JP2003276538A - Obstacle predicting device - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an obstacle predicting device capable of predicting an obstacle ahead of a preceding vehicle. <P>SOLUTION: On a computer device 30, a millimeter wave radar 10 capable of measuring vehicle-to-vehicle distance and relative speed, and an image sensor 20 are provided. Inside the computer device 30, a lateral move detecting part 37 to detect lateral move of the preceding vehicle, and an obstacle determining part 38 are formed. The lateral move detecting part 37 detects lateral move based on change of distance between an image of the preceding car picked up by the image sensor 20 and a lane sideways of it. For detection of that, change of distance on the image is multiplied by vehicle-to-vehicle distance/focal distance. It can thus be converted into actual moving distance whether the preceding car is far or close. In avoiding an obstacle, there are moving patterns of moving for a prescribed distance, stopping, returning, etc., and each is conducted within a prescribed time. The obstacle determining part 38 checks the pattern, lateral move quantity of the preceding car, the prescribed time, etc. The obstacle ahead of the preceding car can thus be predicted. <P>COPYRIGHT: (C)2004,JPO

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an obstacle predicting device for predicting an obstacle on a traveling road during traveling. In particular, the present invention relates to an obstacle prediction device that predicts an obstacle ahead of a preceding vehicle from a lateral movement pattern of the preceding vehicle. The present invention also relates to an obstacle prediction device that predicts the size of an obstacle from the lateral movement amount and the avoidance time. The present invention can be applied to an auto cruise device and an obstacle warning device.

[0002]

2. Description of the Related Art Conventionally, there are various devices for recognizing obstacles on a road. For example, Japanese Patent Laid-Open No. 2001-266160
The peripheral recognition method and the peripheral recognition device disclosed in the publication are examples thereof. This is equipped with an image processing device in the vehicle,
This is an example in which images are continuously captured, three-dimensional information is calculated from the loci of obstacle candidates in the series of image sequences, and the shapes of the candidates are recognized. For example, the relative position of the preceding vehicle and the lane on the traveling road hardly changes in one image sequence. On the other hand, as the obstacle approaches, its position moves and a large image is taken. Position change of this obstacle candidate,
And a method of calculating three-dimensional information of an obstacle from the camera posture at that time and determining whether or not the obstacle. In addition, the vehicle running control device disclosed in JP 2001-171389 A is also an example. This is because an image processing device is mounted on the host vehicle, the device captures an image of a preceding vehicle traveling on an adjacent traveling road, and the amount of approach of the preceding vehicle to the traveling route of the own vehicle from the image change,
This is a device that determines the lane changeability of an adjacent preceding vehicle by detecting the operation of the blinker, the reduction of the width of the adjacent traveling road, and the like. The purpose of the present invention is to optimize the travel control for the interruption of the adjacent vehicle, that is, to correct the drive shaft torque, acceleration / deceleration, inter-vehicle distance, acceleration limit value, inter-vehicle time, and the like.

[0003]

However, Japanese Unexamined Patent Application Publication No.
The peripheral recognition method and the peripheral recognition device disclosed in JP-A No. 01-266160 are effective when there is an obstacle ahead of the own vehicle, but cannot recognize it when there is an obstacle ahead of the preceding vehicle. That is, in order to avoid the obstacle, the preceding vehicle starts recognizing the obstacle in the front only after moving in the lateral direction, and there is a problem that the recognition process cannot be performed in time when the vehicle is traveling at high speed. In particular, when the obstacle is a thin plate-like obstacle, it is presumed that it is difficult to accurately recognize it. Also, Japanese Patent Laid-Open No. 2001-171389
The vehicle travel control device disclosed in the publication is a device for detecting a lateral one-way movement of a vehicle in front of left or in front of right, but its purpose is to recognize the possibility of interruption of the preceding vehicle. However, it is not a device that estimates an obstacle ahead of the preceding vehicle from its movement. That is, an obstacle ahead of the preceding vehicle cannot be predicted.

The present invention has been made to solve the above-mentioned problems, and its purpose is to detect a lateral movement of a preceding vehicle by an image sensor and predict an obstacle ahead of the preceding vehicle from the movement pattern. It is to provide an obstacle predicting device. Further, the approximate size of the obstacle is predicted based on the lateral movement amount and / or the time required from the start to the end of the pattern. Then, this device is mounted on various vehicle-mounted systems, for example, an auto cruise system to further improve traveling safety.

[0005]

An obstacle predicting apparatus according to claim 1 is an obstacle predicting apparatus that predicts an obstacle ahead of a preceding vehicle in vehicle traveling, and detects the preceding vehicle and its surroundings by an image. Image sensor, inter-vehicle distance measuring means for measuring the inter-vehicle distance to the preceding vehicle, and lateral movement detection for detecting the lateral movement of the preceding vehicle based on the detection result of the image sensor and the inter-vehicle distance measuring means. Means and an obstacle judging means for judging an obstacle in front of the preceding vehicle on the basis of the detection result of the lateral movement detecting means.

An obstacle predicting device according to a second aspect of the present invention is the obstacle predicting device according to the first aspect, wherein the obstacle judging means uses the movement pattern of the detection result of the lateral movement detecting means to detect the front of the preceding vehicle. It is characterized by determining obstacles. Further, the obstacle predicting device according to claim 3 is the obstacle predicting device according to claim 2, wherein the movement patterns judged by the obstacle judging means are predetermined when the lateral movement is viewed in time series. It is characterized by moving distance, stopping, and returning to the original running position.

The obstacle predicting device according to claim 4 is the obstacle predicting device according to claim 2 or 3.
The obstacle judging means obtains the relative speed from the vehicle distance measuring means,
It is characterized in that the size of the obstacle is judged from the lateral movement amount in the movement pattern and / or the time required from the start to the end of the movement pattern and the relative speed. Further, the obstacle predicting device according to claim 5 is the obstacle predicting device according to any one of claims 1 to 4, wherein the lateral movement detecting means includes a preceding vehicle and lanes on both sides thereof. The feature is that the amount of movement is detected from the distance. The obstacle predicting device according to claim 6 is the obstacle predicting device according to any one of claims 1 to 5, wherein the lateral movement detecting means is a moving amount average value of the preceding vehicle. It is characterized in that the standard deviation is calculated and the obstacle judging means judges the obstacle based on the average value and the standard deviation.

[0008]

According to the obstacle predicting apparatus of the present invention, an image sensor such as an image processing apparatus is mounted on the vehicle, and the image sensor detects the preceding vehicle and its surroundings. For example, the license plate of the preceding vehicle is detected as the vehicle center. Further, the inter-vehicle distance to the preceding vehicle is measured by the inter-vehicle distance measuring means. This is because the amount of lateral movement for avoiding an obstacle is small when the preceding vehicle is distant, and the amount of avoiding movement of the same amount is large when the preceding vehicle is approaching. That is, the inter-vehicle distance is measured for distance conversion described later. The inter-vehicle distance measuring means is, for example, a millimeter wave radar device, a laser radar device, or the like.

Then, the lateral movement detecting means detects the lateral movement amount of the preceding vehicle based on the detection result of the image sensor. This is calculated, for example, from the lateral displacement of the license plate of two images at predetermined time intervals, the inter-vehicle distance, and the camera parameter (focal length). For example, the movement amount =
It is calculated by the calculation of (positional deviation amount) × (inter-vehicle distance) / focal length. That is, whether the preceding vehicle is distant or close, it is converted into an equivalent amount (actual movement amount).

Finally, the obstacle judging means judges the obstacle based on the detection result of the lateral movement detecting means. For example, if the vehicle moves laterally by a predetermined amount and moves backward by a predetermined amount, it is determined that the preceding vehicle has avoided the obstacle. As a result, the present device can accurately announce the obstacle before the driver visually recognizes the obstacle. For example, when the obstacle has a thin plate shape, it is difficult to make the determination up to a close range. According to the present invention, an obstacle can be predicted by the amount of lateral movement of the preceding vehicle whether the vehicle is distant or approaching.

An obstacle predicting device according to a second aspect of the present invention is the obstacle predicting device according to the first aspect, wherein the obstacle judging means uses the movement pattern of the detection result of the lateral movement detecting means to detect the front of the preceding vehicle. Judge obstacles. When avoiding an obstacle ahead, there is a pattern in its lateral movement. For example, when the traveling road is two lanes, when a preceding vehicle finds an obstacle, it approaches the right or left lane and returns to the original position again after a predetermined time. Alternatively, only one wheel on one side crosses the lane and then returns to the original position again. That is, it is right movement or left movement within a predetermined time, or left movement or right movement within a predetermined time. If any of these movement patterns is used as an obstacle avoidance pattern, it can be distinguished from a simple lane change. Therefore, the obstacle can be predicted without error. Further, since the comparison is made based on the movement pattern, the driver can be more surely notified of the obstacle.

The obstacle predicting device according to a third aspect of the invention is the obstacle predicting device according to the second aspect, wherein the obstacle determining means sequentially moves a predetermined distance when the lateral movement is viewed in time series. The movement patterns for avoiding obstacles are stationary, returning to the original traveling position. When avoiding obstacles, when looking at the lateral movement in detail in time series, it means moving to the right or left for a predetermined distance, stopping at that position after movement, and returning to the original traveling position. There is. Therefore, if the movement consisting of these three elements is used as the movement pattern for obstacle avoidance, the obstacle ahead of the preceding vehicle can be predicted more reliably.

The obstacle predicting device according to claim 4 is the obstacle predicting device according to claim 2 or 3.
The obstacle judging means obtains the relative speed from the inter-vehicle distance measuring means. The relative speed can be obtained, for example, by measuring the time change of the inter-vehicle distance. Therefore, the vehicle speed of the preceding vehicle can be obtained by adding the relative speed to the vehicle speed of the host vehicle. Then, the size of the obstacle is determined from the amount of lateral movement in the movement pattern and / or the time required from the start to the end of the movement pattern and the vehicle speed of the preceding vehicle at that time. The lateral movement amount in the movement pattern is correlated with the width of the front obstacle. Therefore, the width of the obstacle can be predicted based on the lateral movement amount. Also, the product of the time required from the start to the end of the movement pattern and the vehicle speed of the preceding vehicle at that time is correlated with the length of the obstacle. Thereby, the obstacle length can be predicted. Then, if the obstacle determining means calculates the product of the lateral movement amount, the vehicle speed of the preceding vehicle, and the time required from the start to the end of the movement pattern, the size of the obstacle can be roughly estimated. At this time, the size of the obstacle does not include the dimension in the height direction, for example, the size of the obstacle is two-dimensional as viewed from above. Since the width, length, etc. of the obstacle can be predicted, it is possible to assist the driving more safely.

An obstacle predicting device according to claim 5 is the obstacle predicting device according to any one of claims 1 to 4, wherein the lateral movement detecting means is the preceding vehicle and both sides thereof. The amount of movement is detected from the distance to the lane. When the image sensor is, for example, an image processing device using an image sensor and the preceding vehicle is imaged, the lateral movement of the preceding vehicle is affected by the lateral movement of the own vehicle. That is, the movement amount of the preceding vehicle is a relative difference from the movement amount of the own vehicle. That is, the displacement of the preceding vehicle in the image does not always indicate an accurate absolute amount. Therefore, according to the present invention, the lateral movement detecting means is
The moving amount of the preceding vehicle is calculated from the absolute distance between the preceding vehicle on the image sensor and the lane (white line) on the side of the preceding vehicle. That is, the vehicle is not affected by it. As a result, the amount of movement of the preceding vehicle can be accurately detected. Also, the contrast between the lane (white) and the road surface (gray) is strong, and image processing is easy. Therefore, if the change in distance between the preceding vehicle (for example, the license plate) and the lane is obtained, the movement amount of the preceding vehicle can be calculated accurately and reliably.

An obstacle predicting device according to claim 6 is the obstacle predicting device according to any one of claims 1 to 5, wherein the lateral movement detecting means is a moving amount of the preceding vehicle. Calculate the mean and its standard deviation. The lateral vehicle position is not always constant. The position has an error (width) due to road conditions, cross winds, etc. Therefore, for example, the vehicle position is measured at regular time intervals, and the average value and standard deviation thereof are calculated. Then, the obstacle judging means judges the movement of the preceding vehicle (that is, the obstacle) based on the average value and the standard deviation. For example, it is determined whether the absolute value of the difference between the lateral movement amount to be determined and the average value is within the standard deviation (σ). If it is within the standard deviation, it is determined to be normal traveling, and if it is above the standard deviation, it is determined to be obstacle avoidance. Since the obstacle is judged based on the average value and the standard deviation of the movement amount, it is possible to reduce judgment errors. That is, the reliability of obstacle prediction can be improved.

[0016]

DESCRIPTION OF THE PREFERRED EMBODIMENTS (Embodiment) FIG. 1 shows an obstacle predicting apparatus of the present invention. The figure is a system configuration diagram. The obstacle predicting apparatus of this embodiment includes a millimeter wave radar 10 which is an inter-vehicle distance measuring means capable of detecting a relative speed and a distance, an image sensor 20 which captures an image of a preceding vehicle and its surroundings, and a computer apparatus 30. . This computer device 3
0 indicates a lateral movement detection unit 37 which is lateral movement detection means for detecting the lateral movement amount of the preceding vehicle from the output results of the millimeter wave radar 10 and the image sensor 20, and an obstacle is determined from the lateral movement amount. An obstacle determining unit 38 is formed.
The computer device 30 is, for example, a one-chip type microprocessor, as shown in FIG. 2, a ROM 31 in which an arithmetic processing program is written, a CPU 32 for executing the program, and a work area memory R for execution.
It is composed of an AM 33 and an I / O interface 34 for performing input / output with the outside. The millimeter wave radar 10 and the image sensor 20 are mounted on the front part of the own vehicle shown in FIG. 3 and detect the preceding vehicle F.

Next, the operation of the obstacle predicting apparatus of this embodiment will be described with reference to the flowchart of FIG. This embodiment is an example in which an obstacle ahead of the preceding vehicle is predicted during the follow-up control and is output to the driver and the vehicle speed control device (FIG. 1). The obstacle predicting apparatus of this embodiment starts from step s100 when the engine key (not shown) is turned on. First, step s1
At 00, the millimeter wave radar 10 detects the inter-vehicle distance s and the relative speed v with respect to the preceding vehicle. If the preceding vehicle is not detected, the process waits at step s100 until it is detected. Next, in step s102, the image sensor 20 captures an image of the preceding vehicle F and the lane K to the right of the preceding vehicle F as shown in FIG.
As a result, the rear part of the preceding vehicle F and the lane k on the right side thereof are imaged in the frame memory (FIG. 5). Then, the actual distance E between the preceding vehicle F and the lane K is obtained from the distance e between the two and the inter-vehicle distance s on the image.

The method of obtaining the distance E is as follows. First, the preceding vehicle frame W is formed from the vehicle width and vehicle height of the preceding vehicle F in the frame memory. And the preceding vehicle F that divides the vehicle width into two equal parts
The intersection point G (x ₀ , y ₀ ) between the center line of X and the bottom of the frame W is obtained. Next, the bottom of the frame W is extended in the horizontal direction, the intersection H (x ₁ , y ₀ ) with the side lane K is determined, and the difference e (= x ₁ −x ₀ ) between the two on the frame memory. Ask for. In detail,
The difference between the two is calculated by the number of pixels, and the number of pixels is converted into a distance. For example, conversion is performed at 10 μm / pixel. After that, the inter-vehicle distance s and the camera parameter f (focal length) are converted into an actual distance. That is, the magnification is multiplied to calculate the distance E (t) = (x ₁ −x ₀ ) × s / f at the time t.

Then, the process proceeds to step s104. In step s104, if there is past data _{E (t p) (t p} < current time t), is calculated and the average value A and a standard deviation sigma. This is to predict obstacles
This is to reduce judgment errors by using the average value A and the standard deviation σ as the threshold value. Next, step s106.
Then, it is determined whether the follow-up control is currently being performed. If no, the process returns to step s100 to update the average values A and σ. In the case of yes, it transfers to step s108 and it is determined whether the preceding vehicle F is a horizontal movement start. This means that, as shown in FIG. 6, whether or not the difference between the average value A and the current measured value E (t) is not less than the standard deviation σ, that is, | AE (t) |>
It is judged by the judgment formula of σ. In the case of no, since it is within the standard deviation, lateral movement has not started (normal driving).
Then, the process returns to step s100. In the case of yes, since there is a high possibility that the lateral movement starts, step s110
Move to.

In step s110, it is determined whether the movement start time is stored. That is, it is determined whether it is the first move. If the answer is no, it means that it is the first movement start, so the time t at that time is stored as the movement start time (pattern start time) T ₀ . Then, the process returns to step s106, and steps s106 to s110 are repeated. In step s110 of the second time, since the movement start time is already stored, the process proceeds to step s114.

In step s114, it is determined whether or not a predetermined time T ₁ has elapsed after the start of movement. The predetermined time T ₁ is, for example, the time required for the preceding vehicle to avoid an obstacle ahead. For example, as shown in FIG. 7, an obstacle Z is found in front of the preceding vehicle F, and the preceding vehicle F having a vehicle length L and a vehicle speed V (vehicle speed v ₀ + relative vehicle speed v) is a distance D before and after the obstacle Z. Assuming that avoidance is carried out at, the time T ₁ required for avoidance and the vehicle speed V,
The vehicle length L has the following relationship.

[Equation 1] T ₁ > (L + 2D) / V ... (1) Therefore, if the vehicle length L is roughly calculated from the vehicle width of the preceding vehicle F, the preceding vehicle is calculated by the equation (1). Predetermined time T that matches F
₁ can be calculated. For example, if the vehicle length L is 5 m, the obstacle avoidance distance D is 10 m, and the vehicle speed is 80 km / h, T ₁
= 0.675 seconds.

Then, in step s114, (current time t-T ₀ )> T ₁ is determined. If the answer is no, it means that the process is currently being avoided. Therefore, the process returns to step s106, and steps s106 to s114 are repeated. In the case of obstacle avoidance, during this period, in the state of | AE (t) |> σ,
The time t will change. Then, when the predetermined time T ₁ has passed, that is, in the case of yes, the process proceeds to step s116, and this time, the possibility of changing the lane of the preceding vehicle F (change of traveling path) is determined. This is because the lapse of a predetermined time T ₁ or more,
This is because the lane change of the preceding vehicle is included. Therefore, in the next step s116, the moving distance E that means crossing the lane K
(T) <0 and the elapse of a time longer than the predetermined time T ₂ is checked.
At this time, the predetermined time T ₂ is much longer than the predetermined time T ₁ . That is, the moving distance E (t) <0, and t> proceeds to step s118, if the predetermined time T _2, determines that a lane change, a step to capture the new preceding vehicle s1
Return to 00.

In step s116, the moving distance
E (t) <0 and t> predetermined time T ₂Otherwise, immediately
If no, there is a possibility of avoiding obstacles.
Up s120, preceding vehicle F returns to its original running position
It is determined whether or not. This return is | AE (t) |
It can be determined by whether or not it is σ. In the case of no, it is still avoiding
Therefore, the process returns to step s106. If yes, T₁
<T <T₂It means that it returned to the original position under the condition of
(End of movement pattern). This lateral movement within a predetermined time
Return to the original travel position of the
It's a turn. Since it follows this movement pattern, y
In the case of es, the process proceeds to step s122 to predict an obstacle.
Is output to, for example, the driver and the vehicle speed control device (FIG. 1). So
After that, the process proceeds to step s124. Step s124
Now that the obstacle prediction output is complete, the next obstacle prediction
To prepare for measurement, start time T₀And clear again
Return to step s100. Obstacle prediction device of this embodiment
Works this way.

As described above, the obstacle predicting apparatus of the present embodiment detects the lateral movement of the preceding vehicle by the image sensor, and predicts the obstacle ahead of the preceding vehicle from the movement amount and the movement pattern. Therefore, even if there is an obstacle ahead of the preceding vehicle, it can be predicted quickly. Further, when judging the movement of the preceding vehicle, it is judged based on the moving distance average and standard deviation, and a predetermined time from the start to the end of the movement pattern. Therefore, sudden movement is not judged as an obstacle. That is, the obstacle can be predicted stably. Therefore, if the obstacle prediction device of the present embodiment is mounted on an automatic cruise system, an obstacle warning device, etc., the safety of traveling can be further improved.

(Modification) An embodiment representing the present invention has been described above.
However, various other modified examples are possible. example
For example, in the above embodiment, the movement pattern of the preceding vehicle F is shown in FIG.
Lateral movement and a predetermined time T₁Within the return
However, the present invention is not limited to this. For example, in Figure 8
So that the predetermined time T₁Can be classified in more detail
Yes. That is, the three elements of moving, stationary, and returning are combined into one moving pattern.
You can also use it. That is, the rest time T after movement₁₂Measure
Then, it may be used as a criterion. In this case,
4 in place of step s114 in the flowchart of FIG.
Step s114A may be used as shown in FIG. That is,
In step s114A, after standing still for a predetermined time T₁₂But
It is determined whether or not it has passed. If not, that is,
If no, the process is in the process of avoiding, so the process proceeds to step s106.
To do. If yes, the next possibility is a lane change, or
Since it is a return, the process moves to step s116. Equivalent result
The fruit is obtained. Furthermore, in detail, from the start of movement to the standstill
Time T₁₁, Time from rest to recovery T ₁₃Even if you measure
Good. This travel time T₁₁, T₁₃Is related to the width of the obstacle
is there. Therefore, if this is used as a criterion, more detailed obstacles will be
You can predict harmful substances.

Although the size of the obstacle is not mentioned in the above embodiment, the obstacle predicting apparatus of the present invention can roughly predict the size. For example, in the above embodiment, the amount of change in the distance E, that is, the amount of lateral movement of the preceding vehicle has a correlation with the width of the front obstacle. Further, the predetermined time T ₁ in FIG. 7 and FIG. 8 (time from the movement pattern start to the end)
Is related to the length of the obstacle. That is, the obstacle length can be calculated by the product of the vehicle speed of the preceding vehicle (vehicle speed + relative speed) and the predetermined time T ₁ . Therefore, the product of the lateral movement amount, the vehicle speed of the preceding vehicle, and the predetermined time T ₁ is calculated, for example, in step s12 of FIG.
If the calculation is performed by 2, the size of the obstacle can be roughly estimated. Therefore, if the driver is informed of the obstacle information by voice in large, medium, and small, for example, the driver can respond appropriately.

In the above embodiment, the computer device 30 is used.
Although the example in which the vehicle speed control device is provided at the output stage to assist the auto-cruise control has been described, the output stage may be provided with an alarm device for giving a voice warning instead. If constituted in this way, it can be set as an obstacle warning device.

[Brief description of drawings]

FIG. 1 is a system configuration diagram of an obstacle prediction device according to an embodiment.

FIG. 2 is a block diagram of a computer device included in the obstacle prediction device according to the embodiment.

FIG. 3 is a millimeter wave radar of the obstacle prediction device according to the embodiment,
Layout of the image sensor.

FIG. 4 is a flowchart of an example showing the operation of the obstacle prediction device according to the embodiment.

FIG. 5 is a positional relationship diagram of a preceding vehicle and a lane imaged by the image sensor according to the embodiment.

FIG. 6 is a relationship diagram of the distance E to the lane for determining the start of movement, the average value A, and the standard deviation σ according to the embodiment.

FIG. 7 is a relationship diagram of a vehicle length L of a preceding vehicle, an obstacle avoidance distance D, and an avoidance time T ₁ during obstacle avoidance according to an embodiment.

FIG. 8 is a vehicle length L of a preceding vehicle during obstacle avoidance according to a modification, an obstacle avoidance distance D, avoidance times T ₁₁ , T ₁₂ , and T.
₁₃ relationships.

FIG. 9 is a flowchart modification unit of an obstacle prediction device according to a modification.

[Explanation of symbols]

10 ... Millimeter wave radar 20 ... Image sensor 30 ... Computer device 37 ... Lateral movement detection unit 38 ... Obstacle judgment part F ... Leading vehicle K ... Lane

   ─────────────────────────────────────────────────── ─── Continued front page    (72) Inventor Yuji Uchiyama             Aichi Prefecture Nagachite Town Aichi District             Ground 1 Toyota Central Research Institute Co., Ltd. (72) Inventor Shinki Shiraki             Aichi Prefecture Nagachite Town Aichi District             Ground 1 Toyota Central Research Institute Co., Ltd. F-term (reference) 5H180 AA01 CC04 LL01 LL02 LL04                       LL06 LL08

Claims (6)

[Claims] 1. An obstacle predicting device for predicting an obstacle ahead of a preceding vehicle when the vehicle is traveling, an image sensor for detecting an image of the preceding vehicle and its surroundings, and an inter-vehicle distance for measuring an inter-vehicle distance to the preceding vehicle. Based on the detection result of the distance measuring means and the image sensor and the measurement result of the inter-vehicle distance measuring means, the lateral movement detecting means for detecting the lateral movement of the preceding vehicle, and the detection result of the lateral movement detecting means An obstacle predicting device, comprising: an obstacle judging means for judging an obstacle in front of the preceding vehicle based on the obstacle predicting device. 2. The obstacle according to claim 1, wherein the obstacle judging means judges the obstacle in front of the preceding vehicle based on a movement pattern of a detection result of the lateral movement detecting means. Prediction device. 3. The movement pattern judged by the obstacle judgment means is, when the lateral movement is viewed in a time series, a predetermined distance movement, a stationary state, and a return to the original traveling position in order. The obstacle prediction device according to claim 2. 4. The obstacle determining means obtains a relative speed from the inter-vehicle distance measuring means, and based on the lateral movement amount in the movement pattern and / or the time required from the start to the end of the movement pattern and the relative speed. The obstacle prediction device according to claim 2 or 3, wherein the size of the obstacle is determined. 5. The lateral movement detecting means detects the amount of movement from a distance between the preceding vehicle and a lane on the side thereof, according to any one of claims 1 to 4. The obstacle prediction device described. 6. The lateral movement detecting means calculates an average value and a standard deviation of the moving amount of the preceding vehicle, and the obstacle judging means judges an obstacle based on the average value and the standard deviation. The obstacle prediction apparatus according to claim 1, wherein the obstacle prediction apparatus is performed.