JP3626879B2 - Road obstacle detection device, road obstacle detection method, road obstacle detection system, and recording medium recording road obstacle detection program - Google Patents

Description

[0001]
BACKGROUND OF THE INVENTION
The present invention relates to a road obstacle detection device, a road obstacle detection method, and a road obstacle detection system.AndThe present invention relates to a recording medium on which a road obstacle detection program is recorded.
[0002]
[Prior art]
On an expressway or the like, if there is an obstacle on the road, it is very dangerous for the vehicle to travel. Therefore, it is desired to detect the obstacle on the road and notify the driver. As a road obstacle detection system, an object on the road is detected by a sensor, and it is considered to identify whether the object is an obstacle based on whether the object is moving or stationary at a certain speed or higher. It is done.
[0003]
[Problems to be solved by the invention]
For example, when multiple sensors are installed along an expressway and an obstacle detection system that detects obstacles based on measurement data measured by each sensor is constructed, obstacles actually exist on the road However, it is conceivable that an obstacle cannot be detected by the sensor due to the shadow of the vehicle on which the obstacle travels or the obstacle becomes a shadow of another obstacle. In a system for detecting an obstacle on a road, how to detect an obstacle temporarily hidden by such other objects is an important theme.
[0004]
In addition, it is considered useful for road operation management to realize a system that measures the road conditions at regular intervals using the sensors described above and monitors the road conditions. In that case, if the data obtained from the measurement data is updated every time the road condition is measured by the sensor, the data update process is performed even when the car is flowing smoothly and the road condition hardly changes. It can be considered that the data update processing time for displaying the situation increases.
[0005]
An object of the present invention is to be able to detect obstacles hidden by other objects.The
[0006]
[Means for Solving the Problems]
The road obstacle detection device of the present invention detects and detects an obstacle based on a millimeter wave sensor that measures a road condition in a certain range, measurement data measured by the millimeter wave sensor, and background data. Detection means for recording the history of an obstacle as history information, and estimation means for estimating an obstacle that is registered in the history information and does not exist in the current measurement data as an obstacle temporarily hidden by another object And whether the obstacle estimated as the obstacle hidden by the estimation means exists in the area behind the measured object, and there is an obstacle hidden in the area behind the object. If you want toAbout the hidden obstaclesUpdating means for updating the background data based on data weighted to each of the data.
[0007]
According to the present invention, when it is estimated that an obstacle hidden by another object exists, the data is updated so that the obstacle hidden by the other object is not erased from the data obtained from the measurement data. Since it can be performed, it is possible to detect an obstacle that is a shadow of another object and cannot be measured by the sensor. In addition, even when an obstacle hidden by another object is measured again, the data obtained from the measurement data stores the hidden obstacle as it is, so the data is continuous, It will not be measured or not measured.
[0008]
The estimation of whether or not there is an obstacle hidden by another object is performed by, for example, storing attribute information of the obstacle detected in the past and history information indicating the detection history of the obstacle. If an obstacle having the same attribute information has been repeatedly detected more than a predetermined number of times at a measurement timing close to the present time, but the obstacle is not detected this time, the obstacle is hidden by another object. Estimated.
[0009]
In addition, obstacle detection is performed, for example, by calculating a difference between background data obtained from weighted data and measurement data.Yeah.
[0011]
DETAILED DESCRIPTION OF THE INVENTION
Embodiments of the present invention will be described below with reference to the drawings. FIG. 1 is a diagram showing a configuration of a system for monitoring road conditions according to an embodiment of the present invention. This system includes a millimeter wave sensor 11 attached to a drive unit capable of swinging rotation to measure a certain range of road conditions, and measurement data (distance, speed, angle data) measured by the millimeter wave sensor 11. The obstacle detection device 12 that creates display data indicating the road condition, and a message that informs the presence of the obstacle sent from the obstacle detection device 12 is displayed. And an information display board 13.
[0012]
The information display board 13 is installed along a nearby road, such as a place where a road with poor visibility is curved, or at a position crossing the road, to inform the driver of the vehicle that there is an obstacle ahead. belongs to.
[0013]
A number of millimeter wave sensors 11 are installed along the road, and the measurement data of each millimeter wave sensor 11 is transmitted from the obstacle detection device 12 to the central monitoring center that monitors the road condition by wire or wirelessly. Obstacles at each point on the road may be displayed on the display device, and the central monitoring center may display a message or the like informing the presence of the obstacle on the information display board 13.
[0014]
Next, the obstacle detection process in the road condition monitoring system having the above configuration will be described with reference to the flowchart of FIG.
The obstacle detection device 12 acquires measurement data indicating the road condition from the millimeter wave sensor 11 (FIG. 2, S11), and executes background data generation / update processing based on the measurement data. Since this background data generation / update process is related to the occlusion check in step S19, which will be described later, it will be described in detail after the explanation thereof is completed.
[0015]
Next, an object moving at a certain speed or higher in the measurement data is excluded (S13). As a result, measurement data in which stationary vehicles and stationary obstacles such as vehicles with extremely slow speeds, guardrails, speed signs, and the like remain are obtained.
[0016]
FIGS. 3A and 3B are diagrams showing the situation of the roads at the time of background data acquisition and measurement data acquisition. FIGS. 4A and 4B are explanations of background data and measurement data. FIG.
[0017]
FIG. 3A shows the state of the road when there are no vehicles and obstacles, and FIG. 3B shows the state of the road when vehicles or obstacles are present.
When the situation of the road where the vehicle of FIG. 3 (A) does not exist is measured by the millimeter wave sensor 11, the measurement data includes only the guardrail, and background data as shown in FIG. 4 (A) is obtained from the measurement data. Is generated.
[0018]
Assuming that the range indicated by the broken line in FIG. 4A is the scan range of the millimeter wave sensor 11, since there is no obstacle in this case, it is viewed from the millimeter wave sensor 11 at the intersection of the left broken line and the guard rail 31. The dot data obtained from the measurement data in the range from the point represented by the angle and distance to the point represented by the angle and distance viewed from the millimeter wave sensor 11 at the intersection of the broken line on the right side and the guard rail 31 is the background data. Stored as matrix data.
[0019]
When the situation of the road when the vehicle or obstacle A in FIG. 3 (B) is present is measured by the millimeter wave sensor 11, the measurement data is a part of the guardrail and the vehicle or vehicle as shown in black in FIG. 4 (B). The data indicates the obstacle A. In this case, since the central portion of the guardrail 31 becomes a shadow of the obstacle A, it is not measured by the millimeter wave sensor 11.
[0020]
FIG. 5 is a diagram showing a data matrix of background data, and FIG. 5 shows a state in which the background data of FIG. 4A is stored. The measurement data detected by the millimeter wave sensor 11 is obtained as angle and distance data with the position of the millimeter wave sensor 11 as the origin, and the number of dots in the measurement data is determined by the resolution of the millimeter wave sensor 11. A data matrix of the background data shown in FIG. 5 is generated from the dot angle and distance data. In FIG. 5, the vertical axis indicates the angle and the horizontal axis indicates the distance, but this does not mean that the vertical axis and the horizontal axis indicate the angle and the distance, respectively, and the measurement data indicates the angle and the distance. It is obtained as data represented by.
[0021]
Returning to FIG. 2, the difference between the measurement data obtained by removing the moving object at a certain speed or more and the background data is obtained (S14).
When the obstacle A exists, the measurement data includes an image of the obstacle A (L-shaped shape) seen from the millimeter wave sensor 11 and the guardrail 31 as shown in black in FIG. Assuming that the background data before the obstacle exists is as shown in FIG. 4A, if the difference between the two is taken, the center portion 31a of the guardrail 31 and the obstacle A are extracted. However, since the guardrail 31 is a portion that is present in the background data from the beginning, it is excluded from the obstacle, and only L-shaped data as shown in FIG. 6 is extracted as difference data.
[0022]
Next, the extracted difference data is grouped (segmented) (FIG. 2, S15). The grouping is to classify each set of black dots when the difference data includes a plurality of obstacle candidates. After grouping, isolated black dot data that does not belong to the group is removed (S16). This is because isolated data is considered to be noise or the like measured by the millimeter wave sensor 11.
[0023]
Next, attribute information of each obstacle candidate divided into groups is generated (S17). The attribute information is the position of the center of gravity (angle, distance) and speed calculated from the angle and distance data of each detection point of the obstacle (including the obstacle candidate) measured by the millimeter wave sensor 11. And the size of the obstacle. This attribute information is stored in association with information for identifying an obstacle (for example, a label name). The attribute information is not limited to the center of gravity, but may be any information that can identify the position of the obstacle.
[0024]
Next, history information of the detected obstacle candidate is generated (S18). In the history information generation process, the candidate information detected this time is compared with the attribute information of obstacles detected in the past. If there is an obstacle with matching attribute information in the past, it is defined as that obstacle. Recorded label names are recorded as history information at the time of measurement. If there is no obstacle with matching attribute information, a new label name is defined for the obstacle candidate, and the label name is recorded as history information. Here, the history information indicates an obstacle detection history, and is a record in which the label names defined for the obstacles are sequentially recorded corresponding to the respective scan timings.
[0025]
FIG. 7 is a diagram showing an example of the structure of history information. The history information is a pair of the number of scans indicating the number of scans by the millimeter wave sensor 11 and the label name of the obstacle detected at that time. Configured.
[0026]
In the example of FIG. 7, two obstacle candidates are detected at the time of the t-2th scan, and it is stored that the label names are L1 and L2. At the next t-1th scan, only the obstacle candidate corresponding to the label name L2 is detected, and the obstacle candidate corresponding to the label name L1 is not detected. However, the obstacle candidate is temporarily detected by another object. Since it is highly possible that it is hidden, the OC indicating the suspicion of occlusion is stored in the column of the label name L1. At the next t-th scan, since an obstacle candidate is newly detected in addition to the obstacle candidates of the label names L1 and L2, the label name L3 is stored.
[0027]
Returning to FIG. 2, it is checked whether there is an obstacle (occlusion) temporarily hidden by another object with reference to the history information and attribute information described above (S19). The occlusion check is a process for comparing the measurement data and the history information to check whether there is a traveling vehicle or the like that has not been measured in front. If the obstacle is not measured in this measurement, it is estimated that the obstacle is temporarily hidden by another object.
[0028]
Next, the presence or absence of an obstacle is detected (S20). Obstacle detection is performed by referring to history information based on whether or not an obstacle candidate having the same label name, that is, the same attribute information, has been repeatedly detected a certain number of times. If the same obstacle candidate is detected a certain number of times or more (in this embodiment, two or more times), it is determined that an obstacle has been detected.
[0029]
If an obstacle is detected in step S20 (S20, YES), the position of the center of gravity of the obstacle is calculated, and the angle and distance of the center of gravity are calculated (S21). Also, the speed of the obstacle is calculated from the speed of the center of gravity (S22). Further, the size of the obstacle is calculated from the angle and distance data of each detection point of the obstacle (S23).
[0030]
When the detection of the position, speed and size of the obstacle is completed, a message notifying the presence of the obstacle is displayed on the information display board 13 (S24). Further, when a display device is connected to the obstacle detection device 12, the state of the road including the obstacle is displayed on the display device.
[0031]
Further, the position of the detected obstacle and the size of the obstacle are output to the external device via the external I / F of the obstacle detection device 12 (S25). For example, a monitoring center that monitors road conditions receives data related to obstacles output via the external I / F of the obstacle detection device 12, and displays the monitoring center based on the data related to obstacles at each point on the road. The device can display the road conditions including obstacles. As a result, the monitoring center can grasp the presence or absence of an obstacle at each point on the road and inform the driver of the presence of the obstacle in advance by an information display board or the like provided along the road.
[0032]
The contents of the background data generation / update process in step S12 of FIG. 2 will be described with reference to the flowchart of FIG.
First, in step S31, it is determined whether or not an object (including an obstacle) is in a blurred range. The blur range is a predetermined area surrounding the object (area surrounding the obstacle in FIG. 10). When it is determined that the measurement data is in the blurring range (S31, YES), the process proceeds to step S32 and the background data update process is executed. If NO in step S31, the process proceeds to step S33 to determine whether the measurement data is behind the object, and whether there is no occlusion in the rear area. If there is no occlusion in the measurement data behind the object (S33, YES), the background data is not updated, and the process ends there. On the other hand, if there is occlusion even in the measurement data behind the object (S33, NO), the background data update process in step S32 is executed.
[0033]
Here, the background data update process will be described. In the present embodiment, in order to prevent the background data serving as a reference for obstacle detection from being affected even when the road condition suddenly changes. The background data is updated using data obtained by weighting a plurality of measurement data measured at different measurement timings. Specifically, the background data is obtained from the following equation.
[0034]
M^t= (1-α) M^t-1+ ΑW^t-1... (1)
M: Background data
W: Blur data
α: 0 ≦ α ≦ 1
t: Number of scans of the millimeter wave sensor 11
The blur data W is obtained from W = f (D). “D” is measurement data, and “f (D)” is an arithmetic expression for weighting a plurality of measurement data.
[0035]
By calculating background data from the above equation (1), it is possible to obtain background data in which the influence of sudden measurement data is suppressed.
Here, the contents of the occlusion check in step S19 and the background data generation / update process in step S12 will be described with reference to FIGS.
[0036]
FIG. 9A is a diagram showing objects such as vehicles and obstacles actually present on the road, and FIG. 9B is a diagram showing measurement data measured by the millimeter wave sensor 11 at that time. is there. In FIG. 9, the vertical axis represents the angle and the horizontal axis represents the distance. However, this does not mean that the vertical axis and the horizontal axis directly indicate the angle and the distance, respectively. It shows that it is obtained as data expressed by distance.
[0037]
In the area that is not measured by the millimeter wave sensor 11 and becomes a shadow of an object, it is usually unnecessary to update the background data. However, if there is a possibility that part or all of the obstacle is included in the shadow area of the object, the background data of that area must be updated and the obstacle that actually exists in that area The problem of not being included in the data arises.
[0038]
Therefore, in this embodiment, when it is estimated that there is an obstacle (occlusion) hidden in the shadow of the object, the background data update process for a predetermined region surrounding the hidden obstacle is performed. I did it. Thereby, since the obstacle which becomes a shadow of the object and is not measured by the millimeter wave sensor 11 is stored as background data, display data or the like for displaying the obstacle can be created using the background data.
[0039]
When an object A such as a vehicle, a stationary obstacle B, and a stationary obstacle C are present on the road in a positional relationship as shown in FIG. 9A, the object is located near the millimeter wave sensor 11 Although A and the obstacle B are measured by the millimeter wave sensor 11, the obstacle C behind the obstacle B is not measured as a shadow of the obstacle B.
[0040]
Since part of the obstacle B becomes a shadow of the object A, as shown in FIG. 9B, the obstacle in a state where part of the obstacle B is missing by the millimeter wave sensor 11 (referred to as obstacle B ′ Called) is measured. Then, in the occlusion check in step S19 in FIG. 2, the attribute information of the obstacle B ′ is compared with the attribute information of the obstacle measured in the past, and whether or not there is an occlusion hidden by another object. Is estimated. In this case, there is no obstacle whose attribute information is completely the same, but the obstacle B measured at the previous and previous scans is not measured this time from the history information, and the obstacle B ′ and the obstacle B are determined from the attribute information. Since it turns out that it exists in the substantially same position, obstruction B 'estimates that a part of obstruction B was concealed.
[0041]
If it is estimated that the obstacle B ′ is a part of the obstacle B hidden, the background of a predetermined area surrounding the obstacle B based on the attribute information of the obstacle B in step S12 of FIG. Update the data.
[0042]
The contents of the background data update process at this time will be described below with reference to FIG. Regions a and b surrounded by squares around the object A and the obstacle B in FIG. 10 indicate regions in which background data is updated in the present embodiment. In addition, areas d and e indicated by diagonal lines behind the object A and the obstacle B are areas behind the object A and the obstacle B, and are areas in which background data is not normally updated.
[0043]
When the obstacle B is estimated to be occlusion, in the background data generating / updating process in step S12, a predetermined region b surrounding the obstacle B is calculated from the size of the original obstacle B and the position of the center of gravity. Then, the background data in the area b is updated. The background data is updated from the measurement data at time t and the previous background data from the above-described equation (1). At that time, since the measurement data at time t is data in which a part of the obstacle B is missing, for example, the measurement data is corrected based on the attribute information of the obstacle B indicating the original size of the obstacle B ′. The background data including the entire obstacle B is updated by calculating the background data using the corrected measurement data.
[0044]
As a result, even if a part of the obstacle is hidden by another object, the background data includes the entire obstacle that actually exists, so it is possible to detect the obstacle more accurately. It becomes.
[0045]
Next, the occlusion check regarding the obstacle C and the update of the background data will be described with reference to FIG.
In this case, the occlusion check in step S19 in FIG. 2 shows that the obstacle C has been repeatedly measured in the previous measurement from the history information, and the obstacle C has not been measured in the current measurement data. The object C is presumed to be an occlusion temporarily hidden by another object.
[0046]
When the obstacle C is estimated to be occlusion, the size and the position of the center of gravity are acquired from the attribute information of the obstacle C in the background data generation / update process in step S12 of FIG. The background data in the predetermined area c is updated.
[0047]
For updating the occlusion background data, for example, using the data generated from the attribute information of the obstacle C instead of the data of the obstacle C in the current measurement data, the background data is generated from the above-described equation (1). Or may be realized by other methods.
[0048]
Thus, even when the obstacle C is hidden by another object, the history information includes the obstacle C that actually exists, and the obstacle C is not removed from the history information. High obstacles can be detected reliably.
[0049]
In the first embodiment described above, when some or all of the obstacles measured so far are not measured, it is estimated that these obstacles are temporarily hidden by other objects. Based on the obstacle attribute information, the background data of a predetermined area around the position where the obstacles existed is updated, so that the obstacles hidden by other objects are Can be saved as a part. This makes it possible to obtain accurate background data that includes obstacles that are likely to exist, so it is possible to accurately identify obstacles that were not measured as shadows of other objects using the background data. Can be detected and displayed.
[0050]
Next, a second embodiment of the present invention will be described with reference to FIG. This second embodiment is realized in the above-described road condition monitoring system. The road condition is detected by the millimeter wave sensor 11 at regular intervals, and the road condition is determined according to the detected road condition. The data update timing for displaying the status is changed.
[0051]
Road conditions can be classified into two states: a state where cars are flowing smoothly and a state where traffic is congested. Then, considering the change of the state, the road conditions can be classified into the following four.
[0052]
(1) The car is flowing smoothly, (2) The car is gradually flowing from the smoothly flowing state and transitioning to a congested state, (3) The car is gradually smooth from the congested state (4) There are four traffic jams that are continuing.
[0053]
Then, based on the speed of the vehicle detected by the millimeter wave sensor 11, a vehicle having a certain speed or higher is extracted from the measurement data, and these four states are monitored by monitoring the change in the number of vehicles exceeding the certain speed and the change in the average speed. It is judged which one is.
[0054]
Specifically, when the number of vehicles having a speed equal to or greater than a certain value is equal to or greater than a certain value and the number has not changed much, it is determined that the state in which the vehicle of (1) is flowing smoothly is continuing. Further, when the speed is above a certain level and the number of cars is decreasing and the average speed of the cars is also decreasing, it is determined that the vehicle flow gradually deteriorates as described in (2) above and the vehicle is transitioning to a traffic jam state. . Further, when the number of vehicles of a certain speed or more is increasing and the average speed is also increasing, it is determined that the vehicle flow is gradually changing from the traffic congestion state (3) to a state where the vehicle flow gradually improves. Furthermore, when there are almost no cars whose speed is above a certain level and the number has not changed, it is determined that the traffic congestion of (4) is continuing.
[0055]
When the road condition as described above is detected and the road condition is (1) above and the vehicle is flowing smoothly, the road condition does not change abruptly. Data indicating road conditions is updated at a period longer than the measurement period.
[0056]
In the example of FIG. 11, assuming that the data for one scan of the millimeter wave line sensor 11 is one frame, the above-described background data is updated every three frames, that is, every third scan.
[0057]
In the case other than the above (4), that is, the vehicle flow gradually deteriorates in (2), and the vehicle flow gradually improves from the traffic congestion state in (3) when transitioning to the traffic congestion state. When the state of traffic congestion is continuing during the transition to (4), the road condition may change, so the background data is updated for each frame.
[0058]
In the second embodiment, background data is not updated at every measurement timing, but the number of times of update processing is reduced when the road conditions do not change so much by thinning out the number of updates according to the road conditions. Thus, it is possible to reduce the processing time of the update process for the background data, the data for displaying the road condition, and the like.
[0059]
Further, the obstacle detection program according to the present invention or the road condition detection program of the second embodiment is stored in a recording medium 1201 such as a floppy disk or a CDROM shown in FIG. 1201 may be read by a medium driving device (floppy disk driver, CDROM driver, etc.) of an information processing device (personal computer or the like) 1202, the read program may be stored in a storage device 1203 such as a hard disk, and the program may be executed. . Alternatively, the program is stored in a storage device 1204 such as a hard disk of the information provider computer, transferred to the user information processing device 1202 by communication, and stored in the storage device 1203 on the user side to execute the program. May be. Further, the program recorded in the recording medium 1201 may have a partial function of the program described in the embodiment.
[0061]
The sensor to be used is not limited to the millimeter wave sensor 11 and may be any wavelength sensor as long as it can detect the speed, position, etc. of the object.
[0062]
【The invention's effect】
According to the present invention, even when an obstacle hidden by another object exists, for example, the data can be updated so that the obstacle hidden by the other object is not erased. It becomes possible to detect obstacles more accuratelyThe
[Brief description of the drawings]
FIG. 1 is a system configuration diagram of a road condition monitoring system.
FIG. 2 is a flowchart of obstacle detection processing.
FIGS. 3A and 3B are diagrams illustrating road conditions at the time of background data acquisition and measurement data acquisition. FIGS.
FIGS. 4A and 4B are explanatory diagrams of background data and measurement data. FIG.
FIG. 5 is a diagram showing a data matrix of background data.
FIG. 6 is a diagram illustrating an obstacle extracted from a difference between measurement data and background data.
FIG. 7 is a diagram showing a structure of history data.
FIG. 8 is a flowchart of background data generation / update processing.
FIGS. 9A and 9B are diagrams showing obstacles that actually exist and obstacles detected by sensors. FIG.
FIG. 10 is an explanatory diagram of an update range of background data.
FIG. 11 is an explanatory diagram of update processing according to the second embodiment;
FIG. 12 is an explanatory diagram of a recording medium.
[Explanation of symbols]
11 Millimeter wave sensor
12 Obstacle detection device
13 Information display board

Claims (11)

A millimeter wave sensor that measures road conditions in a certain range;
Detecting means for detecting an obstacle based on measurement data and background data measured by the millimeter wave sensor, and recording history of the detected obstacle as history information;
An estimation unit that is registered in the history information and estimates an obstacle that does not exist in the current measurement data as an obstacle temporarily hidden by another object;
When the obstacle estimated as the obstacle hidden by the estimation means is present in the area behind the measured object, and there is an obstacle hidden in the area behind the object The road obstacle further comprises update means for updating the background data based on data obtained by weighting the past background data and the data on the hidden obstacle with a predetermined weight, respectively. Detection device. The updating means includes data obtained by multiplying past background data by a weighting factor (1-α), and data obtained by multiplying past data on the hidden obstacle by a weighting factor α. The road obstacle detection device according to claim 1, wherein the added data is obtained as background data. When there is an obstacle obstructed in the area behind the object measured this time, the updating means determines the past background data of a predetermined range surrounding the hidden obstacle and the hidden obstacle. 3. The road obstacle detection device according to claim 1, wherein the background data is updated based on data weighted with a predetermined weight. The detecting means determines whether or not an obstacle having attribute information matching the attribute information of the detected obstacle exists in the history information, and when there is a matching obstacle, the same obstacle 4. The road obstacle according to claim 1, wherein when there is no obstacle having the same attribute information, the obstacle information is recorded as a new obstacle candidate in the history information. Detection device. Measure road conditions with millimeter wave sensors,
An obstacle is detected based on the measurement data and background data measured by the millimeter wave sensor, and the history of the detected obstacle is stored in the storage means of the computer as history information.
It is registered in the history information stored in the storage means, and an obstacle that does not exist in the current measurement data is estimated as an obstacle temporarily hidden by another object,
It is determined whether an obstacle that is estimated to be a hidden obstacle exists in the area behind the measured object, and if there is an obstacle that is hidden in the area behind the object, A road obstacle detection method comprising: updating background data based on data obtained by weighting the background data and the hidden obstacle data with predetermined weights. Data obtained by multiplying data obtained by multiplying past background data by a weighting factor (1-α) and data obtained by multiplying past data on the hidden obstacle by weighting factor α The road obstacle detection method according to claim 5, which is obtained as data. When there is a hidden obstacle in the area behind the object measured this time, a predetermined range of past background data surrounding the hidden obstacle and the data about the hidden obstacle are predetermined. The road obstacle detection method according to claim 5 or 6, wherein the background data is updated based on the weighted data. Detecting obstacles based on the measurement data indicating the road condition measured by the millimeter wave sensor and the background data, and storing the history of the detected obstacles in the storage means as history information;
Estimating an obstacle that is registered in the history information stored in the storage means and does not exist in the current measurement data as an obstacle temporarily hidden by another object;
When the obstacle estimated as the obstacle hidden in the estimation step is present in the area behind the measured object, and there is an obstacle hidden in the area behind the object Includes a recording of an obstacle detection program for causing a computer to execute a step of updating background data based on past weighted background data and data on the hidden obstacles with predetermined weights. Medium. Data obtained by multiplying data obtained by multiplying past background data by a weighting factor (1-α) and data obtained by multiplying past data on the hidden obstacle by weighting factor α 9. The recording medium according to claim 8, wherein the recording medium is obtained as data. If there is a hidden obstacle in the area behind the object measured this time, a predetermined range of past background data surrounding the hidden obstacle and the data about the hidden obstacle 10. The recording medium according to claim 8, wherein the background data is updated based on the weighted data. A millimeter wave sensor that measures road conditions in a certain range;
Detecting means for detecting an obstacle based on measurement data and background data measured by the millimeter wave sensor, and recording history of the detected obstacle as history information;
An estimation unit that is registered in the history information and estimates an obstacle that does not exist in the current measurement data as an obstacle temporarily hidden by another object;
When the obstacle estimated as the obstacle hidden by the estimation means is present in the area behind the measured object, and there is an obstacle hidden in the area behind the object The present invention further comprises an updating means for updating background data based on data obtained by weighting predetermined background data and past background data and data on the hidden obstacle , respectively. system.

Images