JP2008158969A - Method, device and program for determining risk around moving body - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a method, a device and a program for determining a risk around a moving body, determining the risk around the moving body in consideration of a predicted situation change around the moving body, and attaining safe travel of the moving body. <P>SOLUTION: A collision probability of the moving body with an obstacle is calculated by predicting stochastically possible travelling routes of the obstacle, by generating a travelling route of the moving body and by finding the travelling route interfering with the travelling route of the moving body, out of the possible travelling routes of the obstacle. The risk around the moving body is determined using the latest collision probability of the moving body with the obstacle in its periphery, and a temporal change between the latest collision probability and the collision probability a prescribe time before. <P>COPYRIGHT: (C)2008,JPO&INPIT

Description

  The present invention relates to a moving object surrounding risk determination method, an apparatus, and a program for determining a degree of danger around a moving object from information on obstacles around the moving object.

  2. Description of the Related Art Conventionally, as a technique for improving the safety of a vehicle such as a four-wheeled vehicle, a technique for preventing a collision by determining a risk that the vehicle collides with an obstacle is known. Among these, in the following Patent Document 1, a collision including a yaw rate sensor for detecting the yaw rate of the own vehicle, a speed detecting device for detecting the speed of the own vehicle, and a radar device for detecting the position and speed of surrounding obstacles. Techniques related to the prevention device are disclosed.

  In the technique described in Patent Document 1 below, a predicted traveling locus of the host vehicle is obtained by a yaw rate sensor or a speed sensor, and predetermined areas on both sides of the predicted traveling locus are obtained as predicted traveling areas of the own vehicle, and detected by a radar device. The predicted travel locus and predicted travel area of the obstacle are obtained from the position and speed of the obstacle. Subsequently, the collision point or proximity point between the vehicle and the obstacle is calculated from each predicted travel area, and the collision risk between the vehicle and the obstacle is determined. The determination result is applied to the speed control of the own vehicle.

Japanese Patent No. 2799375

  However, in the conventional collision prevention technology including the above-described technology, although there is a small risk of collision with the own vehicle at the time of prediction, there is a case where an obstacle whose risk becomes large in the future cannot be detected. Hereinafter, this point will be described with reference to FIG.

FIG. 27 is a diagram illustrating a situation where the host vehicle C ₀ and the other vehicles C ₁ and C ₂ are traveling on the road R in the same direction. In the figure, the host vehicle C ₀ and the other vehicles C ₁ , C ₂ are traveling at speeds v _0, v ₁ , v ₂ , respectively (assuming v ₁ > v ₀ , v ₂ > v ₀ ). Further, the distance between the host vehicle C ₀ and the other vehicle C ₁ is L ₁ , and the distance between the host vehicle C ₀ and the other vehicle C ₂ is L ₂ , which satisfies L ₁ <L ₂ .

In the conventional collision prevention technique, in the situation shown in FIG. 27, for example, the collision probability between the own vehicle C ₀ and the other vehicles C ₁ and C ₂ is calculated, and an obstacle having a collision probability larger than a predetermined threshold is calculated. It was detected as a dangerous obstacle for the vehicle. Therefore, in the situation shown in FIG. 27, the collision probability with the other vehicle C ₁ that is relatively close to the own vehicle C ₀ is larger than the collision probability with the other vehicle C _2, and only the other vehicle C ₁ is present. May be detected as a dangerous obstacle.

Here, the subject vehicle C ₀ and the other vehicle C _1, C ₂ is, considering the case of maintaining the running state shown in Figure 27, another vehicle C ₁ going away from the vehicle C ₀ and the subject vehicle C ₀ while collision probability decreases with time, probability of collision with another car C ₂ between the vehicle C ₀ approaching the subject vehicle C ₀ is increased with time. In this case, in the conventional anti-collision techniques described above, since the determined collision risk at that time only by collision probability of the prediction point, as the other vehicle C _2, although the prediction time is less collision probability In some cases, an obstacle whose collision probability increases in the future cannot be detected as a dangerous obstacle.

  As described above, with the conventional collision prevention technology, it is impossible to determine the degree of danger around the vehicle based on the expected change in the surroundings of the vehicle. For this reason, obstacles that may become dangerous for the vehicle in the future cannot be accurately detected, and there is a risk of seriously affecting the safety of the moving object when traveling.

  The present invention has been made in view of the above, and it is possible to determine the degree of danger around the moving body based on the anticipated change in the situation around the moving body, and to ensure the safety of the moving body. An object of the present invention is to provide a moving object surroundings risk determination method, apparatus, and program capable of realizing smooth traveling.

  In order to solve the above-described problems and achieve the object, one aspect of the present invention provides a computer including storage means for storing at least obstacle information including the position and internal state of an obstacle existing around a moving body. Is a moving object surroundings risk determination method for determining a risk around the moving object, and a moving object route generation step for generating a route of the moving object based on a position and an internal state of the moving object; The obstacle course predicting step for probabilistically predicting the course that the obstacle can take based on the obstacle information read from the storage means, and the obstacle predicted by the obstacle course predicting step can be taken. Collision probability for calculating a collision probability between the moving body and the obstacle by obtaining a path that interferes with the path of the moving body generated in the moving body path generation step. The risk level around the moving object is determined by using the calculation step, the latest collision probability calculated in the collision probability calculation step, and the time variation between the latest collision probability and the collision probability before a predetermined time. And a risk determination step.

  In the above invention, the risk determination step detects a dangerous obstacle that should be regarded as dangerous for the moving body from obstacles existing around the moving body according to a predetermined criterion. Also good.

  Further, in the above invention, when the time change amount is positive, the determination criterion is that the minimum value of the latest collision probability when the obstacle is detected as a dangerous obstacle is larger as the time change amount is larger. It may be small.

  In the above invention, the risk determination step may determine the risk of the moving body in an environment composed of all obstacles around the moving body according to a predetermined criterion.

  Further, in the above invention, when the time change amount of the environmental risk level with respect to the moving object of the environment determined by using the collision probability between the moving object and the obstacle is positive, It is good also as the minimum value of the said environmental risk degree at the time of determining with it being dangerous, so that quantity is large.

  In the above invention, the obstacle course predicting step may be performed on a space-time space configured by time and space to change a position that the obstacle can take over time based on the position and internal state of the obstacle. A trajectory generation step generated as a trajectory at, and a prediction calculation step of performing a probabilistic prediction calculation of the course of the obstacle by using the trajectory generated in the trajectory generation step.

  Moreover, in the said invention, you may further have an output step which outputs the information according to the result determined by the said risk determination step.

  According to another aspect of the present invention, there is provided a mobile body surrounding risk determination apparatus, a mobile body path generation unit that generates a path of the mobile body based on a position and an internal state of the mobile body, Obstacle path prediction means for probabilistically predicting a path that the obstacle can take based on the obstacle information read from the storage means for storing at least obstacle information including the position and internal state of the obstacle to be Of the paths that can be taken by the obstacle predicted by the obstacle path prediction means, the path that interferes with the path of the mobile body generated by the mobile body path generation means is obtained, whereby the mobile body and the obstacle A collision probability calculation means for calculating the collision probability of the vehicle, a latest collision probability calculated by the collision probability calculation means, and a time change amount between the latest collision probability and the collision probability before a predetermined time. Ri, and further comprising a, a risk determination means for determining the risk of the surrounding of the movable body.

  In the above invention, the risk determination means detects a dangerous obstacle that should be regarded as dangerous for the moving body from obstacles existing around the moving body according to a predetermined criterion. Also good.

  Further, in the above invention, when the time change amount is positive, the determination criterion is that the minimum value of the latest collision probability when the obstacle is detected as a dangerous obstacle is larger as the time change amount is larger. It may be small.

  In the above invention, the risk determination means may determine the risk for the moving body in an environment composed of all obstacles around the moving body according to a predetermined criterion.

  Further, in the above invention, when the time change amount of the environmental risk level with respect to the moving object of the environment determined by using the collision probability between the moving object and the obstacle is positive, It is good also as the minimum value of the said environmental risk degree at the time of determining with it being dangerous, so that quantity is large.

  Further, in the above invention, the obstacle course predicting means may be configured to change a position that the obstacle can take with time based on the position and the internal state of the obstacle on a space-time space composed of time and space. There may be provided a trajectory generating means for generating a trajectory at the point, and a predictive calculating means for performing a probabilistic predictive calculation of the course of the obstacle by using the trajectory generated by the trajectory generating means.

  Moreover, in the said invention, you may further provide the output means which outputs the information according to the result determined by the said risk degree determination means.

  A moving object surrounding risk determination program according to another aspect of the present invention causes the computer to execute the moving object surrounding risk determination method according to any one of the above inventions.

  According to the moving object surrounding risk determination method, apparatus, and program according to the present invention, when determining the risk around the moving object, the latest collision probability between the moving object and the surrounding obstacle, and the latest By using the amount of time change between the collision probability and the collision probability before a predetermined time, it is possible to determine the degree of danger around the moving object based on the expected change in the surrounding environment. The safe traveling of the moving body can be realized.

  The best mode for carrying out the present invention (hereinafter referred to as “embodiment”) will be described below with reference to the accompanying drawings.

(Embodiment 1)
FIG. 1 is a block diagram showing a functional configuration of a mobile object surrounding risk determination device according to Embodiment 1 of the present invention. A moving object surrounding risk determination device 1 shown in the figure is mounted on a vehicle such as a four-wheeled vehicle (hereinafter referred to as “own vehicle”) as a moving object, and information on obstacles around the own vehicle is obtained. It is a device used to determine the degree of danger around the vehicle.

  The moving object surrounding risk determination device 1 detects the position and internal state of an object existing in a predetermined range and detects the position and internal state of the own vehicle, and the vehicle portion detected by the sensor unit 2. Based on the information, the own vehicle route generation unit 3 (moving body route generation means) that generates a route from the current position of the own vehicle, and the surroundings detected by the sensor unit 2 within a predetermined range An obstacle extraction unit 4 that extracts included obstacles, and an obstacle course prediction unit 5 that performs probabilistic prediction of a course that the obstacles extracted by the obstacle extraction unit 4 can take.

  In addition, the moving object surrounding risk determination device 1 uses the own vehicle route generated by the own vehicle route generation unit 3 and the obstacle route predicted by the obstacle route prediction unit 5 to collide with the own vehicle and the obstacle. A collision probability calculation unit 6 for calculating the probability, a dangerous obstacle detection unit 7 (risk degree determination means) for detecting an obstacle with a high risk of collision with the host vehicle, using the calculation result of the collision probability calculation unit 6; The storage unit 8 stores various information including the detection result of the sensor unit 2 and the calculation result of the collision probability calculation unit 6, and writes the calculation result of the collision probability calculation unit 6 into the storage unit 8 and information on necessary collision probability. Is output from the storage unit 8 and output to the dangerous obstacle detection unit 7, and an output unit 10 that outputs various types of information including detection results of the dangerous obstacle detection unit 7 is provided.

  The sensor unit 2 is configured by an appropriate combination capable of detecting the vehicle and surrounding objects among a speed sensor, an acceleration sensor, a steering angle sensor, an angular velocity sensor, an image sensor, a millimeter wave radar, a laser radar, and the like. Note that the internal state of the object detected by the sensor unit 2 is a useful state that can be used for prediction of the object, and preferably the speed (with speed and direction) and angular speed (size and size) of the object. Physical quantity).

  The storage unit 8 includes a collision probability storage unit 81 that stores the collision probability between the own vehicle and the obstacle calculated by the collision probability calculation unit 6 together with obstacle information including the position and internal state of the obstacle. The storage unit 8 also stores information on the position and internal state of the host vehicle detected by the sensor unit 2, information on the host vehicle route generated by the host vehicle route generation unit 3, and the like. Such a storage unit 8 includes a ROM in which a moving object surrounding risk determination program according to the first embodiment, a program for starting a predetermined OS, and the like are stored in advance, and a RAM in which calculation parameters and data of each process are stored. Etc. The storage unit 8 can also be realized by providing an interface on which a computer-readable recording medium can be mounted on the moving object surrounding risk determination device 1 and mounting a recording medium corresponding to this interface.

  The output unit 10 generates an image based on the processing result performed by the dangerous obstacle detection unit 7, and displays the generated image using a display such as liquid crystal, plasma, organic EL, A warning sound generation unit 102 that generates a predetermined warning sound according to the processing result of the dangerous obstacle detection unit 7.

  The moving object surroundings risk determination device 1 having the above-described functional configuration is realized using an electronic device (computer) including a CPU having calculation and control functions. The CPU included in the moving object surrounding risk determination device 1 reads out various programs including the moving object surrounding risk determination program and other various information from the storage unit 8, thereby moving the moving object surrounding risk degree according to the first embodiment. Arithmetic processing relating to a determination method (described later) is executed.

  The mobile object surrounding risk determination program according to the first embodiment is recorded on a computer-readable recording medium such as a hard disk, a flexible disk, a CD-ROM, a DVD-ROM, a flash memory, or an MO disk, and widely distributed. It is also possible to make it.

  Next, the moving body surrounding risk determination method according to the first embodiment will be described. FIG. 2 is a flowchart showing an outline of processing of the moving object surrounding risk determination method according to the first embodiment. In the following description, it is assumed that all objects to be processed move on a two-dimensional plane, but the moving object surrounding risk determination method according to the first embodiment moves in a three-dimensional space. It can also be applied to objects. Further, the present invention can be applied to a case where one object has a plurality of degrees of freedom (for example, an object such as a robot arm having 6 degrees of freedom).

  In FIG. 2, the sensor part 2 detects the position and internal state of the own vehicle, and stores the detected information in the storage part 8 (step S1). Hereinafter, it is assumed that the position of the object is a value at the center of the object. The internal state of the object includes at least the speed of the object, and may include acceleration, angular velocity, and the like.

  Thereafter, the host vehicle course generation unit 3 generates a course of the host vehicle based on the detection result of the sensor unit 2 (step S2). Specifically, the own vehicle course generation unit 3 generates a trajectory when the own vehicle travels as it is. In addition, as long as the sensor unit 2 can detect a road surface environment such as a white line, a plurality of trajectories corresponding to the number of lanes that can travel may be generated as the own vehicle course.

  The moving object surrounding risk determination device 1 probabilistically predicts a path that an obstacle present around the host vehicle can take in parallel with the processes of steps S1 and S2 (steps S3 to S5). Specifically, first, the sensor unit 2 detects the position and internal state of an object within a predetermined range from the own vehicle with respect to the own vehicle (step S3).

  Next, the obstacle extraction unit 4 extracts obstacles within a predetermined range based on the result detected by the sensor unit 2, and stores obstacle information including the position and internal state of the extracted obstacles in the storage unit 8. Is written and stored (step S4). In step S4, an object that can be regarded as an obstacle under a predetermined condition is extracted from the objects detected in step S3, and other objects are excluded.

  Thereafter, the obstacle course prediction unit 5 reads the obstacle information at the current time from the storage unit 8, and probabilistically predicts a plurality of paths that the obstacle can take (step S5). In this step S5, various conventionally known methods can be applied. For example, the route prediction may be performed by assigning a predetermined probability distribution to a plurality of routes that the obstacle can take according to the current situation.

FIG. 3 is a diagram illustrating an example of a probability distribution given to the path of an obstacle in step S5. Specifically, a case where a probability distribution curve ρ ₁ that maximizes the straight direction is given to the obstacle is illustrated. In this sense, the x coordinate in FIG. 3 is a coordinate in the width direction of the road R, and its origin represents the current position of the obstacle. Note that the probability distribution to be given to the obstacle is preferably unimodal as typified by, for example, a normal distribution, but is not limited to the distribution function.

  Thereafter, the collision probability calculation unit 6 obtains a path that interferes with the path of the host vehicle generated by the host vehicle path generation unit 3 from the path of the obstacle predicted by the obstacle path prediction unit 5. The collision probability between the obstacle and the obstacle is calculated (step S6). When the distance between the own vehicle path and the obstacle path is smaller than a predetermined distance (referred to as an interference distance), the collision probability calculation unit 6 causes the two paths to interfere with each other, Consider it a collision. The interference distance is determined according to conditions such as the size of the vehicle and the obstacle (in the case of a vehicle, the width and length of the vehicle). The calculation result of the collision probability calculation unit 6 is written into the collision probability storage unit 81 by the collision probability input / output unit 9.

FIG. 4 is a diagram schematically illustrating a collision probability storage mode in the collision probability storage unit 81. Collision probability storage unit 81, the time t the vehicle in _n and the obstacle _{O j (j = 1,2,3, ···} ) collision probability CP _j a (n), the time t _n (n is an integer) In the table Tb (n). In the table Tb (n), the obstacle information at time t _n of the obstacle O _j X _j (n) at time t _{n +} obstacle in _one information X _j (n + 1) and is, collision probability CP _j (n) It is recorded with. Therefore, as shown in FIG. 5, the obstacle information X _j (n) at time t _n is recorded in duplicate in a maximum of two tables Tb (n) and Tb (n−1). By configuring the table Tb (n) having such characteristics for each time, the collision probability CP _j (n) between the own vehicle and the obstacle O _j is within a range recorded in the collision probability storage unit 81. You can follow up to any past.

FIG. 6 to FIG. 8 are diagrams showing a state in which information is written to two tables Tb (n) and Tb (n−1) in chronological order as time passes. First, FIG. 6 shows a state in which obstacle information and collision probability up to time t _n−1 are written. Although not shown in the figure, writing of all the information in the tables Tb (n−2), Tb (n−3),... Has been completed at this time (time t _n−1 ).

Thereafter, when the collision probability calculating unit 6 calculates the time t _n of the collision probability CP _j (n), the collision probability output section 9, the time t _n-1 of the table Tb (n-1) to the obstacle information X _j Write (n). At this time, the collision probability input / output unit 9 takes correspondence with the obstacle information X _j (n−1) at time t _n−1 one step before. Specifically, the collision probability input / output unit 9 regards the obstacle information X _j (n) as a vector and takes an inner product with the obstacle information X _{j ′} (n−1) at time t _n−1. The combination that minimizes the inner product is associated with the same obstacle.

The association of obstacles at different times may be performed by other methods. For example, the obstacle information X _j (n) is regarded as a point on the multidimensional space-time, and the distance between the points on the multidimensional space-time is calculated based on a predetermined definition (for example, Euclidean distance). Obstacles may be dealt with according to the minimum combination. If the sensor unit 2 includes an image sensor, the same obstacle can be tracked by image recognition processing such as pattern matching.

Thereafter, the collision probability input / output unit 9 reads the collision probability CP _j (n−1) at time t _n−1 one step before the obstacle that has been dealt with by the above-described association, and detects the dangerous obstacle. Output to unit 7. FIG. 7 shows a state where the table Tb (n−1) has been associated. For obstacle information that cannot be handled between time t _n−1 and time t _n , null is output to the dangerous obstacle detection unit 7. Thus, in general, not all obstacles are always associated with each other at different times.

Subsequently, the collision probability input / output unit 9 writes the obstacle information X _j (n) and the collision probability CP _j (n) of the obstacle O _j at time t _{n in} the table Tb (n). Thereby, the tables Tb (n) and Tb (n−1) stored in the collision probability storage unit 81 are in the state shown in FIG.

  The collision probability input / output unit 9 sequentially writes the obstacle collision probability and the obstacle information in the collision probability storage unit 81 by repeatedly performing the same processing as described with reference to FIGS. .

Next, using the collision probability calculated by the collision probability calculation unit 6, the dangerous obstacle detection unit 7 detects a dangerous obstacle with a high risk of collision with the own vehicle (step S7). In the following description, the latest collision probability is CP _j (0), and the time corresponding to this collision probability (current time) is t ₀ . In step S _< b> 7, the dangerous obstacle detection unit 7 detects a dangerous object using two amounts of the collision probability CP _j (0) at the current time t ₀ and the time variation ΔCP _j of the collision probability. Here, the time variation Delta] CP _j, the collision probability CP _j at time t ₀ (0), as a change amount of the collision probability CP _j (-1) of Earlier time t _-1, equation (1) Defined by
ΔCP _j = CP _j (0) −CP _j (−1) (1)

When detecting the dangerous obstacle, the dangerous obstacle detection unit 7 uses the collision probability CP _j (0) at time t ₀ and the time variation ΔCP _{j of the} collision probability defined by Expression (1). Detect dangerous obstacles. FIG. 9 is a diagram schematically illustrating a determination criterion when the dangerous obstacle detection unit 7 detects a dangerous obstacle. The dangerous obstacle detection unit 7 has a set (CP _j (0), ΔCP _j ) of the collision probability CP _j (0) at time t ₀ and the time variation ΔCP _j of the collision probability belongs to the region Sα in FIG. If it is, the obstacle O _j is determined as a dangerous obstacle.

The curve α that defines the boundary of the region Sα is a region in which ΔCP _j > 0 and the collision probability CP _j (0) decreases as the time variation ΔCP _j increases. By using the region Sα delimited by such a curve α as a determination criterion, the larger the time change amount ΔCP _j (> 0) of the collision probability, the larger the latest collision probability when determining a dangerous obstacle. The minimum value of CP _j (0) is small. Even if the collision probability at the current time is small, the amount of time change of the collision probability is large, and obstacles that are likely to be dangerous for the vehicle in the future are displayed. It becomes possible to detect it as a dangerous obstacle at an early stage.

It should be noted that the criterion for determining a dangerous obstacle in the dangerous obstacle detection unit 7 is not limited to that described above. FIG. 10 is a diagram schematically illustrating another determination criterion when the dangerous obstacle detection unit 7 detects a dangerous obstacle. The dangerous obstacle detection unit 7 has a set (CP _j (0), ΔCP _j ) of the collision probability CP _j (0) at time t ₀ and the time variation ΔCP _j of the collision probability belongs to the region Sβ in FIG. If it is, the obstacle O _j is determined as a dangerous obstacle.

In FIG. 10, a curve β that defines the boundary of the region Sβ is a region where ΔCP _j > 0, and the collision probability CP _j (0) approaches a constant value while decreasing as the time change amount ΔCP _j increases. It is a curved line. Further, the curve β is a curve in which CP _j (0) approaches a constant value while increasing as ΔCP _j decreases in a region where ΔCP _j <0. Even when the region Sβ delimited by the curve β showing such a behavior is adopted as the determination criterion, even if the collision probability at the current time is small, the collision can be performed as in the case where the region Sα is employed as the determination criterion. It is possible to detect an obstacle that has a large probability of time change and is likely to be dangerous for the vehicle in the future as a dangerous obstacle.

  After step S7 described above, the output unit 10 outputs information according to the detection result of the dangerous obstacle detection unit 7 (step S8). Specifically, the display unit 101 displays a dangerous obstacle, and the warning sound generation unit 102 generates a warning sound to notify that there is a dangerous obstacle around.

Incidentally, as the display unit 101, a projector is installed at the rear upper portion of the driver's seat, it may be performed superimposed display of the windshield of the vehicle C ₀ by this projector. Thereby, the driver of the own vehicle C ₀ can recognize the dangerous obstacle while driving. Accordingly, the driver of the subject vehicle C ₀ is, by reflecting the recognition result to the immediately driving operation, it is possible to avoid the risk of becoming approaching the subject vehicle C ₀ to accurately. Further, for example, a display screen of a car navigation system mounted on the host vehicle C ₀ may be provided with a function as the display unit 101 to display dangerous obstacles, or dangerous obstacle information is displayed on a meter panel. You may make it do.

  Furthermore, the output unit 10 may be provided with a function of vibrating the handle or seat of the own vehicle, and when a dangerous obstacle is detected, the handle or seat may be vibrated to notify the danger.

  The processes in steps S1 to S8 described above are repeatedly performed at predetermined time intervals, and information in accordance with the latest road environment is always output. For this reason, according to the moving body surrounding risk determination method according to the first embodiment, it is possible to assist the driver of the own vehicle to appropriately operate in response to the road environment that changes every moment.

Here, as the moving object surrounding risk determination method according to the first embodiment, an advantage of taking into account the temporal change amount of the collision probability will be described. FIG. 11 is a diagram showing a time change example of the collision probability CP _j (n) of the other vehicles C ₁ and C ₂ as obstacles with the own vehicle C ₀ under the road environment shown in FIG. First, the other vehicle C ₁ is traveling ahead by a distance L ₁ (<L _th ) from the _host vehicle C ₀ . Therefore, as shown by the curve 201, although the collision probability at the present time is high, when the vehicle continues to move in the same situation, the vehicle travels at a higher speed than the host vehicle C ₀ (v ₁ > v ₀ ). Gradually decreases. On the other hand, the succeeding other vehicle C ₂ is traveling behind the own vehicle C ₀ by a distance L ₂ (> L _th ). Therefore, as shown by the curve 202, although the collision probability at the present time is low, the vehicle is traveling at a higher speed than the host vehicle C ₀ (v ₂ > v ₀ ), so the collision probability gradually increases as time elapses. I will do it.

In the conventional collision determination technique, for example, when the collision probability CP _j (0) at the current time t ₀ exceeds the threshold value CP _th , it is determined as dangerous. Therefore, in the case shown in FIG. 11, the other vehicle C _{1 that} is close to the own vehicle C ₀ is recognized as a dangerous obstacle, but the other vehicle C ₂ that is far from the own vehicle C ₀ is a dangerous obstacle. It was never recognized as a thing.

On the other hand, according to the moving object surrounding risk determination method according to the first embodiment, the danger is obtained by using the relationship between the latest collision probability CP _j (0) and the time variation ΔCP _j of the collision probability. Since an obstacle is detected, a determination result that takes into account future changes in the degree of risk can be obtained. For example, when the dangerous obstacle detection unit 7 detects a dangerous obstacle by using the region Sα of FIG. 9 as a criterion, as shown in FIG. 12, the other vehicle C ₂ is detected as a dangerous obstacle ( Point P ₂ ) is not detected as a dangerous obstacle because the other vehicle C ₁ is ΔCP _j <0 (point P ₁ ). On the other hand, when the dangerous obstacle detection unit 7 detects a dangerous obstacle by using the region Sβ of FIG. 10 as a criterion, the other vehicles C ₁ , C ₂ , as indicated by points P ₁ , P ₂ in FIG. Both ₂ are detected as dangerous obstacles.

  As described above, according to the moving object surrounding risk determination method according to the first embodiment, even if an obstacle has a low collision probability at the present time, an obstacle whose collision probability increases remarkably with the passage of time. It becomes possible to detect an object as a dangerous obstacle at an early stage.

In the above description, as the time variation Delta] CP _j of collision probability has been described with reference to the amount of change in collision probability between time t _-1 ~t ₀ adjacent to determine the time variation Delta] CP _j The time interval used at the time can be arbitrarily set. Moreover, you may define the amount of time change using the collision probability in three or more time.

Further, in the above description, the preceding vehicle traveling in the same direction as the own vehicle C ₀ has been described as an obstacle. For example, an oncoming vehicle traveling in the opposite direction to the own vehicle is referred to as an obstacle. It can also be. In addition, a stationary object can be used as an obstacle.

  According to the first embodiment of the present invention described above, the latest collision probability between the moving body (own vehicle) and the surrounding obstacle, and the amount of time change between the latest collision probability and the collision probability before the predetermined time. According to the criteria set forth using, the situation surrounding the moving object is predicted by detecting dangerous obstacles that should be considered dangerous to the moving object from the obstacles that exist around the moving object. Therefore, it is possible to determine the danger level around the moving object based on the change of the above, and to detect an obstacle that may increase the risk level for the moving object in the future. Therefore, for example, even when the vehicle is controlled based on the presence or absence of a dangerous obstacle, it becomes possible to perform control that avoids danger at an early stage, thereby realizing safe traveling of the moving object. Can do.

  Further, according to the first embodiment, as a criterion for determining a dangerous obstacle, when a predetermined time change amount of the collision probability is positive, the obstacle is more dangerous as the time change amount is larger. Because the criterion that minimizes the minimum value of the latest collision probability when detecting as an object is used, an obstacle whose collision probability tends to increase in the future as a dangerous obstacle even if the collision probability is small It becomes possible to detect accurately at an early stage.

  Furthermore, according to the first embodiment, when a dangerous obstacle is detected, by outputting information corresponding to the dangerous obstacle, the driver of the mobile object receiving the information is driving It is possible to drive while avoiding dangers that may occur in the near future. In addition, for dangerous obstacles that increase in risk, various information can be output (early) before the level of increase in risk. Respond to changing circumstances.

(Embodiment 2)
The second embodiment of the present invention generates a change in position that an obstacle extracted in the same manner as in the first embodiment can take as time passes, as a trajectory on time and space composed of time and space, The route is predicted using the generated trajectory. In the second embodiment, it is assumed that the position of the object is a value at the center of the object, and the internal state of the object is specified by the speed (speed v, direction θ).

  FIG. 14 is a block diagram illustrating a functional configuration of the moving object surrounding risk level determination device according to the second embodiment. In the moving object surrounding risk determination device 11 shown in the figure, the configuration other than the obstacle course prediction unit 12 is the same as that of the moving object surrounding risk determination device 1 according to the first embodiment. For this reason, the same code | symbol is attached | subjected to the site | part which has the same function as the mobile body surrounding risk determination apparatus 1, respectively.

  The obstacle course prediction unit 12 generates a change in position that each obstacle extracted by the obstacle extraction unit 4 can take with time as a locus on a time space composed of time and space. And a prediction calculation unit 14 that performs a probabilistic prediction calculation of the course of the obstacle by using the path of the obstacle output from the path generation unit 13.

  The trajectory generation unit 13 predicts and generates a trajectory that an obstacle can take until a predetermined time elapses, and an operation for selecting an operation for virtually operating the obstacle on a simulation from a plurality of operations. The selection unit 131, the object operation unit 132 that performs the operation selected by the operation selection unit 131 for a predetermined time, and whether the position and internal state of the obstacle after the operation by the object operation unit 132 satisfy a predetermined condition And a determination unit 133 for determining

  Next, a moving object surrounding risk determination method according to Embodiment 2 of the present invention will be described. The moving object surrounding risk determination method according to the second embodiment is the same as the moving object surrounding risk determination method according to the first embodiment, except for the obstacle course prediction process and the collision probability calculation process. (See the flowchart in FIG. 2). Therefore, in the following description, the obstacle course prediction process (corresponding to step S5 in FIG. 2) and the collision probability calculation process (corresponding to step S6 in FIG. 2) will be described in detail.

  FIG. 15 is a flowchart showing an outline of obstacle course prediction processing. In the following description, the course prediction process for one obstacle is described, but in practice, the course prediction process is performed for all obstacles. First, the trajectory generation unit 13 generates a plurality of trajectories of obstacles (step S51).

FIG. 16 is a flowchart showing details of the locus generation processing in the locus generation unit 13. In FIG. 16, assuming that the total number of objects (including the own vehicle) detected by the sensor unit 2 is K, an operation for generating a locus for one obstacle O _k (1 ≦ k ≦ K, k is a natural number) is M. _It is assumed that _k is performed (M _k is a natural number). Further, the time for generating the locus (trajectory generation time) is T (> 0). By appropriately determining the trajectory generation time T (and an operation time Δt described later), a series of course prediction processes can be performed in a practical calculation time.

The trajectory generation unit 13 first initializes the value of the counter _k for identifying the obstacle Ok to 1 and sets the value of the counter m indicating the number of trajectory generations for the same obstacle to 1 (step S501). .

Next, the trajectory generation unit 13 reads the result detected by the sensor unit 2 from the storage unit 8, and sets the read detection result as an initial state (step S502). Specifically, the time t is set to 0, and the initial position (x _k (0), y _k (0)) and the initial internal state (v _k (0), θ _k (0)) are respectively transmitted from the sensor unit 2. Input information (x _k0 , y _k0 ) and (v _k0 , θ _k0 ).

Subsequently, the operation selection unit 131 selects one operation u _k (t) to be performed during the subsequent time Δt from a plurality of selectable operations according to the operation selection probability given in advance to each operation. (Step S503). Operation u operation selection probability for selecting the _kc p (u _kc) is defined by associating the elements with a predetermined random number, for example a set of selectable operations as _{u k (t) {u kc} }. In this sense, a different operation selection probability p (u _kc ) may be given for each operation u _kc , or an equal probability may be given to all elements of the operation set {u _kc }. In the latter case, p (u _kc ) = 1 / (total number of selectable operations). The operation selection probability p (u _kc ) can be defined as a function that depends on the position and internal state of the vehicle and the surrounding road environment.

In general, the operation u _kc is composed of a plurality of elements, and the contents of selectable operations differ depending on the type of the obstacle O _k . For example, when the obstacle _Ok is a four-wheeled vehicle, the acceleration and angular velocity of the four-wheeled vehicle are determined by the degree of steering and the degree of depression of the accelerator. In view of this, the operation u _kc to be performed on the obstacle O _k is a four-wheeled vehicle is determined by elements including acceleration and angular velocity. On the other hand, when the obstacle _Ok is a person, the operation u _kc can be designated by the speed.

A more specific setting example of the operation u _kc will be given. When the obstacle O _k is an automobile takes the acceleration -10~ + 30 (km / h / sec), a steering angle in a range of -7~ + 7 (deg / sec) ( specifying the direction in either sign) , if the obstacle O _k is a human, the speed 0~36 (km / h), take an orientation in a range of 0~360 (deg). In addition, all the quantities described here are continuous quantities. In such a case, an appropriate set of discretizations may be used to make the number of elements of each operation finite, and a set of operations {u _kc } may be configured.

Thereafter, the object operation unit 132 operates the operation u _kc selected in step S503 for a time Δt (step S504). The time Δt is preferably smaller in terms of accuracy, but may be a value of about 0.1 to 0.5 (sec) in practice. In the following description, the trajectory generation time T is an integral multiple of Delta] t, the value of T is may be variable depending on the speed of the obstacle O _k, may not be an integer multiple of Delta] t .

Subsequently, the determination unit 133, the internal state of the obstacle O _k after operating the operation u _kc in step S504 determines whether or not satisfies a predetermined control condition (step S505). Determines the control condition in this step S505 is defined according to the type of the obstacle O _k, for example, when the obstacle O _k is a four-wheeled vehicle, and the speed range after the operation of step S504, step S504 The maximum acceleration G after the operation is determined by the vehicle G or the like.

Determined in step S505, if the internal state of the obstacle O _k satisfies a predetermined control condition (in step S505 Yes), the determination unit 133, the obstacle O _k after operating the operation u _kc It is determined whether or not the position is within the movable area (step S506). The movable area determined in step S506 refers to an area such as a road (including a roadway and a sidewalk). Hereinafter, the case where the object is located in the movable region is expressed as “the moving condition is satisfied”.

Result of the determination in step S506, when the obstacle O _k are located movable region (Yes at step S506), advance the trajectory generating unit 13 only time Δt (t ← t + Δt) , after the operation of step S504 Is set to (x _k (t), y _k (t)), and the internal state is set to (v _k (t), θ _k (t)) (step S507).

  Note that if there is a condition that is not satisfied even in step S505 or S506 (No in step S505 or No in step S506), the process returns to step S502.

The processes in steps S502 to S507 described above are repeated until the trajectory generation time T is reached. That is, when the time t newly defined in step S507 has not reached T (No in step S508), the process returns to step S503 and is repeated. On the other hand, if the newly defined time t in step S507 reaches T (Yes at step S508), and outputs the trajectory with respect to the obstacle O _k, stored in the storage unit 8 (step S509).

Figure 17 is a time t = 0, Δt, 2Δt, ···, is a locus of the obstacle O _k generated by repeating a series of processes ranging from step S503 to step S507 in FIG schematically shown by T . The trajectory TR _k (m) (1 ≦ m ≦ M _k , where m is a natural number) shown in the figure is a three-dimensional space-time (x, y, t) of two-dimensional space (x, y) and one-dimensional time (t). ) If this trajectory TR _k (m) is projected onto the xy plane, a predicted course of the obstacle O _k in the two-dimensional space (x, y) can be obtained.

After step S509, if the value of the counter m has not reached M _k (No in step S510), the trajectory generating unit 13 increases the value of the counter m by 1 (step S511), returns to step S502, and returns to step S502 described above. Step S508 is repeated until the trajectory generation time T is reached.

When the counter m reaches M _k in step S510 (Yes in step S510), the generation of all trajectories for the obstacle O _k is completed. Figure 18 is one obstacle O M generated for _{k k} pieces of locus _{TR k (1), TR k} (2), ···, trajectory group consisting of _{TR k (M k) {TR} k } Is a diagram schematically showing three-dimensional space-time. The starting point, that is, the initial position (x _k0 , y _k0 , 0) of each trajectory forming the elements of the trajectory set {TR _k } is the same (see step S502). Note that FIG. 18 is a schematic diagram to the last, and the value of M _k can take, for example, a value of about several hundred to several tens of thousands.

When the counter m reaches M _k in step S510, if the obstacle identification counter k has not reached the total number K of obstacles (No in step S512), the value of the counter k is incremented by 1 and a trajectory is generated. The value of the counter m for the number of times is initialized to 1 (step S513), and the process returns to step S502 to repeat the process. On the other hand, when the counter k reaches K (Yes in step S512), the trajectory generation for all obstacles is completed, so the trajectory generation process in step S2 is terminated and the process proceeds to the subsequent step S3. .

In this manner, with respect to all the obstacles O _k of the sensor unit 2 detects, by performing trajectory generation processing for a predetermined number of times, within the predetermined range when the three-dimensional space (x, y, t) A spatiotemporal environment consisting of a set of trajectories that a plurality of obstacles can take is formed. FIG. 19 is an explanatory diagram schematically illustrating a configuration example of a spatiotemporal environment. The spatio-temporal environment Env (TR ₁ , TR ₂ ) shown in the figure corresponds to the road environment shown in FIG. 27, and the trajectory set {TR ₁ } of the obstacle O ₁ (indicated by a broken line in FIG. 19) and It consists of a trajectory set {TR ₂ } of obstacle O ₂ (indicated by a solid line in FIG. 19). In the second embodiment, since the trajectory generation is performed independently for each obstacle without considering the correlation between the obstacles, the trajectories of different obstacles may intersect in time and space.

In FIG. 19, the density per unit volume of the trajectory set {TR _k } (k = 1, 2) in each region of the three-dimensional space-time (x, y, t) is the obstacle O _k in each region of the space-time. Density of existence probability (hereinafter referred to as “spatiotemporal probability density”). Therefore, by using the space environment Env (TR _1, TR ₂₎ when configured by the trajectory generation processing at step S51, it is determined the probability that the obstacle O _k passes through the predetermined area of the spatio-temporal 3-dimensional It becomes possible. Note that the spatio-temporal probability density is merely a concept of probability in spatio-temporal, so it is not always 1 when the sum of the values of one object in spatio-temporal is taken.

By the way, when a specific value of the trajectory generation time T is set in advance as a fixed value, the probability density distribution in space-time becomes uniform when the trajectory is generated beyond the value T. It is preferable to set a value that has no meaning even if it is calculated. For example, when the obstacle is a four-wheeled vehicle and the four-wheeled vehicle is running normally, it may be at most about T = 5 (sec). In this case, when the operation time Δt in step S504 is about 0.1 to 0.5 (sec), a series of processing from step S503 to step S507 is performed to generate one trajectory TR _k (m). Repeat 10 to 50 times.

  It should be noted that a method for reading different types of roads currently being traveled from map data using position data by setting different trajectory generation times T for different types of roads such as expressways, ordinary roads, and two-lane roads, image recognition, etc. It is preferable to perform switching by a method of reading the type of road by a road recognition device to which is applied.

  Also, the probability density distribution in space-time is statistically evaluated by using the trajectory calculated up to the trajectory generation time T. When the distribution is constant, the trajectory generation time T is reduced and the distribution becomes constant. If not, it is also preferable to perform adaptive control that increases the generation time.

After the trajectory generation process for each obstacle described above (step S51), the prediction calculation unit 14 performs probabilistic prediction of the course that each obstacle can take (step S52). Hereinafter, as a specific prediction calculation process in the prediction calculation unit 14, a case where a probability that a specific trajectory TR _k (m) is selected from the trajectory set {TR _k } generated for the obstacle O _k will be described. However, of course, this prediction calculation is only an example.

と求められる。したがって、障害物Ｏ_kにＭ_k本の軌跡集合｛ＴＲ_k｝が与えられたとき、障害物Ｏ_kが取りうる一つの軌跡ＴＲ_k（ｍ）が選ばれる確率ｐ（ＴＲ_k（ｍ））は、

となる。 When M _k trajectories of the obstacle O _k are generated, the probability p (TR _k (m)) that one of the trajectories TR _k (m) is an actual trajectory is calculated as follows. . First, the operation sequence {u _km (t)} for realizing the trajectory TR _k obstacle O _k (m) _{is, {u km (0),} u km (Δt), u km (2Δt), ·· , U _km (T)}, since the probability that the operation u _km (t) is selected at time t is p (u _km (t)), the operation sequence { The probability that u _km (t)} is executed is

Is required. Therefore, when the trajectory set of obstacle O _k to M _k the {TR _k} is given, the probability one trajectory TR _k (m) is selected to obstacle O _k can take p (TR _k (m)) Is

It becomes.

となる。ここで、ｓはｔ＝０からＴまでの操作時間Δｔの総数すなわち操作回数である。したがって、障害物Ｏ_kが取りうるＭ_k本の軌跡に含まれる軌跡ＴＲ_k（ｍ）の確率の総和はＭ_kｐ₀ ^sとなり、そのうちの１本の軌跡ＴＲ_k（ｍ）が選ばれる確率ｐ（ＴＲ_k（ｍ））は、式（４）を式（３）に代入することによって、

となる。すなわち、確率ｐ（ＴＲ_k（ｍ））は、軌跡ＴＲ_k（ｍ）に依存しない。 Here, when all operations u _km (t) are selected with equal probability p ₀ (where 0 <p ₀ <1), Equation (2) becomes

It becomes. Here, s is the total number of operation times Δt from t = 0 to T, that is, the number of operations. Therefore, the sum of the probabilities of the trajectories TR _k (m) included in the M _k trajectories that the obstacle O _k can take is M _k p ₀ ^s , and the probability that one of the trajectories TR _k (m) is selected. p (TR _k (m)) is obtained by substituting Equation (4) into Equation (3),

It becomes. That is, the probability p (TR _k (m)) does not depend on the trajectory TR _k (m).

In Equation (5), if the number of trajectories generated for all objects is the same (M), M ₁ = M ₂ =... = M _K = M (constant), so p. (TR _k (m)) = 1 / M, which is constant regardless of the obstacle O _k . In this case, by normalizing the value of the probability p (TR _k (m)) to 1, the prediction calculation in the prediction calculation unit 14 can be simplified, and a predetermined prediction calculation can be executed more quickly. .

Thereafter, the prediction calculation unit 14 determines each region of the three-dimensional space-time based on the probability p (TR _k (m)) calculated for each obstacle O _k (k = 1, 2,..., K). The existence probability of the obstacle Ok per unit volume at is _obtained . This existence probability corresponds to the spatio-temporal probability density in the three-dimensional space-time of the trajectory set {TR _k }, and the existence probability is generally large in the region where the density of the trajectory passing therethrough is high. The calculation result in the prediction calculation unit 14 is output to the collision probability calculation unit 6.

FIG. 20 is a diagram schematically illustrating a configuration example of the spatiotemporal environment formed by adding the predicted course TR ₀ of the host vehicle C _{0 to} the spatiotemporal environment Env (TR ₁ , TR ₂ ) illustrated in FIG. . The spatio-temporal environment Env (TR ₀ , TR ₁ , TR ₂ ) shown in the figure is that the own vehicle C ₀ and obstacles O ₁ , O ₂ are in a + y-axis direction on a flat and straight road R like an expressway. headed to are those showing a space-time environment when moving, 27, the other vehicle C ₁ is regarded as an obstacle O _1, corresponding to when regarded another vehicle C ₂ and the obstacle O ₂ ing.

  Next, calculation of the collision probability (corresponding to step S6 in FIG. 2) performed by the collision probability calculation unit 6 when the course of the obstacle based on the trajectory generation described above is performed will be described.

FIG. 21 is a flowchart showing details of the collision probability calculation process. In the following description, for convenience of description, in addition to the own vehicle C ₀ , the obstacles O _k (k = 1, 2,..., K) are all four-wheeled vehicles and are referred to as other vehicles O _k . Collision probability calculation process shown in FIG. 21 is composed of three loop processing, the locus TR ₀ of the subject vehicle C ₀ generated in step S3, all the trajectory set of the other vehicle O _k predicted in step S5 { The collision probability with TR _k } is calculated individually.

At this time, the collision probability calculating unit 6, the trajectory TR ₀ of the subject vehicle C _0, the trajectory set of the other vehicle O _k {TR _k}, and the subject vehicle C ₀ and the evaluation function for evaluating the collision probability of another vehicle O _k Use to calculate the collision probability. In the second embodiment, the description will be made assuming that the collision probability calculation unit 6 includes an evaluation function. However, by providing an input unit in the moving object surrounding risk determination device 11, the evaluation function is externally provided. It is good also as a structure which inputs. Further, it may be configured to adaptively vary the evaluation function by the speed of the road type and the vehicle C _0.

First, the collision probability calculating unit 6 starts the repetitive process (Loop1) for the other vehicle O _k (step S601). In Loop 1, the counter k for identifying other vehicles is initialized as k = 1, and the value of k is increased each time the repetition process is completed.

In Loop 1, the moving object surrounding risk determination device 11 determines, for each other vehicle O _k , all elements TR _k (m) (m = 1, 2,...) Of the trajectory set {TR _k } generated in step S5. An iterative process (Loop 2) is performed on M _k ) (step S602). In this repetitive processing, as the amount of vehicle C ₀ and the other vehicle O _k gives a degree interfere quantitatively, by introducing a degree of interference r (k) defined by the counter k for other vehicles identification, the degree of interference The initial value of r (k) is set to 0 (step S603).

Subsequently, the collision probability calculating unit 6 starts the iterative process of evaluating the interference between the trajectory TR _k trajectory TR ₀ and the other vehicle O _k of the vehicle _{C 0 (m) (Loop3)} ( step S604). In Loop 3, the distances at the same time between the two trajectories TR ₀ and TR _k (m) are sequentially obtained at times t = 0, Δt,. Since the position of each trajectory in the two-dimensional space is defined as the center of each vehicle, when the spatial distance between the two trajectories becomes smaller than a predetermined interference distance, the course of the own vehicle C ₀ and the like interfere with the path of the vehicle O _k, can be regarded as the subject vehicle C ₀ and the other vehicle O _k have collided.

Figure 22 is a diagram schematically showing the relationship between the on time 3-dimensional space of the locus TR ₀ and the other vehicle O _k trajectory TR _k (m) of the vehicle C _0. In the case shown in the figure, the trajectory TR ₀ and the trajectory TR _k (m) intersect at two points a ₁ and a ₂ . Therefore, in the vicinity of the two points a ₁ and a ₂ , there are regions A ₁ and A ₂ in which the distance at the same time between the two loci is smaller than the interference distance. That is, it is determined that the host vehicle C ₀ and the other vehicle O _k have collided at times when the two tracks TR ₀ and TR _k (m) are included in the areas A ₁ and A ₂ , respectively. In this sense, time t = 0,
Delta] t, · · ·, among T, then the number of passes through the area A ₁ and A _2, which is the number of times collision between the host vehicle C ₀ and the other vehicle O _k.

  As apparent from FIG. 22, in the spatiotemporal environment formed in the second embodiment, even if two trajectories collide once, the subsequent trajectory is generated. This is because the trajectory for each object is generated independently.

とする（ステップＳ６０６）。ここで、第２項目ｃ_k・ｐ（ＴＲ_k（ｍ））・Ｆ（ｔ）について説明する。係数ｃ_kは正の定数であり、例えばｃ_k＝１とおくことができる。また、
ｐ（ＴＲ_k（ｍ））は式（３）で定義される量であり、他車Ｏ_kで１本の軌跡ＴＲ_k（ｍ）が選ばれる確率である。最後のＦ（ｔ）は、１回の衝突における物体間の干渉の時間依存性を与える量である。したがって、自車Ｃ₀と他車Ｏ_kとの間の干渉に時間依存性を持たせない場合には、Ｆ（ｔ）の値を一定とすればよい。これに対して、自車Ｃ₀と他車Ｏ_kとの間の干渉に時間依存性を持たせる場合には、例えば図２３に示すように、時間が経過するとともに値が徐々に小さくなっていくような関数としてＦ（ｔ）を定義してもよい。図２３に示すＦ（ｔ）は、より直近の衝突を重要視する場合に適用される。 Thereafter, the collision probability calculating unit 6, the results of obtaining the distance of the subject vehicle C ₀ and the other vehicle O _k, when the vehicle C ₀ and the other vehicle O _k is determined to have collided in the sense described above (at step S605 Yes), the value of the interference r (k) is

(Step S606). Here, the second item c _k · p (TR _k (m)) · F (t) will be described. The coefficient c _k is a positive constant, and for example, c _k = 1 can be set. Also,
p (TR _k (m)) is an amount defined by the equation (3), and is a probability that one trajectory TR _k (m) is selected for the other vehicle O _k . The last F (t) is an amount that gives time dependency of interference between objects in one collision. Therefore, when the time dependency is not given to the interference between the host vehicle C ₀ and the other vehicle _Ok , the value of F (t) may be constant. On the contrary, when to have time-dependent interference between the subject vehicle C ₀ and the other vehicle O _k, for example, as shown in FIG. 23, gradually decreases the value with the passage of time F (t) may be defined as any function. F (t) shown in FIG. 23 is applied when the most recent collision is regarded as important.

  If the time t has not reached T after step S606, the collision probability calculation unit 6 does not end the repetition (No in step S607), increases the value of t by Δt (step S608), and step S604. Return to and repeat Loop3.

  On the other hand, if the time t has reached T after step S606, the collision probability calculation unit 6 ends Loop3 (Yes in step S607).

In the case where the subject vehicle C ₀ and the obstacle O _k do not collide (No at step S605) in step S605, the process proceeds directly to one of the determination processing whether to repeat the Loop3 (step S607).

Through the loop 3 iteration process described above, the value of the interference degree r (k) becomes larger as the number of collisions increases. After this Loop3 is completed, the collision probability calculation unit 6 performs a process of determining whether or not to repeat Loop2 (Step S609). That is, of the locus generated for the other vehicle O _k, if any are not subjected to interference evaluation with the trajectory TR ₀ of the subject vehicle C ₀ (in step S609 No), the m and m + 1 (step S610) Returning to Step S602, Loop 2 is repeated.

In contrast, when the interference evaluation of the locus TR ₀ of the subject vehicle C ₀ was performed for all the trajectories generated for the other vehicle O _k (Yes in step S609), the collision probability calculating unit 6, grant final interference degree r (k) to evaluate interference between total trajectory trajectory TR ₀ and the other vehicle O _k of the vehicle C ₀ (step S611), this imparting to interference degree r (k) Is stored in the storage unit 8 (step S612).

式（７）の右辺の和は、自車Ｃ₀の軌跡ＴＲ₀が他車Ｏ_kが取りうる軌跡と衝突する確率に他ならない。したがって、干渉度ｒ（ｋ）は衝突確率に等しいかまたは比例する量である。結局、式（７）により、ある時刻における自車Ｃ₀と他車Ｏ_kとの衝突確率が得られる。 In the formula (6), and constant regardless of coefficients c _k in k (e.g. c _k = 1), F (t) constant (e.g., 1) Distant, trace TR ₀ of the subject vehicle C ₀ and the other vehicle O _Assuming that the number of collisions of _{k with} the trajectory TR _k (m) is b _0k (m), the final value of the interference degree r (k) output in step S612 is the probability p for each trajectory TR _k (m). A value obtained by multiplying (TR _k (m)) by b _0k (m) is the sum of all the elements of the trajectory set {TR _k }.

The sum of the right side of Expression (7) is nothing but the probability that the trajectory TR ₀ of the host vehicle C ₀ collides with the trajectory that the other vehicle _Ok can take. Accordingly, the interference degree r (k) is an amount equal to or proportional to the collision probability. After all, the equation (7), a collision probability between the subject vehicle C ₀ and the other vehicle O _k at a certain time is obtained.

Subsequent to step S612, the collision probability calculation unit 6 performs a process of determining whether or not to repeat Loop1 (step S613). If the other vehicle O _k should perform interference evaluation with subject vehicle C ₀ is left (No at step S613), the collision probability calculating unit 6, the value of k is increased by one (step S614), returns to step S601 Repeat Loop1. On the other hand, if the other vehicle O _k should perform interference evaluation with subject vehicle C ₀ is not left (Yes at step S613), the collision probability calculating unit 6 terminates a series of collision probability calculation process.

  Thereafter, the dangerous obstacle detection unit 7 detects the dangerous obstacle (corresponding to step S7 in FIG. 2), and outputs information on the dangerous obstacle from the output unit 10 (corresponding to step S8 in FIG. 2). These processes are the same as the obstacle risk detection method according to the first embodiment.

  According to the second embodiment of the present invention described above, the latest collision probability between the moving body (own vehicle) and surrounding obstacles, and the amount of time change between the latest collision probability and the collision probability before a predetermined time. According to the criteria set forth using, the situation surrounding the moving object is predicted by detecting dangerous obstacles that should be considered dangerous to the moving object from the obstacles that exist around the moving object. Therefore, it is possible to determine the danger level around the moving object based on the change of the above, and to detect an obstacle that may increase the risk level for the moving object in the future. Therefore, for example, even when the vehicle is controlled based on the presence or absence of a dangerous obstacle, it becomes possible to perform control that avoids danger at an early stage, thereby realizing safe traveling of the moving object. Can do.

  Further, according to the second embodiment, when the predetermined time change amount of the collision probability is positive as a criterion for determining the dangerous obstacle, the obstacle is more dangerous as the time change amount is larger. Because the criterion that minimizes the minimum value of the latest collision probability when detecting as an object is used, an obstacle whose collision probability tends to increase in the future as a dangerous obstacle even if the collision probability is small It becomes possible to extract accurately.

  Furthermore, according to the second embodiment, a change in position that an obstacle can take with time is generated as a trajectory in time and space, and by using this generated trajectory, the path of a moving obstacle can be determined. Probabilistic prediction can be performed with high accuracy.

  In addition, according to the second embodiment, the degree of interference that quantitatively indicates the degree of interference between the trajectory that the obstacle can take in time and space and the trajectory that the vehicle can take is calculated. By using to calculate the collision probability, it is possible to calculate the collision probability between the moving object and the obstacle under a situation that can actually occur in a practical time.

In the second embodiment, when performing the trajectory generation process in the space-time of the obstacle, the trajectory may be generated by operating all selectable operations. An algorithm for realizing such a trajectory generation process can be realized by applying a recursive call by vertical search or horizontal search, for example. In this case, the number of elements i.e. the number of trajectories of the trajectory set of obstacle O _k {TR _k} is not known until the trajectory generation process with respect to the obstacle O _k is completed. Therefore, when a trajectory that each object can take is generated by searching all possible operations, the number of elements of the operation u _c (t) in the operation time Δt (operation u _c (t) is a continuous amount. In this case, a search method having an optimal calculation amount may be selected according to the degree of discretization.

  Moreover, this Embodiment 2 is applicable also in four-dimensional space-time (space three dimensions, time one dimension) like the case where it applies to the vehicle which is drive | working the road with a height difference.

(Embodiment 3)
In the third embodiment of the present invention, the entire obstacles around the moving body are regarded as one environment, and the environmental risk for the moving body in this environment is used to determine the risk around the moving body. It is characterized by.

  FIG. 24 is a block diagram illustrating a functional configuration of the moving object surrounding risk level determination device 21 according to the third embodiment. The moving object surrounding risk determination device 21 shown in the figure includes an environmental risk determining unit 22 that determines the risk of the environment around the moving object as a risk determination unit. The configuration of the other moving object surrounding risk determination device 21 is substantially the same as the structure of the moving object surrounding risk determination device 1 according to the first embodiment. For this reason, the same code | symbol is attached | subjected to the site | part which has the same function as the mobile body surrounding risk determination apparatus 1, respectively.

  FIG. 25 is a flowchart showing an outline of the process of the moving object surrounding risk determination method according to the third embodiment. In FIG. 25, the process of step S11-step S16 is the same as the process of step S1-step S6 in the mobile body surrounding risk determination method which concerns on the said Embodiment 1 (refer FIG. 2). Therefore, in the following, the processing after step S17 will be described. The course prediction process (step S15) for each obstacle and the collision probability calculation process (step S16) between the host vehicle and the obstacle may be performed in the same manner as in the second embodiment.

ここで、Ｎ（ｎ）は、時刻ｔ（ｎ）で環境を構成する障害物の数である。ＣＰ_k（ｎ）は、時刻ｔ_nにおける自車と障害物Ｏ_k（ｋ＝１,２,・・・,Ｎ（ｎ））との衝突確率であり、１−ＣＰ_k（ｎ）は、自車が障害物Ｏ_kと衝突しない確率を与える。この意味で、式（８）の右辺第２項の積は、自車が、環境を構成する全ての障害物と衝突しない確率を与える。したがって、式（８）で定義される環境危険度Ｄ（ｎ）は、少なくとも一つの障害物と衝突する確率に他ならない。 In step S17, the environmental risk determination unit 22 determines the risk of the surrounding environment with respect to the host vehicle (step S17). Specifically, the environmental risk determination unit 22 first calculates an environmental risk D (n), which is the overall environmental risk at an arbitrary time t (n), according to the following equation (8).

Here, N (n) is the number of obstacles that make up the environment at time t (n). CP _k (n) is a collision probability between the own vehicle and the obstacle O _k (k = 1, 2,..., N (n)) at time t _n , and 1-CP _k (n) is It gives the probability that the vehicle does not collide with an obstacle O _k. In this sense, the product of the second term on the right side of Equation (8) gives the probability that the vehicle will not collide with all obstacles that make up the environment. Therefore, the environmental risk D (n) defined by the equation (8) is nothing but the probability of colliding with at least one obstacle.

Next, the environmental risk determination unit 22 defines an environment defined as the amount of change between the environmental risk D (0) at the current time t ₀ and the environmental risk D (−1) at time t ₋₁ before that. Risk change over time ΔD = D (0) −D (−1) (9)
Is used to determine whether the environment at the current time t ₀ is dangerous for the vehicle.

FIG. 26 is a diagram schematically illustrating a determination criterion used when the environmental risk determination unit 22 determines the environmental risk. The environmental risk level determination unit 22 determines that the set (D (0), ΔD) of the environmental risk level D (0) at time t ₀ and the time change amount ΔD of the environmental risk level defined by Expression (9) is as shown in FIG. If it belongs to 26 areas Sγ, it is determined that the environment is dangerous for the vehicle.

  A curve γ defining the boundary of the region Sγ is a region in which ΔD> 0 and the environmental risk D (0) decreases as the time change amount ΔD increases. By using the region Sγ delimited by such a curve γ as a determination criterion, the larger the time change amount ΔD (> 0) of the environmental risk, the larger the latest in determining that the environment is more dangerous. The minimum value of the environmental risk degree D (0) becomes small, and it becomes possible to detect an environment that is likely to be dangerous for the vehicle in the future.

  Thereafter, the output unit 10 outputs information according to the determination result of the environmental risk determination unit 22 (step S18). Specifically, the display unit 101 displays that the present is in a dangerous environment, and the warning sound generating unit 102 generates a warning sound to notify that the present is in a dangerous environment.

  According to the third embodiment of the present invention described above, the entire obstacle existing around the moving body (own vehicle) is regarded as one environment, and the latest collision probability between the moving body and the surrounding obstacle, And by determining the environmental risk around the vehicle according to the criteria determined using the amount of time change between the latest collision probability and the collision probability before a predetermined time. It is possible to determine the degree of danger around the moving object after capturing the change in the situation of

  Further, according to the third embodiment, when the time change amount of the environmental risk level determined using the collision probability between the moving object and the obstacle is positive as a determination criterion when determining the dangerous obstacle, Since the criterion is such that the minimum value of the environmental risk level when determining that the risk is higher is greater as the amount of time change is larger, the risk level will increase in the future even if the risk level is small at this time. It becomes possible to determine a tendency environment as a dangerous environment at an early stage. Therefore, for example, even when the own vehicle is controlled according to the environmental risk level, it is possible to perform control so as to avoid danger at an early stage, thereby realizing safe traveling of the moving object. it can.

  In addition, as a criterion for determining the environmental risk by the environmental risk determination unit 22, an area having a boundary with a curve exhibiting the same behavior as the curve β (see FIG. 10) described in the first embodiment is used. It may be set. The curve that defines the boundary of this region is a region where ΔD> 0, and the environmental risk D (0) approaches a constant value while decreasing with increasing time variation ΔD, while ΔD <0. In this area, the environmental risk D (0) approaches a constant value while increasing as the time variation ΔD decreases.

この場合には、式（９）よりも計算量が少なくて済むため、現時刻における環境が、自車にとって危険な環境であるか否かを、より迅速に判定することができる。 Moreover, you may employ | adopt definitions other than Formula (9) as environmental risk degree D (n). For example, an environmental risk D (n) defined by a simple sum of the collision probabilities CP _k (n) of the obstacle O _k may be applied as in the following equation (10).

In this case, since the calculation amount is smaller than that in the equation (9), it can be more quickly determined whether or not the environment at the current time is a dangerous environment for the host vehicle.

(Other embodiments)
Up to this point, the first to third embodiments have been described in detail as the best mode for carrying out the present invention. However, the present invention should not be limited only by these three embodiments. For example, the present invention can be applied to an automatic driving system. In this case, an operation signal for operating the own vehicle is generated in response to the output of the moving object surrounding risk determination device (the course prediction result or the collision probability with the own vehicle), and this operation signal is provided in the own vehicle. What is necessary is just to make it transmit to the predetermined | prescribed actuator apparatus.

  In addition, when predicting the path of an obstacle in the moving object surrounding risk determination device according to the present invention, an imaginary obstacle is arranged in addition to the actual obstacle detected by the sensor unit, and the imaginary obstacle is detected. The course prediction may be performed together. More specifically, an imaginary model that exhibits unfavorable behavior for the host vehicle may be configured, and the route prediction may be performed by placing this model at a predetermined position. Such an imaginary model, for example, by colliding with an obstacle that may jump out of the intersection by placing it at a position where it cannot be detected from the own vehicle traveling in the vicinity of the intersection where there is a shield or the like and the visibility is bad It is possible to predict such dangers. Note that the information of the imaginary model may be stored in advance in the storage unit, and may be arranged at a desired position according to the condition setting from the input unit provided separately.

  When the mobile object surrounding risk determination device according to the present invention is applied in a region such as an expressway on which traveling of only a vehicle is assumed, each vehicle is provided with communication means for inter-vehicle communication. Alternatively, vehicles traveling close to each other may exchange each other's traveling status by inter-vehicle communication. In this case, the operation history of each vehicle is stored in its own storage unit, an operation selection probability for each operation is given based on the operation history, and information on the operation selection probability is also added to other vehicles. You may make it transmit. As a result, the accuracy of the route prediction is increased, and it is possible to more reliably avoid danger during traveling.

  By the way, in this invention, it is also possible to use GPS (Global Positioning System) as a position detection means. In this case, the position information and movement information of the object detected by the sensor unit can be corrected by referring to the 3D map information stored in the GPS. Furthermore, it is also possible to function as a sensor unit by communicating GPS outputs with each other. In any case, highly accurate course prediction can be realized by using GPS, and the reliability of the prediction result can be further improved.

  The mobile object surrounding risk determination device according to the present invention can be mounted on a mobile object such as a vehicle, a person, or a robot other than a four-wheeled vehicle.

  Moreover, the mobile body surrounding risk determination apparatus according to the present invention does not need to be mounted on the mobile body. For example, when the own vehicle can use inter-vehicle communication or road-to-vehicle communication, the mobile object surrounding risk determination device according to the present invention includes a route interference evaluation system including the own vehicle, other vehicles around the own vehicle, and infrastructure. It can consist of In this case, the course prediction calculation of the obstacle is performed on the infrastructure side, and for the own vehicle, a prediction calculation request is made to request and receive the prediction calculation result from the infrastructure side and perform processing based on the received prediction calculation result. It can also be specified as a vehicle.

  As is apparent from the above description, the present invention can include various embodiments and the like not described herein, and within the scope not departing from the technical idea specified by the claims. Various design changes and the like can be made.

It is a block diagram which shows the function structure of the mobile body surrounding risk determination apparatus which concerns on Embodiment 1 of this invention. It is a flowchart which shows the outline | summary of a process of the moving body surrounding risk determination method which concerns on Embodiment 1 of this invention. It is a figure which shows the example of the probability distribution provided with respect to the course of an obstruction. It is a figure which shows typically the memory | storage aspect of the collision probability in a collision probability memory | storage part. It is a figure which shows the characteristic of the table which memorize | stores a collision probability. It is a figure which shows the condition where information is written in time series with respect to two tables which memorize | store the collision probability which adjoins temporally (the 1). It is a figure which shows the condition where information is written in time series with respect to two tables which memorize | store the collision probability which adjoins temporally (the 2). It is a figure which shows the condition where information is written in time series with respect to two tables which memorize | store the collision probability which adjoins temporally (the 3). It is a figure which shows typically the criteria used when a dangerous obstacle detection part detects a dangerous obstacle. It is a figure which shows typically another criteria when a dangerous obstacle detection part detects a dangerous obstacle. It is a figure which shows the example of a time change of the collision probability of the own vehicle and an obstruction. It is a figure which shows the example of a detection of a dangerous obstacle using the criterion shown in FIG. It is a figure which shows the example of detection of a dangerous obstacle using the criterion shown in FIG. It is a block diagram which shows the function structure of the mobile body surrounding risk determination apparatus which concerns on Embodiment 2 of this invention. It is a flowchart which shows the outline | summary of the obstacle course prediction process in the mobile body surrounding risk determination method which concerns on Embodiment 2 of this invention. It is a flowchart which shows the detail of the locus | trajectory production | generation process of an obstruction. It is a figure which shows typically the locus | trajectory of an obstruction on three-dimensional space time. It is a figure which shows typically the locus | trajectory set produced | generated with respect to the obstacle on three-dimensional space time. It is a figure which shows typically the structural example of the spatiotemporal environment formed by the locus | trajectory set of a several obstruction. It is a figure which shows typically the structural example of the spatiotemporal environment formed by adding the course of the own vehicle to the locus | trajectory set of a several obstruction. It is a flowchart which shows the outline | summary of the collision probability calculation process of the mobile body surrounding risk determination method which concerns on Embodiment 2 of this invention. It is a figure which shows typically the relationship in the three-dimensional space-time with the locus | trajectory of the own vehicle, and the locus | trajectory of an obstruction. It is a figure which shows the example of the function which gives the time dependence of the interference between objects. It is a block diagram which shows the function structure of the mobile body surrounding risk determination apparatus which concerns on Embodiment 3 of this invention. It is a flowchart which shows the outline | summary of a process of the moving body surrounding risk determination method which concerns on Embodiment 3 of this invention. It is a figure which shows typically the criteria used when an environmental risk determination part determines an environmental risk. It is a figure which shows the condition where the own vehicle and two other vehicles are drive | working the road.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1,11,21 Moving body surrounding risk determination apparatus 2 Sensor part 3 Own vehicle course production | generation part 4 Obstacle extraction part 5,12 Obstacle course prediction part 6 Collision probability calculation part 7 Danger obstacle detection part 8 Storage part 9 Collision Probability input / output unit 10 Output unit 13 Trajectory generation unit 14 Prediction calculation unit 22 Environmental risk level determination unit 81 Collision probability storage unit 101 Display unit 102 Warning sound generation unit 131 Operation selection unit 132 Object operation unit 133 determination unit

Claims (15)

A computer having a storage means for storing at least obstacle information including the position and internal state of obstacles existing around the mobile body is a mobile body ambient risk determination method for determining the risk around the mobile body. There,
Based on the position and internal state of the mobile body, a mobile body path generation step for generating a path of the mobile body;
An obstacle course prediction step for probabilistically predicting a course that the obstacle can take based on the obstacle information read from the storage means;
Of the paths that can be taken by the obstacle predicted in the obstacle course prediction step, by obtaining a path that interferes with the path of the mobile body generated in the mobile body path generation step, the mobile body and the obstacle A collision probability calculation step for calculating the collision probability of
The risk determination for determining the risk around the moving object by using the latest collision probability calculated in the collision probability calculation step and the time variation between the latest collision probability and the collision probability before a predetermined time. Steps,
A moving object surrounding risk determination method characterized by comprising: The risk determination step includes
2. A moving object surrounding hazard according to claim 1, wherein a dangerous obstacle to be considered dangerous for the moving object is detected from obstacles existing around the moving object according to a predetermined criterion. Degree determination method. The criterion is
The minimum value of the latest collision probability when detecting the obstacle as a dangerous obstacle is smaller as the time change amount is larger when the time change amount is positive. A method for judging the risk around moving objects. The risk determination step includes
The method according to claim 1, wherein a risk level of the mobile object in an environment including all obstacles existing around the mobile object is determined according to a predetermined criterion. The criterion is
When the time change amount of the environmental danger level of the environment determined by using the collision probability between the moving object and the obstacle is positive, it is determined that the larger the time change amount is, the more dangerous it is. 5. The method of determining a surrounding environmental risk according to claim 4, wherein a minimum value of the environmental risk is small. The obstacle course prediction step includes:
A trajectory generating step for generating, based on the position and internal state of the obstacle, a change in position that the obstacle can take as time elapses as a trajectory on time and space composed of time and space;
A prediction calculation step for performing a probabilistic prediction calculation of the course of the obstacle by using the trajectory generated in the trajectory generation step;
The mobile surrounding risk determination method according to any one of claims 1 to 5, wherein:   The mobile body surrounding risk determination method according to any one of claims 1 to 6, further comprising an output step of outputting information according to a result determined in the risk determination step. Based on the position and internal state of the moving body, a moving body path generating means for generating a path of the moving body;
Based on the obstacle information read from the storage means for storing at least obstacle information including the position and internal state of obstacles existing around the mobile body, the path that the obstacle can take is predicted probabilistically. Obstacle course prediction means,
Of the paths that can be taken by the obstacle predicted by the obstacle path prediction means, the path that interferes with the path of the mobile body generated by the mobile body path generation means is obtained, whereby the mobile body and the obstacle A collision probability calculation means for calculating the collision probability of
The risk determination for determining the risk around the moving object by using the latest collision probability calculated by the collision probability calculation means and the time variation between the latest collision probability and the collision probability before a predetermined time. Means,
A moving object surrounding risk determination device comprising: The risk determination means includes
9. The danger around the moving body according to claim 8, wherein a dangerous obstacle that should be regarded as dangerous to the moving body is detected from obstacles around the moving body according to a predetermined criterion. Degree determination device. The criterion is
The minimum value of the latest collision probability when detecting the obstacle as a dangerous obstacle is smaller as the time change amount is larger when the time change amount is positive. Mobile object surrounding risk assessment device. The risk determination means includes
9. The mobile object surrounding risk determination device according to claim 8, wherein a risk level for the mobile object in an environment including all obstacles existing around the mobile object is determined according to a predetermined determination criterion. The criterion is
When the time change amount of the environmental danger level of the environment determined by using the collision probability between the moving object and the obstacle is positive, it is determined that the larger the time change amount is, the more dangerous it is. 12. The moving object surrounding risk determination device according to claim 11, wherein a minimum value of the environmental risk is small. The obstacle course prediction means includes
Based on the position and internal state of the obstacle, a trajectory generating means for generating a change in position that the obstacle can take as time passes as a trajectory on time and space composed of time and space;
Prediction calculation means for performing probabilistic prediction calculation of the path of the obstacle by using the trajectory generated by the trajectory generation means;
The mobile object surrounding risk determination device according to any one of claims 8 to 12, characterized by comprising:   The mobile body surrounding risk determination device according to any one of claims 8 to 13, further comprising output means for outputting information according to a result determined by the risk determination means.   8. A moving object surrounding risk determination program that causes the computer to execute the moving object surrounding risk determination method according to claim 1.

Images