KR20190140491A - Method for generating predicted path data of objects, apparatus therefor, and unmanned ground vehicle - Google Patents

Abstract

According to one embodiment of the present invention, an optimal path for minimizing a collision risk with an object when a path of an unmanned vehicle is planned may generated by generating predicted path data in consideration of uncertainty according to the position, speed, direction angle information of an object around the unmanned vehicle and type of object obtained from environmental awareness information of an unmanned vehicle and displaying collision risk information between an unmanned vehicle and an object on a continuous space-time over a predicted path through online learning.

Description

METHOD FOR GENERATING PREDICTED PATH DATA OF OBJECTS, APPARATUS THEREFOR, AND UNMANNED GROUND VEHICLE}

TECHNICAL FIELD The present invention relates to autonomous driving of an unmanned vehicle, and more particularly, to a method, an apparatus, and an unmanned vehicle for generating predicted path data of an object around an unmanned vehicle.

In general, in order for an unmanned vehicle to minimize the risk of external collisions and to carry out safe autonomous driving, it must be expressed in a manner that can easily reflect the collision risk information within several tens of meters during the route planning, and service it in real time. In addition, in order to ensure stable autonomous driving in complex driving situations in which not only obstacles in the stationary state but also moving obstacles exist, based on several pieces of moving obstacle information acquired through environmental awareness every time, time and space are reflected by reflecting future prediction information including uncertainty of moving obstacles. The ability to predict the likelihood of a collision is essential.

In this regard, a technique for recognizing an absolute position of an obstacle and an unmanned vehicle and calculating a moving path of the moving obstacle to avoid collision with the moving obstacle at a future time has been proposed. Here, when calculating the movement path of the moving obstacle, the approximation is performed by the cubic function to reliably predict the position of the moving obstacle at a certain time.

However, since the conventional method does not include the uncertainty in the movement path of the moving obstacle, there is a problem that it is difficult to apply to the path optimization problem of minimizing the maximum risk due to the uncertainty.

(Patent literature)

Republic of Korea Patent No. 10-1207903 (Registration date November 28, 2012)

Therefore, according to an embodiment of the present invention, the prediction path data is generated with respect to the object around the unmanned vehicle obtained from the environmental awareness information of the unmanned vehicle in consideration of the position, the speed, the direction angle information, and the uncertainty according to the type of the object. By displaying the collision risk information between the unmanned vehicle and the object in continuous space-time on the prediction path through online learning, the predicted path data of the object is generated to generate the optimal path that minimizes the risk of collision with the object when planning the path of the unmanned vehicle. It is intended to provide a method, apparatus, and driverless vehicle to create.

A method of generating predicted path data of objects according to an embodiment of the present invention as described above, based on environmental recognition information about a driving path of an unmanned vehicle, movement information of at least one object around the unmanned vehicle And obtaining a type of the object, and generating prediction path data of the object based on the motion information of the object and the uncertainty according to the type of the object.

The environmental recognition information may be obtained using an environmental recognition sensor mounted on the unmanned vehicle.

The motion information of the object may include at least one of a position, a speed, and a direction angle of the object.

The generating of the prediction path data may include generating a plurality of prediction path data at predetermined time intervals.

In addition, when the type of the object is a pedestrian, the uncertainty coefficient associated with the uncertainty is set to be lower than a predetermined reference value.

In addition, when the type of the object is a vehicle, the uncertainty coefficient associated with the uncertainty is set to be larger than a predetermined reference value.

The generating of the prediction path data may include calculating an average and a variance of the plurality of prediction path data generated at each predetermined time interval, and corresponding to the prediction path data based on the average and the variance. The method may further include generating and storing label information indicating whether or not the risk is at the absolute coordinate of the position.

The method may further include calculating a collision probability between the unmanned vehicle and the object with respect to the continuous space between the absolute coordinates based on the label information.

)로 구성된 로지스틱 회귀 분류기를 이용하여 산출되는 것을 특징으로 한다.In addition, the collision probability is a weight assigned to a plurality of sparse kernels and each sparse kernel with respect to absolute coordinates.

It is characterized by using a logistic regression classifier consisting of).

In addition, the weight is characterized by being updated through online learning using stochastic gradient descent (SGD).

The method may further include generating and displaying risk information of the object based on the collision probability.

In addition, an apparatus for generating prediction path data of an object according to an embodiment of the present invention, based on environmental recognition information on a driving route of an unmanned vehicle, motion information of at least one object around the unmanned vehicle and the object And an object path predictor configured to generate a predicted path data of the object based on the motion information of the object and the uncertainty according to the type of the object.

The motion information of the object may include at least one of a position, a speed, and a direction angle of the object.

The object path predictor may generate a plurality of prediction path data at predetermined time intervals.

The object path predictor may generate the prediction path data by setting an uncertainty coefficient related to the uncertainty lower than a predetermined reference value when the type of the object is a pedestrian.

The object path predictor may generate the predicted path data by setting an uncertainty coefficient related to the uncertainty greater than a predetermined reference value when the type of the object is a vehicle.

The object path predictor may be further configured to calculate an average and a variance of the plurality of prediction path data generated at the predetermined time intervals, and based on the average and the variance, in the absolute coordinates of the position corresponding to the prediction path data. And generating and storing label information indicating whether or not the risk.

The apparatus may further include an object risk generation unit configured to calculate a collision probability between the unmanned vehicle and the object with respect to the continuous space between the absolute coordinates based on the label information.

)로 구성된 로지스틱 회귀 분류기를 이용하여 상기 충돌 확률을 산출하는 것을 특징으로 한다.The object risk generator may include a plurality of sparse kernels and weights assigned to each sparse kernel with respect to absolute coordinates.

The collision probability is calculated using a logistic regression classifier consisting of a).

In addition, as an unmanned vehicle according to an embodiment of the present invention, the unmanned vehicle includes a device for generating predictive path data of an object, and the device for generating predictive path data of the object is a driving path of the unmanned vehicle. On the basis of the environmental recognition information about the at least one object around the unmanned vehicle and the object recognition unit for obtaining the type of the object, based on the motion information of the object and the uncertainty according to the type of the object, And an object path predictor for generating prediction path data of the object.

According to an embodiment of the present invention, for the object around the unmanned vehicle obtained from the environmental awareness information of the unmanned vehicle, generating prediction path data in consideration of the position, velocity, direction angle information and uncertainty of the type of the object, By displaying the collision risk information between the unmanned vehicle and the object in the continuous space-time on the prediction path through the online learning, it is possible to generate the optimal path to minimize the collision risk with the object when planning the path of the unmanned vehicle.

1 is a detailed block diagram of an apparatus for generating prediction path data of an object according to an embodiment of the present invention.
2 to 6 are diagrams illustrating the predicted movement position and the risk information of the object for each time interval predicted by the apparatus for generating prediction path data of the object according to an embodiment of the present invention.
7 is a flowchart illustrating an operation control in the apparatus for generating prediction path data of an object according to an embodiment of the present invention.

Hereinafter, with reference to the accompanying drawings will be described in detail the operating principle of the present invention. In the following description of the present invention, when it is determined that a detailed description of a known function or configuration may unnecessarily obscure the subject matter of the present invention, the detailed description thereof will be omitted. Terms to be described later are terms defined in consideration of functions in the present invention, and may be changed according to intentions or customs of users or operators. Therefore, the definition should be made based on the contents throughout the specification.

1 illustrates a detailed block diagram of an apparatus for generating prediction path data of an object according to an embodiment of the present invention.

Hereinafter, an operation of each component of the apparatus 100 for generating prediction path data of objects according to an embodiment of the present invention will be described in detail with reference to FIG. 1.

First, the object recognizing unit 110 recognizes the absolute positions of at least one or more objects positioned on the driving path of the unmanned vehicle and the unmanned vehicle, and recognizes the movement information about the object and the type of the object through the environmental recognition information. In this case, the motion information of the object may include information such as the position, speed, and direction angle of the object, the type of the object may be a pedestrian, a vehicle, and the like, and the environmental recognition information may be a camera mounted on an unmanned vehicle. It may be obtained through environmental recognition sensors such as Lidar and Radar, and provided to the object recognizing unit 110, but is not limited thereto.

The object path predictor 120 generates the predicted path data of the object based on information obtained from the object recognizer 110, such as the position, velocity, and direction of the object, and uncertainty associated with the type of the object.

In addition, in generating the prediction path data of the object, the object path predictor 120 may generate a plurality of prediction path data at predetermined time intervals.

In this case, the type of object may be classified into a case where the object is a pedestrian and a vehicle, but is not limited thereto.

In addition, when generating the prediction path data, the object path predictor 110 may generate the prediction path data by setting an uncertainty coefficient related to the uncertainty of the object lower than a predetermined reference value when the type of the object is a pedestrian. In addition, when the type of the object is a vehicle, prediction path data may be generated by setting an uncertainty coefficient related to the uncertainty of the object to be larger than a predetermined reference value. At this time, since the vehicle has a relatively low probability of rapidly changing speed, direction angle, etc. at predetermined time intervals, compared to the pedestrian, the uncertainty coefficient is set lower than the predetermined reference value for the vehicle.

In addition, the object path predictor 120 calculates an average and a variance of the plurality of predicted path data generated at predetermined time intervals, and whether the risk is at an absolute coordinate of a position corresponding to the predicted path data based on the average and the variance. Label information indicating a may be generated.

Hereinafter, an operation of generating the prediction path data of the object in consideration of the uncertainty in the object path predictor 120 will be described in more detail.

The object path predictor 120 generates the predicted path data of the object based on information obtained from the object recognizer 110, such as the position, speed, and direction of the object, and uncertainty associated with the type of the object.

스텝에서 대상체의 타입에 따라

개의 샘플 경로에 대해 불확실성을 고려한 예측 경로 데이터를 생성한 후, 샘플링된 예측 경로 데이터로부터 각 스텝에서의 평균과 분산을 계산하는 과정을 나타낸다. Pseudo codes as shown in [Table 1] below are all divided by a predetermined time interval in the object path predictor 120.

Depending on the type of object in the step

After generating prediction path data in consideration of uncertainty for the three sample paths, the process of calculating the mean and the variance at each step from the sampled prediction path data is shown.

TABLE 1

값으로

사이의 균등 난수를 생성하여 대상체의 예측 경로 데이터에 불확실성을 부여한다. 여기서,

는 불확실성 강도조절 계수이며 시간 간격

는 0.1초로 정한다. 이때, 본 발명의 일실시예에서는 위 시간 간격을 위에서 보여지는 바와 같이"0.1"로 설정한 것을 예를 들어 설명하나 이에 한정되는 것은 아니다.As shown in Table 1 above, the object path predictor 120 is

By value

Generate even random numbers between to give uncertainty to the subject's prediction path data. here,

Is the uncertainty intensity factor and the time interval

Is set to 0.1 second. At this time, in the embodiment of the present invention, the time interval is set as "0.1" as shown above, for example, but is not limited thereto.

번째 스텝에서의 평균과 분산에 따라 아래의 [수학식 1]에서와 같은 정규분포로부터

개의 대상체 각각에 대한 2차원 공간상에 위치

정보와 해당 위치에서의 무인 차량과 대상체간 충돌 등과 같은 위험 여부에 대한 레이블 정보

를 저장한다.In addition, the object path predictor 120 is obtained as described above.

From the normal distribution as in Equation 1 below, depending on the mean and the variance in the first step

In two-dimensional space for each of the two objects

Information and label information about the hazard, such as collision between unmanned vehicles and objects at that location

Save it.

2 to 6 illustrate the predicted movement position and risk information of an object for each time interval predicted by the apparatus 100 for generating prediction path data of an object, according to an embodiment of the present invention.

Figure 2 shows the position and the risk of the object obtained in the first step (0.1 seconds).

Referring to FIG. 2, the initial positions 200 and 210 in the two-dimensional space of the first object are x (east) = 25 and y (north) = 22, the speed is 1 m / s, and the direction angle is ?? 90 °. The initial position in the two-dimensional space of the second object is x (east) = 12, y (north) = 15, the speed is 4m / s, and the direction angle is + 90 °.

3 to 6 show estimated positions of the first object and the second object in the 10th step (1 second), the 20th step (2 seconds), the 30th step (3 seconds), and the 40th step (4 seconds), respectively. (200 ', 210') and prediction path data and risk.

6 to 6 show the predicted path data and the risk of the first object and the second object in the 40th step, and are the predicted path data of each of the D objects indicated by black dots. The yellow light indicates the risk based on the probability of collision with an unmanned vehicle.

As shown in FIG. 6, in the exemplary embodiment of the present invention, uncertainty of each object is taken into consideration so that prediction path data is randomly distributed from the initial positions 200 and 210 of the respective objects to the expected positions 200 'and 210'. In this case, when the above information is used, the unmanned vehicle can more accurately calculate a path that does not collide with the object.

Next, the object risk generator 130 performs online learning based on the predicted path data generated by the object path predictor 120 and label information indicating whether or not the risk is at an absolute coordinate of a position corresponding to the predicted path data. The collision probability between the unmanned vehicle and the object in the continuous space between the absolute coordinates is calculated, and the risk of the object in the continuous space-time is generated and displayed based on the collision probability. Such a risk may mean a collision probability between an unmanned vehicle and an object, and as shown in FIG. 6, the risk may be displayed highest at the center point of the object, and may be displayed as the distance is farther from the center point.

)로 구성된 로지스틱 회귀 분류기(Logistic regression classifier)를 이용하여 충돌 확률을 산출할 수 있다. 또한, 대상체 위험도 생성부(130)는 SGD(stochastic gradient descent)를 이용한 온라인 학습을 통해 가중치를 갱신할 수 있다.In this case, in calculating the collision probability, the object risk generator 130 may assign a plurality of sparse kernels and weights assigned to each sparse kernel with respect to absolute coordinates corresponding to continuous spaces in the prediction path data.

The probability of collision can be calculated using a logistic regression classifier composed of In addition, the object risk generation unit 130 may update the weight through online learning using stochastic gradient descent (SGD).

Hereinafter, an operation of generating the risk of the object on the driving route based on the predicted path data generated by the object path predictor 120 in the object risk generator 130 will be described in detail.

과 학습이 필요한 가중치(

)로 구성된 로지스틱 회귀 분류기를 통해 온라인 학습을 수행하여 무인 차량과 대상체간 충돌 확률을 산출하고, 충돌 확률을 기초로 연속적인 시공간상 대상체의 위험도를 표시한다. 이때, 대상체 위험도 생성부(130)는 연속 공간상에 충돌 확률에 기반한 위험도를 정의하기 위해 Hilbert maps 기법을 활용할 수 있으나, 이에 한정되는 것은 아니다.The object risk generation unit 130 is based on the label information indicating whether the collision risk on the prediction path data and the prediction path data of the object, as shown in [Equation 2] below Sparse kernel

And weights that require learning (

On-line learning is performed through a logistic regression classifier consisting of) to calculate the collision probability between the unmanned vehicle and the object, and to display the risk of the subject in continuous space-time based on the collision probability. In this case, the object risk generator 130 may use the Hilbert maps technique to define the risk based on the collision probability in the continuous space, but is not limited thereto.

은 아래의 [수학식 3]과 같이 나타낼 수 있다.Where sparse kernel

Can be expressed as shown in Equation 3 below.

)를 학습하기 위한 목적함수

를 아래의 [수학식 4]에서와 같이 Negative log-likelihood로 정의할 수 있다.In addition, the object risk generator 130 may include a weight (

Objective function for learning

Can be defined as negative log-likelihood as shown in [Equation 4] below.

)를 점진적으로 갱신시킬 수 있다.Here, the object risk generation unit 130 uses Stochastic Gradient Descent (SGD) as shown in Equation 5 below to obtain a weight (

) Can be updated gradually.

는 아래의 [수학식 6]과 같이 [수학식 7]의 공간에 대한 학습율과 [수학식 8]의 시간에 대한 학습율의 곱으로 정의될 수 있다.Where learning rate

May be defined as the product of the learning rate for the space of [Equation 7] and the learning rate for time of [Equation 8] as shown in [Equation 6] below.

는 강도조절 계수이고

는 감소 비율이다. 학습된

는 전체

개의 각

번째 스텝에서의 대상체의 위험도 정보는 무인 차량의 경로 계획 시 비용함수로 활용될 수 있다. here

Is the intensity control factor

Is the rate of decrease. Learned

Is full

Angles

Risk information of the object in the first step may be used as a cost function when planning the route of the unmanned vehicle.

7 illustrates an operation control flow in the apparatus for generating prediction path data of an object according to an embodiment of the present invention. Hereinafter, an embodiment of the present invention will be described in detail with reference to FIGS. 1 to 7.

First, the apparatus 100 for generating a prediction path data of an object recognizes an absolute position of at least one object located on a driving path of an unmanned vehicle and an unmanned vehicle (S700), motion information about the object obtained through environment recognition information, and Based on the uncertainty associated with the type of the object, prediction path data of the object is generated (S702).

At this time, the information on the object may include information such as the type, location, speed, direction angle of the object, and the above environmental recognition information such as camera (Camera), Lidar (Radar), Radar (mounted on the unmanned vehicle) It may be obtained through environmentally aware sensors, such as), but is not limited thereto.

Here, in generating the prediction path data, the apparatus 100 for generating prediction path data of the object analyzes information such as the type, speed, and direction angle of the object based on the environmental recognition information, and predicts a plurality of predictions at predetermined time intervals. You can create route data.

In addition, the apparatus 100 for generating prediction path data of the object calculates an average and a variance of the plurality of prediction path data generated at predetermined time intervals, and based on the average and the variance, the absolute path of the position corresponding to the prediction path data is calculated. Label information indicating whether or not the risk can be generated (S704). In this case, the above risk may represent a collision risk between the unmanned vehicle and the object, but is not limited thereto.

Subsequently, the apparatus 100 for predicting path data generation of the object may perform online learning based on the predicted path data generated by the object path predictor 120 and label information indicating whether or not the risk is at an absolute coordinate of a position corresponding to the predicted path data. Next, the collision probability between the unmanned vehicle and the object in the continuous space between the absolute coordinates is calculated (S706), and the risk of the object in the continuous space-time is generated and displayed based on the collision probability (S708).

In this case, the risk may mean a collision probability between the unmanned vehicle and the object, and may be displayed highest at the center point of the object as shown in FIG. 6, and may be displayed as far away from the center point.

As described above, according to an embodiment of the present invention, the prediction path considering the position, velocity, direction angle information and uncertainty of the object type relative to the object around the unmanned vehicle obtained from the environmental awareness information of the unmanned vehicle By generating the data and displaying the collision risk information between the unmanned vehicle and the object in continuous space-time on the predicted path through online learning, it is possible to create the optimal route that minimizes the risk of collision with the object when planning the path of the unmanned vehicle. have.

Combinations of the steps of each flowchart attached to the present invention may be performed by computer program instructions. These computer program instructions may be mounted on a processor of a general purpose computer, special purpose computer, or other programmable data processing equipment such that the instructions performed through the processor of the computer or other programmable data processing equipment are described in each step of the flowchart. It will create a means to perform them. These computer program instructions may be stored in a computer usable or computer readable memory that can be directed to a computer or other programmable data processing equipment to implement functionality in a particular manner, and thus the computer usable or computer readable memory. It is also possible for the instructions stored therein to produce an article of manufacture containing instruction means for performing the functions described in each step of the flowchart. Computer program instructions may also be mounted on a computer or other programmable data processing equipment, such that a series of operating steps may be performed on the computer or other programmable data processing equipment to create a computer-implemented process to create a computer or other programmable data. Instructions for performing the processing equipment may also provide steps for executing the functions described in each step of the flowchart.

In addition, each step may represent a module, segment or portion of code that includes one or more executable instructions for executing a specified logical function (s). It should also be noted that in some alternative embodiments, the functions noted in the steps may occur out of order. For example, the two steps shown in succession may in fact be performed substantially simultaneously or the steps may sometimes be performed in the reverse order, depending on the function in question.

Meanwhile, in the above description of the present invention, specific embodiments have been described, but various modifications may be made without departing from the scope of the present invention. Therefore, the scope of the invention should be determined by the claims rather than by the described embodiments.

100: apparatus for generating prediction path data of an object
110: object recognition unit
120: object path predictor
130: object risk generator

Claims (20)

A method of generating prediction path data of objects,
Acquiring motion information of at least one or more objects around the unmanned vehicle and the type of the object based on environment recognition information on a driving route of the unmanned vehicle;
Generating prediction path data of the object based on the motion information of the object and the uncertainty according to the type of the object.
Method of generating prediction path data of an object. The method of claim 1,
The environment recognition information is obtained using an environment recognition sensor mounted on the unmanned vehicle.
Method of generating prediction path data of an object. The method of claim 1,
The motion information of the object includes one or more of the position, velocity, and direction angle of the object.
Method of generating prediction path data of an object. The method of claim 1,
The generating of the prediction path data may include:
Generating a plurality of prediction path data every predetermined time interval.
Method of generating prediction path data of an object. The method of claim 4, wherein
When the type of the object is a pedestrian, the uncertainty coefficient associated with the uncertainty is set lower than a predetermined reference value.
Method of generating prediction path data of an object. The method of claim 4, wherein
When the type of the object is a vehicle, an uncertainty coefficient related to the uncertainty is set to be larger than a predetermined reference value.
Method of generating prediction path data of an object. The method of claim 4, wherein
The generating of the prediction path data may include:
Calculating an average and a variance of the plurality of prediction path data generated at each predetermined time interval;
Generating and storing label information indicating whether a risk is at an absolute coordinate of a position corresponding to the prediction path data based on the mean and the variance;
Method of generating prediction path data of an object. The method of claim 7, wherein
Calculating a collision probability of the unmanned vehicle and the object with respect to the continuous space between the absolute coordinates based on the label information;
Method of generating prediction path data of an object. The method of claim 8,
The collision probability is a weight assigned to a plurality of sparse kernels and each sparse kernel with respect to absolute coordinates.

Calculated using a logistic regression classifier
Method of generating prediction path data of an object. The method of claim 9,
The weight is updated through online learning using stochastic gradient descent (SGD).
Method of generating prediction path data of an object. The method of claim 10,
Generating and displaying risk information of the object based on the collision probability;
Method of generating prediction path data of an object. An apparatus for generating prediction path data of an object,
An object recognition unit obtaining motion information of at least one or more objects around the unmanned vehicle and the type of the object based on environment recognition information on a driving route of the unmanned vehicle;
And an object path predictor configured to generate prediction path data of the object based on the motion information of the object and the uncertainty according to the type of the object.
Object prediction path generation device. The method of claim 12,
The motion information of the object includes one or more of the position, velocity, and direction angle of the object.
Object prediction path generation device. The method of claim 12,
The object path predictor generates a plurality of prediction path data at predetermined time intervals.
Object prediction path generation device. The method of claim 14,
The object path predicting unit generates the prediction path data by setting an uncertainty coefficient related to the uncertainty lower than a predetermined reference value when the type of the object is a pedestrian.
Object prediction path generation device. The method of claim 14,
The object path predicting unit generates the prediction path data by setting an uncertainty coefficient related to the uncertainty larger than a predetermined reference value when the type of the object is a vehicle.
Object prediction path generation device. The method of claim 14,
The object path predictor,
Calculating an average and a variance of the plurality of prediction path data generated at each predetermined time interval,
Generating and storing label information indicating whether a danger is at an absolute coordinate of a position corresponding to the prediction path data based on the mean and the variance
Object prediction path generation device. The method of claim 17,
The apparatus may further include an object risk generation unit configured to calculate a collision probability of the unmanned vehicle and the object with respect to the continuous space between the absolute coordinates based on the label information.
Object prediction path generation device. The method of claim 18,
The object risk generator may include a plurality of sparse kernels and weights assigned to each sparse kernel with respect to absolute coordinates.

Calculating the probability of collision using a logistic regression classifier
Object prediction path generation device. As an unmanned vehicle,
The unmanned vehicle includes an apparatus for generating prediction path data of an object.
The apparatus for generating the prediction path data of the object,
An object recognition unit obtaining motion information of at least one or more objects around the unmanned vehicle and the type of the object based on environment recognition information on the driving route of the unmanned vehicle;
And an object path predictor configured to generate prediction path data of the object based on the motion information of the object and the uncertainty according to the type of the object.
Driverless vehicles.

Images