a ring type.

FIGS. 3 and 4 show a flowchart for explaining the software processing of the traveling vehicle detection method by the millimeter wave radar according to the present invention. Of these, FIG. 3 determines only when a stopped / low-speed vehicle is found as a result of the tracking process, and outputs the obtained result (information data indicating that there is a stopped / low-speed vehicle) to the external I / F. Figure 4 shows the case
Shows the case where all the obtained results are output to the external I / F regardless of the result of the tracking process (whether the stopped / low-speed vehicle is detected or not). As shown in the flow charts of FIGS. 3 and 4, in the traveling vehicle detection method by the millimeter wave radar of the present invention, data input, background difference, segmentation, attribute generation, tracking processing, stop / low speed vehicle recognition, external I / F transmission. And the background update processing (see FIGS. 17 and 18) are the eight processings (S1 to S8). An area for storing the attribute data and the tracking data is secured at the start of the flowchart.

FIG. 5 is a schematic diagram for explaining a traveling vehicle detection method by the millimeter wave radar of the present invention. Here, as the set scene, the millimeter wave radar 1 according to the present invention is installed on the shoulder side of the road (the left end side in FIG. 5), and three vehicles 5, 6, and 7 are on the road (3 lanes). It is supposed to drive from the left side to the right side. Therefore, in this case, the scanning of the fan beam emitted from the millimeter wave radar 1 is performed from the rear of each vehicle 5, 6, 7. Further, 8 is a guard rail, and 9 is an obstacle falling on the road.
6 and 7 are schematic diagrams showing time-series scenes (changes in situation) accompanying the traveling of the vehicles 5, 6 and 7, and FIG. 6 shows the first scan by the millimeter wave radar 1. FIG. 7 is a schematic diagram showing an image at the time of tracking (N-1), and FIG. 7 is an image at the time of performing the second scan (N). Further, in these FIGS. 6 and 7, as an image showing the measurement result, a thick line shows the distance measurement result, and a thick arrow shows the speed measurement result. Hereinafter, the details of the present invention will be described with reference to FIGS. 6 and 7 and the functions of the eight processes (S1 to S8).

First, data input (S1) is a process of receiving polar coordinate system data sent from the millimeter wave sensor 1. FIG. 8 shows an example of the transmitted data. Then, after the data is input, a processing step based on the background difference (S2) is performed. This background difference is a step of subtracting the input data from the background, and by performing this processing, the measurement result of the background (guardrail) shown in FIG. Only the data of the object 9 can be left. Then, FIG. 9 shows the segmentation (S
The conceptual diagram of the process in 3) is shown. Here, first, three three-dimensional memory planes of the distance and the azimuth direction (scanning direction) are prepared, and the presence / absence information (presence of presence) is present for the presence / absence plane using the measurement data (one example) shown in FIG. (Presence / absence) and speed information are written in the speed plane. Then, labels that are connected or substantially match and whose speed information substantially match are labeled as the same segment (same grouping).

When this processing is performed in the state shown in FIG. 9, three speed data (29, 30,
Since 32) are approximated (connected), a label 1 is given to them, and a different label, label 2, is given to different speed data (10) different from these three speed data. In the case of the scene shown in FIG. 6, labels 1, 2, 3
Are given to the vehicles 5, 6 and the obstacle 9, respectively. Although three memory planes are prepared in the present embodiment, the above processing may be performed with less than three memory planes shown in FIG. 9 by margining the presence information and the speed information. You may increase the memory plane according to the attribute of. Further, the above processing may be performed by a memory plane having a Cartesian coordinate system.

FIG. 10 is a conceptual diagram of the structure of attribute data in the attribute generation (S4). This attribute generation is the process of generating the center of gravity position, width, number of measurement points, and velocity of each group from the measurement data grouped by segmentation (the center of gravity position, width, number of measurement points, and velocity are referred to as "attribute", respectively). I'm calling it.) That is, in this attribute generation, the center of gravity position, width, number of measurement points, and
Calculate the velocity, etc., and convert the polar coordinates to world coordinates in the Cartesian coordinate system (the generated attribute data is
Stored in 10 structure tables. ). The coordinate conversion may be executed immediately after the data input (S1). Next, the tracking process (S5) will be described with reference to FIGS. 6 and 7 and FIGS. The center of gravity may be the average.

Here, FIG. 11 shows the structure of tracking data, and FIG.
Shows an example of data at the first time (N-1) by the millimeter wave radar 1. Incidentally, in this figure, since (N) and subsequent ones have not been substituted yet, they are indicated by an asterisk "*". Further, FIG. 13 shows a data flow of tracking processing, and FIG. 14 shows a conceptual diagram of prediction by tracking. That is, when the vehicle 6 of FIG. 6 is taken as an example, the tracking data when the first scan by the millimeter wave radar 1 is performed (N−1) is shown by “a solid star mark” in FIG. 7. According to the present invention, the tracking data is used to predict the position / velocity (star mark and thick arrow) in FIG. 7, and the attribute obtained in the vicinity of the predicted position is obtained from the data obtained by the second scan in FIG. Search for matching data (eg velocity and position).

Further, the vehicle 7 that has entered when the second scan by the millimeter wave radar 1 is performed (N) is added to the new tracking data. Further, new data is stored in the additional data at the time of the second scan (“white star”).

As shown in FIG. 13, the tracking processing described above includes position / velocity prediction processing, matching processing, shadowing processing, tracking continuation determination, tracking data update processing, new tracking data addition processing, registration deletion processing, and filter processing. The nine processings (a) to (i) of processing and tracking data output are performed. Each of these processes will be described in detail below.

(A) Position / velocity prediction processing FIG. 14 shows a conceptual diagram of prediction in the position / velocity prediction processing. That is, as shown in FIG. 14, the tracking data stores the loci up to now, and it is possible to make a prediction using the data obtained from this approximate curve. For example, assuming that the point P is the tracking position obtained by the first scan, the subsequent estimated value can be predicted. Below is an example of a prediction method using this data
Two examples are given. (A-1) Kalman prediction algorithm As shown in FIG. 14, assuming that the traveling direction of the vehicle is the X direction and the vehicle width direction is the Y direction, the discrete model for observing the vehicle with the millimeter wave radar 1 is a three-dimensional model. Then, it can be expressed as the following "Equation 1" and "Equation 2". Here, the dimension may be n-dimensional.

[0018]

[Equation 1]

[0019]

[Equation 2] However, t is the time when the latest (for example, the first scan) is sampled, T is the sampling time, “′” is the first derivative, “″” is the second derivative, and “v *”. "
Is system noise, “w *” is observation noise, “xo”,
[Xo '] and "yo" are millimeter-wave measurement data. We use this model to apply a Kalman filter,
By substituting the result into the following “Equation 3”, the velocity / position can be estimated.

[0020]

[Equation 3] However, “xf” and “yf” are processing results by the Kalman filter, and “xfp” and “yfp” are prediction results. (A-2) Tangent algorithm tracking data (tracking data obtained by the first scan) and past data for a certain number of samplings,
The position and the velocity are approximated by an n-dimensional function, and the tangential direction at the latest position is obtained. In the case of the position, the position advanced by the sampling time T at the speed obtained by the millimeter wave radar 1 in that direction is set as the predicted position. In the case of speed, an approximated function is used to calculate the speed advanced by the sampling time T, and this is used as the predicted speed.

(B) Matching process, that is, this matching process is to match all of the prediction data obtained from the tracking data with all of the attribute data, and the matching method in this case is within a certain speed range. If they match, the difference (distance) between the positions is divided by either (first tracking data and second attribute data) measurement points, and the value is stored. Then, the matching processing is performed by operating the savings data so that the savings data becomes smaller and has a one-to-one correspondence. Then, if the prediction data obtained from the tracking data that cannot be matched is set to the shadowing process, and if the matching data is able to be matched, the continuous shadowing count of the header shown in FIG. 11 is set to "0". Then, the process proceeds to the tracking data update process, or to the new tracking data addition process if the attribute data cannot be matched.

(C) Shadowing processing Shadowing does not match the prediction data obtained from the tracking data because the measured object is out of the measurement range like the vehicle 5 in FIG. Ing)
It shows the situation that was disturbed by and could not be measured. Therefore, since the vehicle and the like are naturally traveling even during such shadowing, the millimeter wave radar 1 can be used again for measurement after a certain time elapses. The processing for setting the number of consecutive shadowing times / total number of shadowing times of the header portion shown in 11 to "1" is performed.

(D) Tracking continuation judgment processing The number of continuous shadowing is judged by a certain threshold value.
Then, if the number of continuous shadowing is less than or equal to the threshold value, it is determined that the shadowing is performed, and the tracking data updating process is performed. On the other hand, if the number of consecutive shadowing is equal to or more than the threshold value, it means that the measurement cannot be performed even after a certain time has passed, and whether the vehicle is out of the measurement range (vehicle 5 in FIG. 7) or this tracking is performed. It is determined that the data is noise, and the process proceeds to the next registration deletion process (g).

(E) Tracking data update processing When the matching is performed by the matching processing, the value of the attribute data is substituted for the position / speed / measurement points of the data section (array) shown in FIG. 11, and the time is when the data is received. Substitute the value of the timer of. Also, a value indicating the tracking state is substituted into the tracking status of the header section to set the tracking count to "1". If, for example, shadowing is determined, the predicted data is substituted for the position / velocity of the data section, the previous number of measurements is substituted for the number of measurement points, and the time is the time of the timer when the data is received. Substitute the value etc. Further, a value indicating the shadowing state is substituted for the tracking status of the header portion shown in FIG. 11, and the tracking count, the continuous shadowing count, and the total shadowing count are incremented.

(F) New Tracking Data Addition Process The number of measurement points of the attribute data is judged by a certain threshold value. If it is less than the threshold value, it is judged as noise. It is judged that it has appeared, and the attribute data is substituted for the position / velocity of the data part shown in FIG. Also, the value indicating the tracking status is substituted into the tracking status, and the tracking count is set to "1".
To

(G) Registration Deletion Processing This registration deletion processing is initialization of the data section shown in FIG. Further, in this registration deletion processing, a value indicating an unused state is substituted for the tracking status of the header part to initialize the tracking count and the like.

(H) Filtering Since the tracking data includes observation noise, filtering is performed. The moving average method, the Kalman filter method, or the like is used as the filtering method. The Kalman filter model and method are the same as those in the case of the matching process prediction. Since the observation noise is effectively removed by this filter processing, the accuracy of the obtained position and speed can be improved.

(I) Tracking data output process This is an output process of passing the processing result of each processing described above to another function.

(J) Low speed / stopped vehicle recognition The processing steps in this low speed / stopped vehicle recognition are shown in FIGS.
Its outline is shown in 16. Of these, FIG. 15 is a flowchart in the case of transmitting only low speed / stopped vehicle information recognized as dangerous to a higher rank, and FIG. 16 is a flowchart in the case of transmitting all vehicle information. The procedure is as follows: Tracking data input, low speed / stopped vehicle recognition by speed,
It consists of a net appearance frequency calculation process, an appearance frequency threshold process, a measurement point threshold process, and data output. Here, in the net appearance number calculation process, net data is obtained by a process by shadowing.

That is, the speed of the tracking data and the past speed are threshold values at which it can be recognized as low speed / stop (for example, 5).
If it is Km / h), the one indicating the low speed / stop condition is substituted into the recognition result of the header part in the tracking data structure of FIG. Next, the number of appearances and the number of measurement points are thresholded with a threshold value that is a function of position / velocity, and data satisfying these values (measurement point> Th2 (x, y, V _X , V _Y )) are data. Output by output. Therefore, noise reduction (noise reduction) is performed in the threshold processing of the number of measurement points here.

(K) External Output This external output processing is processing for outputting data from an external interface such as a LAN (see FIG. 2).

(L) Background update processing This background update processing is outlined in FIGS. 17 and 18. This processing method is the same as that described in the above-mentioned "JP-A-2000-172980" which is a conventional example. In addition, the background update process here is set by
It is updated at a certain interval, and is not necessarily performed each time the millimeter-wave radar travels.

[0033]

As described above, the present invention can be applied to roads.
Using the installation the millimeter-wave radar, predicts the position and speed by using the tracking data obtained by the first previous scan when the tracking vehicle for each scanning, thereby resulting position
Segmentation process and the position and speed data obtained by the second scanning of the location and speed data and the next, and attribute generation process, the position of each segment resulting
・ A matching process is performed in which the distance from the speed data is divided by the first measurement point and the second measurement point and compared with each second candidate, and the smaller one is selected .
Thus , under all weather conditions, tracking data can be used to reliably detect and recognize a vehicle running at a low speed or a stopped vehicle, and it is possible to prevent a rear-end collision at a meeting point or an intersection with poor visibility. It has excellent effects such as expected effects.

[Brief description of drawings]

FIG. 1 is an assumed view showing a state in which a millimeter wave radar according to the present invention is applied.

FIG. 2 is a schematic diagram showing the overall configuration of the millimeter-wave radar device.

FIG. 3 is a flowchart showing a configuration procedure of the present invention (No. 1).

FIG. 4 is a flowchart showing a configuration procedure of the present invention (No. 2).

FIG. 5 is a schematic diagram illustrating a traveling vehicle detection method by the millimeter wave radar according to the present invention.

FIG. 6 is a schematic diagram for explaining a traveling vehicle detection method by the millimeter wave radar (N-1).

FIG. 7 is a schematic diagram for explaining a traveling vehicle detection method by the millimeter wave radar (N).

FIG. 8 is a conceptual diagram showing an example of the measurement data.

FIG. 9 is a conceptual diagram of the same segmentation.

FIG. 10 is a conceptual diagram of an attribute data structure when the same attribute is generated.

FIG. 11 is a structural diagram of the tracking data.

FIG. 12 is a conceptual diagram showing an example of a measurement result by the millimeter wave radar.

FIG. 13 is a detailed flowchart of the tracking process.

FIG. 14 is a conceptual diagram of prediction by the tracking.

FIG. 15 is a flowchart for recognizing the stopped / low-speed vehicle (part 1).

FIG. 16 is a flowchart of the same stop / low-speed vehicle recognition (part 2).

FIG. 17 is a conceptual diagram of the background data generation process.

FIG. 18 is a detailed diagram of the background data updating unit.

[Explanation of symbols]

1 Millimeter wave radar 2 millimeter wave sensor 3 Vehicle detection computer 4 Upper computer 5,6,7 vehicle 8 guardrails 9 obstacles

─────────────────────────────────────────────────── ─── Continuation of the front page (58) Fields surveyed (Int.Cl. ⁷ , DB name) G08G 1/00-1/16 G01S 13/34 G01S 13/91

Claims (5)

(57) [Claims] 1. A millimeter-wave radar installed on a road is used,
When tracking the vehicle for each scan, the position and speed are predicted using the tracking data obtained from the scans before the first scan,
Thus resulting position and velocity data and segmentation processes the resulting position and velocity data for the following second time scanning, and attribute generation process, and the resulting position and velocity data of each segment Traveling with a millimeter-wave radar characterized by performing a matching process of selecting a smaller one by comparing a distance with each of the candidates of the second time by a value obtained by dividing the distance by the number of measurement points of the first time and the number of measurement points of the second time. Vehicle detection method. 2. The traveling vehicle detection method by the millimeter wave radar according to claim 1, wherein the matching process is performed by focusing on the number of measurement points belonging to the same segment and the tracking data obtained by the second scanning. 3. A segment having a small number of measurement points obtained by the second scan when there is no match between the trace data obtained by the first scan and the trace data obtained by the second scan. Is judged as noise, and a segment with many measurement points is newly added to the tracking data.
A traveling vehicle detection method using the millimeter wave radar according to claim 1 . 4. When there is no match with the tracking data obtained by the first scanning due to the effect of shadowing that cannot be observed by the millimeter wave radar, etc., the position / velocity data and shadowing predicted in the tracking data are obtained. the number of times stored, and to claim 1 for holding the tracking data
A traveling vehicle detection method using the described millimeter wave radar. 5. The method for detecting a traveling vehicle by a millimeter wave radar according to claim 4 , wherein the observation noise of the tracking data caused by shadowing or the like is separated in the traveling direction and the vehicle width direction of the vehicle and then filtered.