KR20190083556A - Method for avoiding collision with a shielded moving object using time series sensing data - Google Patents

Abstract

The present invention relates to a method for avoiding a collision with a shielded moving object using time series sensing data and, more specifically, relates to a method for avoiding a collision with a shielded moving object using time series sensing data, which can avoid a collision by estimating a moving direction and a shape of a front moving object that is a shielded moving object and is hardly shown by being covered by a surrounding environment by using time series sensing data acquired with a detection means such as a radar sensor mounted on a vehicle.

Description

TECHNICAL FIELD [0001] The present invention relates to a shielded moving object collision avoidance method using time series sensing data,

The present invention relates to a shielded moving object collision avoidance method using time series sensing data, and more particularly, to a shielded moving object collision avoidance method using time series sensing data for estimating a moving direction and a shape of a forward moving object obscured by a surrounding environment, .

The automobile is provided with a safety device for the safety of the driver and the passenger, and a safety device for avoiding the collision with the obstacle by warning the driver of the obstacle located in the blind spot which the driver can not correctly recognize For example, a rear warning device that detects and alerts a rear obstacle.

Many of the related arts related to the rear warning device have been using only a detection means such as a camera to detect only precisely recognizable obstacles.

On the other hand, in the case of a moving object that is moving among the obstacles obstructing the running of the vehicle, it is often hidden by the surrounding environment.

In particular, in the case of a road without traffic lights or a residential area where the distinction between a road and an Indian road is unclear, the moving object may be hidden by the surrounding environment such as a parked vehicle, a street store, a pole, It is difficult to accurately detect the moving object by only the detection means installed in the vehicle as well as the driver of the vehicle in which the vehicle is running.

Korean Patent No. 10-1394862

SUMMARY OF THE INVENTION The present invention has been made in view of the above-mentioned problems, and it is an object of the present invention to provide a front moving vehicle which is hidden by the surrounding environment and is invisible by using a sensing data of time series obtained by using a detecting means such as a radar sensor, The present invention provides a method for avoiding a collision by estimating a moving direction and a shape of a mobile body by using time series sensing data.

Accordingly, in the present invention, a method of detecting a shielded moving body existing in front of a vehicle using a radar sensor mounted on a vehicle; Estimating a moving direction of the shielded moving body by using an electromagnetic wave signal of a radar sensor reflected from the shielded moving body; Determining whether or not the shielded moving object is obstructed in traveling of the vehicle on the basis of the arrangement relationship between the moving direction of the shielded moving object and the traveling direction of the vehicle; And a step of warning the driver if the shielded moving object is judged to be an obstacle to the driving of the vehicle. The present invention provides a method for avoiding collision of a shielded moving object using time series sensing data.

Specifically, in the process of estimating the moving direction of the shielded moving object, the moving direction of the shielded moving object is estimated based on the time series moving direction of the point cloud matched with the electromagnetic wave signals reflected from the shielded moving object, The moving direction of the shielded moving body can be estimated based on the time series moving direction of a point (feature point) selected from a point group matched with the signals.

According to the shielded moving object collision avoidance method of the present invention, if it is determined that the shielded moving object is obstructive to the running of the vehicle, if it is determined that the shielded moving object does not obstruct the running of the vehicle, The shielded moving object is re-detected, and if it is determined that the shielded moving object is an obstacle to the traveling of the vehicle, it is possible to prevent the collision with the shielded moving object, Provide information.

According to another aspect of the present invention, there is provided a shielding moving object collision avoiding method including estimating a shape of a shielded moving object by using an electromagnetic wave signal of a radar sensor reflected from the shielded moving object; And informing the driver of the shape of the shielded moving body to induce manual operation of the driver for avoiding collision with the shielded moving body,

Wherein the step of estimating the shape of the shielded moving object acquires time series of point cloud data from the electromagnetic wave signals of the radar sensor and, based on a plurality of points selected from the time series of point cloud data, .

According to another aspect of the present invention, there is provided a shielded moving object collision avoidance method for a shielded moving object collision avoidance method, the shielded moving object collided avoidance method comprising: And initializing the detected information.

According to the present invention, the moving direction and the shape of the shielded moving body present in front of the vehicle are estimated using the time series sensing data to predict the possibility of collision between the vehicle and the shielded moving body to inform the driver or automatically control the driving of the vehicle So that collision avoidance can be performed.

FIG. 1 is a conceptual view illustrating a shielded moving object collision avoiding method using time series sensing data according to the present invention.
2 is a conceptual diagram illustrating a moving direction and a shape estimation method of a shielded moving body using time series sensing data according to the present invention.
3 is a flowchart showing a method for avoiding a collision of a shielded moving object using time series sensing data according to the present invention
4 is a flowchart showing a method of estimating a shape of a shielded moving object using time series sensing data according to the present invention.

Hereinafter, the present invention will be described with reference to the accompanying drawings.

The present invention can be applied not only to the visibility of a driver but also to a vehicle (e.g., a vehicle) when a moving vehicle existing in front of the vehicle being driven or a vehicle being stopped can not be accurately recognized, It is possible to estimate the movement direction and the shape of the shielded moving object by using the time series sensing data so that the collision with the shielded moving object can be avoided.

For this purpose, in the present invention, as shown in Fig. 1, for a shielded moving body which is partially exposed to surrounding environments and other obstacles such as parked vehicles, street shops, electric poles, Time-series sensing data is secured by using a detection means such as a 3D radar sensor mounted on the vehicle, and the moving direction and shape of the shielded moving body can be estimated based on the secured time-series sensing data.

This makes it possible to avoid collision with the shielded moving body by predicting approaching of the shielded moving body obscured by other non-moving obstacles (i.e., non-moving bodies) located in the periphery.

As is known, a radar is an apparatus that receives electromagnetic waves reflected from the surface of a target object by radiating electromagnetic waves, and uses the reflected electromagnetic waves to measure a distance to the target object, a moving direction of the target object, The relative speed of the object can be measured.

In the present invention, an electromagnetic wave signal for a shielded moving object partially exposed through other obstacles is measured using a radar sensor of a vehicle to which the operating principle of the radar is applied, and time-series sensing data The moving direction of the shielded moving object is estimated through the point group processing of the point cloud data of the current point (see t + alpha in FIG. 2) and the shape estimated through the point cloud data at the previous point (see t in FIG. 2) And estimates the overall shape of the shielded moving body by continuously integrating the estimated shapes (see Fig. 2).

Hereinafter, the present invention will be described in more detail with reference to FIGS. 3 and 4 attached hereto.

As shown in FIG. 3, an object existing in front of the vehicle is first detected using detection means such as a 3D radar sensor mounted on a vehicle (a car) (S100). At this time, the vehicle may be stopped or running.

At this time, the radar sensor can receive an electromagnetic wave signal reflected from a surface of an object in a line scan method for detecting time series information only for a certain line (range) selected from a sensing range in front of the vehicle ).

Next, whether or not a moving object is present in the object detected within the sensing range of the detection means is determined based on the signal (radar signal) of the radar sensor (S110). In the case of the radar sensor, an electromagnetic wave signal (i.e., a radar signal) capable of discriminating whether or not an object moves can be provided by receiving an electromagnetic wave reflected from the surface of the object. In the in-vehicle control unit, It is judged whether or not to move.

At this time, if it is determined that there is no moving object in front of the vehicle, step S100 of detecting an object existing in front of the vehicle is performed again.

If it is determined that a moving object is present in the front of the vehicle, synthesis processing for estimating the shape of the moving object is started using time series data of the electromagnetic wave signal (i.e., radar signal) reflected from the surface of the moving object (S120).

At this time, in the case where the moving object is not a moving object (a shielded moving object) partially hidden by surrounding obstacles (i.e., non-moving objects), that is, an unshielded movement in which the entire shape is exposed, Since the shape of the moving object can be accurately confirmed, the synthesis processing for estimating the shape of the moving object using the time series sensing data is not started. Therefore, it can be seen that the object in which the synthesis processing for shape estimation is started in S120 is a shielded moving object.

In addition, in the synthesis process S120 for estimating the shape of the moving object, the shape of the shielded moving object may be synthesized using the synthesizing process of the point cloud acquired through feature point matching or time series sensing. In step S120, Will be described later with reference to FIG.

Synthesis processing for estimating the shape of the shielded moving object is started (S120), and a moving direction (VECTOR) of the shielded moving object is estimated (S130).

The electromagnetic wave signals of the radar sensor reflected from the shielded moving object are used to estimate the moving direction of the shielded moving object and the electromagnetic wave signals reflected from the shielded moving object are used to estimate the moving direction of the shielded moving object, The moving direction of the shielded moving object can be estimated based on the time series moving direction of the point cloud matched with the electromagnetic waves reflected from the shielded moving object.

That is, the movement direction of the feature point or the movement direction of the point cloud can be calculated based on the electromagnetic wave signal reflected from the shielded mobile body to estimate the movement direction of the shielded mobile body.

Here, the feature point may be a point selected from among a point cloud matched with the electromagnetic wave signals reflected from the shielded moving object.

Next, the traveling direction of the shielded moving body is compared with the traveling direction of the vehicle (the own vehicle), and it is determined whether or not the shielded moving body is obstructed to drive the own vehicle based on the relationship between the traveling direction of the shielded moving body and the traveling direction of the vehicle ).

For example, when the angle between the moving direction of the shielded moving body and the traveling direction of the vehicle lies within the set angular range, it is determined that the shielded moving object is an obstacle to the traveling of the vehicle, and the angle formed by the traveling direction of the shielded moving object and the traveling direction It can be determined that the shielded moving object does not obstruct the traveling of the own vehicle.

Specifically, when the moving direction of the shielded moving body and the traveling direction of the own vehicle are vertically arranged, it can be judged that the shielded moving body is an obstacle obstructing the running of the own vehicle. When the moving direction of the shielded moving body and the traveling direction of the own vehicle are arranged horizontally, it can be judged that the shielded moving body is a movement that does not hinder the traveling of the own vehicle.

At this time, if it is determined that the shielded moving object does not obstruct the driving of the own vehicle, step S100 of detecting an object existing in front of the vehicle is performed again. This is because the moving direction of the shielded moving body can be changed over time.

If it is determined that the shielded mobile object is an obstacle obstructing the driving of the own vehicle, the driver is warned that there is a risk of appearance of a front obstacle, that is, there is a front obstacle (i.e., a shielded moving object) (S150) to avoid collision with the shielded moving object (front obstacle) by performing automatic operation control through coordinated control with a safety device mounted on the vehicle or the like based on information on the front obstacle at the same time .

When the cooperative control with the safety device mounted on the vehicle is performed so that the automatic operation control avoiding the collision with the shielded moving object can be performed, information on the moving direction of the shielded moving object is provided to the safety device, For example, an automatic speed control device, an automatic steering device, or the like can be applied as the safety device.

Subsequently, it is determined whether the radar sensor senses the end of the shielded moving object based on the time series sensing data of the electromagnetic wave signal (that is, the radar signal) reflected from the surface of the shielded moving object (S160).

If it is determined that the radar sensor has not yet detected the end of the shielded moving object, step (S100) of detecting an object present in front of the vehicle is performed again, and the radar sensor completes detection until the end of the shielded moving object The synthesizing process for estimating the shape of the shielded moving object S120 is terminated and the detected information about the vehicle front object (including the shielded moving object) detected using the radar sensor is deleted and initialized.

Next, with reference to FIG. 4, a description will be made of a combining process (S120) for estimating the shape of the shielded moving object by using the time-series electromagnetic signals of the radar sensor reflected from the shielded moving object.

An ICP (Iterative Closest Point) algorithm can be used in the shape estimation of the shielded moving object, and the shape of the shielded moving object is estimated by synthesizing a point cloud using an open source point cloud library.

As shown in FIG. 4, point cloud data for a part of the shielded moving object, which is detected between obstacles at time t, is first acquired using the radar sensor of the own car (S200).

Next, point cloud data for a part of the shielded moving object detected between the obstacles at time t + alpha is acquired using the radar sensor (S210). That is, the point cloud data of the moving mobile body being moved is continuously acquired by sensing at the same observation point at intervals of a predetermined time (?).

That is, the point cloud data of the time series (t, t + alpha, t + 2 alpha, ...) is obtained from the electromagnetic wave signal of the radar sensor. Here,? Is a time interval for acquiring the point cloud data.

Next, a plurality of points are selected and extracted from the point group data (source side point group data) acquired at the time t and the point cloud data (destination side point cloud data) acquired at the time t + alpha, respectively (S220). At this time, a plurality of points can be selected and extracted by using a uniform sampling which selects and extracts regularly according to a predetermined condition, or a random sampling by randomly selecting and extracting.

 Next, a corresponding point between the source side point group data and the destination side point group data is obtained by the normal shooting method (S230).

According to the normal shooting method, the intersection point of the points intersecting the plane of the destination point data group mesh is extended by extending the normal of the most recently selected point among a plurality of points selected from the source side point group data And determines the intersection as the corresponding point of the destination side point group data for the source side point group data.

Here, the point cloud data mesh is a data mesh constructed by connecting points constituting point cloud data. That is, the point cloud data mesh is a data mesh consisting of points of point cloud data.

Then, a weight for the pair of the corresponding points is added (S240). For example, a weight proportional to the distance between two points is added to a pair that is paired with the corresponding point (i.e., the most recently selected point among a plurality of points selected from the source side point cloud data), or a weight A weight proportional to the inner product / length of the normal can be added.

The process of steps S200 to S240 is repeated several times (or a set number of times) for the point cloud data of the time series to obtain a plurality of corresponding points and pairs, and unnecessary pairs are deleted from the plurality of corresponding points and pairs (S250). For example, it is possible to set a threshold value of a constant value or set a worst value, and to delete an unnecessary pair exceeding a setting range based on the threshold value or the worst value. When the above-described steps S200 to S240 are repeated with respect to the time-series point cloud data, the point cloud data obtained at the time t + alpha is the source side point cloud data, and the point cloud data acquired at the time t + And becomes point cloud data.

When the unnecessary pair is deleted using the worst value as a reference, the distance between the corresponding point and the pair is deleted as an unnecessary pair since the distance is out of the set distance. For example, if a pair (n%) among the plurality of pairs is unnecessary As shown in FIG.

Subsequently, an erroneous vehicle is obtained between each corresponding point and the pair, and the erroneous vehicle is minimized (S260). For example, a point-to-point method may be used to calculate the error between a corresponding point and a pair, and to use select-match-minimize (which uses the translation and rotation of the current reference transformation matrix) ) Method, it is possible to minimize the error between the corresponding point and the pair to be less than the set threshold by repeating the process of finding a new conversion matrix in which the error becomes smaller by evaluating the error between the corresponding point and the pair.

More specifically, when calculating the error between the corresponding point and the pair using the point-to-point method, the square of the distance between the corresponding point and the corresponding point of the pair is calculated as the error value. In this case, the distance may be the distance of Euclid. When evaluating the erroneous vehicle between the corresponding point and the pair using the select-matching-minima method, a reference conversion matrix is set for the source-side point cloud data and the destination-side point cloud data, Rotation can be used to evaluate the correctness of the corresponding points and pairs.

Subsequently, the partial shape of the shielded moving body is synthesized by using a pair of corresponding points and the pair of corresponding points whose error is minimized, that is, a part shape of the shielded moving body is estimated through synthesis processing of a plurality of points, The entire shape of the shielded moving body can be estimated by integrating the partial shape of the shielded moving body continuously synthesized with the lapse of time (S270).

At this time, it is also possible to estimate the shape of the shielded moving body by using a program capable of estimating the shape using point cloud data composed of a plurality of corresponding points and pairs or by mounting artificial intelligence.

Alternatively, it is also possible to estimate the shape of the shielded moving body from the database of the shape of the vehicle moving body based on a plurality of points selected from the time series of point cloud data acquired from the electromagnetic wave signal of the radar sensor. The database may be pre-constructed so as to include point cloud data on various shapes of the shielded moving object and stored in the in-vehicle control unit, and the shape of the shielded moving object can be estimated by matching processing based on the data base. In addition, based on the estimated shape of the shielded moving object, it is possible to grasp the movement characteristics depending on the type and the type of the shielded moving object.

When the database is used, a moving object matching and matching based on the database is estimated and determined. At this time, a moving object having the highest matching rate among the moving objects included in the database can be selected and determined as a shielded moving object .

Accordingly, by notifying the driver of the type of the shielded mobile body before the driver visually recognizes the type of the shielded mobile body, the driver can estimate the risk, possibility of shape change, possibility of speed change, etc. of the shielded mobile body. It is possible to estimate the possibility of fluctuation of the movement direction of the shielded moving body based on the movement characteristics according to the type of the shielded moving body, thereby enabling the manual operation of the driver for avoiding collision with the shielded moving body to be guided, It is possible to secure an allowance distance and avoidance time with the moving object.

In addition, the in-vehicle control unit can warn the driver of the presence of the shielded moving object having the estimated shape and the possibility of collision in accordance with the moving direction of the shielded moving object, and the driver recognizing the warning can collide with the shielded moving object You can take action to avoid.

While the present invention has been particularly shown and described with reference to exemplary embodiments thereof, it is to be understood that the scope of the present invention is not limited to the disclosed exemplary embodiments. Modifications are also included in the scope of the present invention.

Claims (8)

Detecting a shielded moving body existing in front of the vehicle using a radar sensor mounted on the vehicle;
Estimating a moving direction of the shielded moving body by using an electromagnetic wave signal of a radar sensor reflected from the shielded moving body;
Determining whether or not the shielded moving object is obstructed in traveling of the vehicle on the basis of the arrangement relationship between the moving direction of the shielded moving object and the traveling direction of the vehicle;
A step of warning the driver if the shielded moving object is judged to be an obstacle to the running of the vehicle;
The method according to any one of claims 1 to 3,
The method according to claim 1,
Wherein the step of estimating the moving direction of the shielded moving object estimates the moving direction of the shielded moving object based on the time series moving direction of the point cloud matched with the electromagnetic waves reflected from the shielded moving object. Collision avoidance method.
The method according to claim 1,
Wherein the step of estimating the moving direction of the shielded moving object estimates the moving direction of the shielded moving object based on the time-series moving direction of a point selected from the point cloud matched with the electromagnetic wave signals reflected from the shielded moving object. Method for avoiding collision of shielded moving object using data.
The method according to claim 1,
Wherein when the shielded moving object is determined not to obstruct the running of the vehicle as a result of the determination whether the shielded moving object obstructs the running of the vehicle, the shielded moving object which is present in front of the vehicle is detected again using the radar sensor A method for avoiding collision of a shielded moving object using sensing data.
The method according to claim 1,
The method further comprises the step of initializing the detection information of the vehicle ahead object detected using the radar sensor if it is determined that the end of the shielded moving object is sensed based on the electromagnetic wave signal reflected from the shielded moving object Wherein the moving object collision avoidance method uses time series sensing data.
The method according to claim 1,
Further comprising the step of providing information on the direction of movement of the shielded moving body to a safeguard mounted on the vehicle so as to enable automatic operation control to avoid collision with the shielded moving body when it is determined that the shielded moving body obstructs the running of the vehicle Wherein the moving object collision avoiding method is characterized by using time series sensing data.
The method according to claim 1,
Estimating a shape of the shielded moving body by using an electromagnetic wave signal of a radar sensor reflected from the shielded moving body;
A step of informing the driver of the shape of the shielded moving body to induce manual operation of the driver for avoiding collision with the shielded moving body;
The method of claim 1, further comprising the step of:
The method of claim 7,
Wherein the step of estimating the shape of the shielded moving object acquires time series of point cloud data from the electromagnetic wave signals of the radar sensor and, based on a plurality of points selected from the time series of point cloud data, And estimating a shape of the shielded moving object by using the time series sensing data.

Images