Electronic device, program, and information processing method
Patent Information
- Application Number
- JP2024118927
- Authority / Receiving Office
- JP · JP
- Patent Type
- Applications
- Current Assignee / Owner
- Filing Date
- 2024-07-24
- Publication Date
- 2026-02-05
Smart Images
Figure 2026017882000001_ABST
Abstract
Description
[Technical Field]
[0001] The present disclosure relates to an electronic device, a program, and an information processing method. [Background technology]
[0002] For example, it is important for an automatic delivery robot to recognize pedestrians and predict their movement paths (trajectories) in order to avoid collisions. For example, Patent Document 1 discloses a moving object prediction device that ensures the accuracy of prediction of the future positions of moving objects while reducing the calculation load. [Prior art documents] [Patent documents]
[0003] [Patent Document 1] Japanese Patent Application Publication No. 2018-124663 Summary of the Invention [Problem to be solved by the invention]
[0004] Here, the Social Force Model (SFM) is known as a method for predicting a pedestrian's movement location. The Social Force Model models a pedestrian as being subject to an attractive force (attractive force) that moves them toward their destination, a repulsive force (first repulsive force) that tries to avoid obstacles, and a repulsive force (second repulsive force) that tries to avoid other pedestrians. The Social Force Model then predicts the pedestrian's next movement location by calculating the resultant force of the attractive force, the first repulsive force, and the second repulsive force.
[0005] In general, the use of social force models can improve the accuracy of predicting pedestrian movement locations. However, it is known that the first repulsive force can act excessively. This can lead to a problem of reduced accuracy in predicting pedestrian trajectories.
[0006] In view of the above circumstances, an object of the present disclosure is to provide an electronic device, a program, and an information processing method that can improve the accuracy of predicting the location of a pedestrian. [Means for solving the problem]
[0007] (1) An electronic device according to an embodiment of the present disclosure includes: an acquisition unit that acquires image data of a pedestrian and the surroundings of the pedestrian; a prediction unit that predicts the position where the pedestrian will move using a social force model that calculates a resultant force of an attractive force that moves the pedestrian toward the destination, a first repulsive force that is a repulsive force that the pedestrian tries to avoid an obstacle, and a second repulsive force that is a repulsive force that the pedestrian tries to avoid other pedestrians; The prediction unit Recognizing the pedestrian and the obstacle based on the image data, and calculating a velocity vector calculated from the velocity of the pedestrian and an obstacle position vector directed from the pedestrian to the obstacle; The weight based on θ, which is the angle between the velocity vector and the obstacle position vector, is set as ω θ The repulsive force vector from the obstacle determined according to the distance between the pedestrian and the obstacle is defined as f1, and a predetermined weight is defined as ω φ The prediction is performed using f1', which is a corrected repulsive force vector calculated by the following formula, as the first repulsive force.
number
[0008] (2) As one embodiment of the present disclosure, in (1), Said ω θ The value of decreases as the value of θ increases.
[0009] (3) As an embodiment of the present disclosure, in (1) or (2), The predetermined values θ1 and θ2 satisfy 0<θ1≦θ2, Said ω θThe value of is 1 when the θ satisfies 0≦θ<θ1, and is 0 when the θ satisfies θ2≦θ.
[0010] (4) An electronic device according to an embodiment of the present disclosure includes: an acquisition unit that acquires image data of a pedestrian and the surroundings of the pedestrian; a prediction unit that predicts the position where the pedestrian will move using a social force model that calculates a resultant force of an attractive force that moves the pedestrian toward the destination, a first repulsive force that is a repulsive force that the pedestrian tries to avoid an obstacle, and a second repulsive force that is a repulsive force that the pedestrian tries to avoid other pedestrians; The prediction unit Recognizing the pedestrian and the obstacle based on the image data, and calculating a destination direction vector from the pedestrian to the destination and a normal vector determined based on the shape of the obstacle; A weight based on φ, which is the angle between the destination direction vector and the normal vector, is set as ω φ The repulsive force vector from the obstacle determined according to the distance between the pedestrian and the obstacle is defined as f1, and a predetermined weight is defined as ω θ The prediction is performed using f1', which is a corrected repulsive force vector calculated by the following formula, as the first repulsive force.
number
[0011] (5) As an embodiment of the present disclosure, in (4), The normal vector is calculated based on a plurality of detection points including a detection point on the obstacle and other detection points present around the detection point.
[0012] (6) As an embodiment of the present disclosure, in (5), The normal vector is calculated by principal component analysis using the plurality of detection points.
[0013] (7) As an embodiment of the present disclosure, in any one of (4) to (6), Said ω φ The value of decreases as φ increases.
[0014] (8) As an embodiment of the present disclosure, in any one of (4) to (7), The predetermined values φ1 and φ2 satisfy 0<φ1≦φ2, Said ω φ The value of is 1 when φ satisfies 0≦φ<φ1, and is 0 when φ satisfies φ2≦φ.
[0015] (9) As an embodiment of the present disclosure, in any one of (1) to (8), The prediction unit causes a display unit to display a prediction result image indicating a predicted trajectory of the position of the pedestrian.
[0016] (10) As an embodiment of the present disclosure, in any one of (1) to (9), When i is an integer equal to or greater than 1 and the corrected repulsive force vector targeting the i obstacles is F, F is calculated by the following formula.
number
[0017] (11) A program according to an embodiment of the present disclosure includes: Computer, an acquisition unit that acquires image data of a pedestrian and the surroundings of the pedestrian; a social force model that calculates the resultant force of an attractive force that moves the pedestrian toward the destination, a first repulsive force that is a repulsive force that the pedestrian tries to avoid an obstacle, and a second repulsive force that is a repulsive force that the pedestrian tries to avoid other pedestrians, and causes the social force model to function as a prediction unit that predicts the position where the pedestrian will move, The prediction unit Recognizing the pedestrian and the obstacle based on the image data, and calculating a velocity vector calculated from the velocity of the pedestrian and an obstacle position vector directed from the pedestrian to the obstacle; The weight based on θ, which is the angle between the velocity vector and the obstacle position vector, is set as ω θ The repulsive force vector from the obstacle determined according to the distance between the pedestrian and the obstacle is defined as f1, and a predetermined weight is defined as ω φ The prediction is performed using f1', which is a corrected repulsive force vector calculated by the following formula, as the first repulsive force.
number
[0018] (12) A program according to an embodiment of the present disclosure includes: Computer, an acquisition unit that acquires image data of a pedestrian and the surroundings of the pedestrian; a social force model that calculates the resultant force of an attractive force that moves the pedestrian toward the destination, a first repulsive force that is a repulsive force that the pedestrian tries to avoid an obstacle, and a second repulsive force that is a repulsive force that the pedestrian tries to avoid other pedestrians, and causes the social force model to function as a prediction unit that predicts the position where the pedestrian will move, The prediction unit Recognizing the pedestrian and the obstacle based on the image data, and calculating a destination direction vector from the pedestrian to the destination and a normal vector determined based on the shape of the obstacle; A weight based on φ, which is the angle between the destination direction vector and the normal vector, is set as ω φ The repulsive force vector from the obstacle determined according to the distance between the pedestrian and the obstacle is defined as f1, and a predetermined weight is defined as ω θ The prediction is performed using f1', which is a corrected repulsive force vector calculated by the following formula, as the first repulsive force.
number
[0019] (13) An information processing method according to an embodiment of the present disclosure includes: An information processing method executed by an electronic device, an acquisition step of acquiring image data of a pedestrian and the surroundings of the pedestrian; a prediction step of predicting a position where the pedestrian will move using a social force model that calculates a resultant force of an attractive force that moves the pedestrian toward the destination, a first repulsive force that is a repulsive force that the pedestrian tries to avoid an obstacle, and a second repulsive force that is a repulsive force that the pedestrian tries to avoid other pedestrians; The prediction step Recognizing the pedestrian and the obstacle based on the image data, and calculating a velocity vector calculated from the velocity of the pedestrian and an obstacle position vector directed from the pedestrian to the obstacle; The weight based on θ, which is the angle between the velocity vector and the obstacle position vector, is set as ω θ The repulsive force vector from the obstacle determined according to the distance between the pedestrian and the obstacle is defined as f1, and a predetermined weight is defined as ω φ The prediction is performed using f1', which is a corrected repulsive force vector calculated by the following formula, as the first repulsive force.
number
[0020] (14) An information processing method according to an embodiment of the present disclosure includes: An information processing method executed by an electronic device, an acquisition step of acquiring image data of a pedestrian and the surroundings of the pedestrian; a prediction step of predicting a position where the pedestrian will move using a social force model that calculates a resultant force of an attractive force that moves the pedestrian toward the destination, a first repulsive force that is a repulsive force that the pedestrian tries to avoid an obstacle, and a second repulsive force that is a repulsive force that the pedestrian tries to avoid other pedestrians; The prediction step Recognizing the pedestrian and the obstacle based on the image data, and calculating a destination direction vector from the pedestrian to the destination and a normal vector determined based on the shape of the obstacle; A weight based on φ, which is the angle between the destination direction vector and the normal vector, is set as ω φ The repulsive force vector from the obstacle determined according to the distance between the pedestrian and the obstacle is defined as f1, and a predetermined weight is defined as ω θ The prediction is performed using f1', which is a corrected repulsive force vector calculated by the following formula, as the first repulsive force.
number
[0021] According to the present disclosure, it is possible to provide an electronic device, a program, and an information processing method that can improve the accuracy of predicting the location of a pedestrian. [Brief explanation of the drawings]
[0022] [Figure 1] FIG. 1 is a diagram illustrating an example of the configuration of an electronic device according to an embodiment of the present disclosure. [Figure 2A] FIG. 2A is a diagram for explaining the problem with the conventional technology. [Figure 2B] FIG. 2B is a diagram for explaining the problem with the conventional technology. [Figure 3] FIG. 3 is a diagram for explaining the repulsive force vector after correction. [Figure 4] FIG. 4 is a flowchart showing the processing of an information processing method according to an embodiment of the present disclosure. [Figure 5] FIG. 5 is a diagram comparing an example of the prior art with an example of this embodiment. DETAILED DESCRIPTION OF THE INVENTION
[0023] Hereinafter, an electronic device 10 (see FIG. 1), a program, and an information processing method according to an embodiment of the present disclosure will be described with reference to the drawings.
[0024] 1 is a diagram showing an example of the configuration of an electronic device 10 according to this embodiment. The electronic device 10 is used, for example, in an automatic delivery robot, and executes an information processing method (see FIG. 4) described below to avoid collisions with people, and predicts the next location (future location) to which people will move. In this embodiment, the people are described as pedestrians.
[0025] 1, electronic device 10 includes an acquisition unit 11, a prediction unit 12, and a storage unit 13. The hardware configuration of electronic device 10 may be, for example, a computer. The computer may be, for example, a small computer mounted on an automatic delivery robot or the like, or may be, for example, a server computer. The components of electronic device 10 will be described in detail below.
[0026] Here, electronic device 10 may not be a single device, but may be composed of multiple devices located in multiple locations and capable of transmitting and receiving data to and from each other via a network. In other words, multiple devices connected via a network may function as electronic device 10 as a whole as shown in Fig. 1. Therefore, for example, electronic device 10 may be composed of a single computer as a hardware configuration, or may be composed of multiple computers connected via a network. When composed of multiple computers, a shared memory accessible by each computer or a storage device on the network may be used.
[0027] Generally, a computer includes, for example, a memory, a hard disk drive (storage device), a CPU (processing device), a display, etc. When the electronic device 10 is realized by a computer, a program may cause the processing device to function as the acquisition unit 11 and the prediction unit 12. The storage device may function as the memory unit 13. Here, the memory unit 13 may be any other storage medium such as RAM, ROM, HDD, SSD, etc. The memory unit 13 may store various programs and data, such as an OS and application software, used by the electronic device 10.
[0028] The electronic device 10 according to this embodiment acquires image data of a pedestrian and the pedestrian's surroundings from a camera, recognizes the pedestrian, and predicts the pedestrian's future location. The predicted pedestrian's future location is used, for example, as data for collision avoidance by an automatic delivery robot. Here, the camera may be a camera that captures images of the road or a camera mounted on the automatic delivery robot. The image data includes not only images but also information that allows the positions of objects contained in the images, such as pedestrians and obstacles, to be calculated. For example, the camera may be a stereo camera or a device with a LiDAR (Light Detection and Ranging) function. When a stereo camera is used, the position and shape of an object are calculated based on parallax data contained in the image data. When LiDAR is used, the position and shape of an object are calculated based on a set of detection points (point cloud) contained in the image data. The electronic device 10 may recognize objects, such as pedestrians and obstacles, from the image data and calculate their positions and shapes using a known method. When the electronic device 10 is used in an automatic delivery robot, for example, the electronic device 10 may display a prediction result image on a display used by an administrator managing the automatic delivery robot. The prediction result image is an image showing the predicted locus of the pedestrian's movement position (see FIG. 5).
[0029] Here, the social force model is known as a method for predicting a pedestrian's movement location. In the social force model, a pedestrian is modeled as being subject to an attractive force (attractive force) that moves them toward their destination, a repulsive force (first repulsive force) that tries to avoid obstacles, and a repulsive force (second repulsive force) that tries to avoid other pedestrians. The social force model then predicts the pedestrian's next movement location by calculating the resultant force of the attractive force, the first repulsive force, and the second repulsive force.
[0030] 2A and 2B are diagrams illustrating the problems of conventional technology using a social force model. FIG. 2A shows a reference trajectory indicating the actual movement position of a pedestrian and a trajectory (predicted trajectory) predicted by conventional technology using a social force model. In the example shown in FIG. 2A, there is a parked vehicle to the right of the pedestrian's direction of travel. However, the pedestrian does not take evasive action because the vehicle does not obstruct his or her progress. However, conventional technology using a social force model sets a large first repulsive force for such a vehicle. FIG. 2B shows the first repulsive force calculated by the conventional social force model, separated into the X and Y directions. As shown in FIG. 2A, the X and Y directions correspond to the up / down and left / right directions in the image, respectively. In FIG. 2B, at time 0 (when the pedestrian is next to the parked vehicle), the first repulsive force in the X direction is large. Here, a negative value in the X direction indicates a force acting downward in the image, and the absolute value indicates the magnitude of the force. Because the first repulsive force acts excessively, the predicted trajectory deviates significantly in the negative X direction and deviates from the pedestrian's actual trajectory (reference trajectory). In other words, the prediction accuracy of the pedestrian's future position is reduced.
[0031] In contrast, the electronic device 10 according to the present embodiment uses a weight (ω θ and ω φ ) is introduced to calculate the first repulsive force (f i In the electronic device 10 according to the present embodiment, the corrected repulsive force vector (f i By using the first repulsive force, it is possible to improve the accuracy of predicting the position where the pedestrian will move. Here, the subscript i is an integer equal to or greater than 1 and is used to indicate each obstacle. There may be multiple obstacles, but in the following, when focusing on one obstacle, the explanation will be given by setting "i=1" (i.e., replacing i with 1) without any special mention.
[0032] Referring again to FIG. 1 , the components of the electronic device 10 will be described. The acquisition unit 11 acquires image data from a camera. The image data includes information on images of a pedestrian and the pedestrian's surroundings. As described above, the image data includes not only the image but also information that allows the positions of objects, such as the pedestrian and obstacles, included in the image to be calculated. Furthermore, the acquisition unit 11 acquires the image data, which is time-series data, so that the speed and direction of movement of the object can be identified.
[0033] The prediction unit 12 predicts the position where the pedestrian will move using a social force model that calculates the resultant force of an attractive force that moves the pedestrian toward the destination, a first repulsive force that is a repulsive force that moves the pedestrian to avoid an obstacle, and a second repulsive force that is a repulsive force that moves the pedestrian to avoid other pedestrians. The prediction unit 12 recognizes objects using a known method based on the image data acquired by the acquisition unit 11, calculates their positions and shapes, and predicts the pedestrian's future position using the social force model, but corrects for the first repulsive force as described above.
[0034] FIG. 3 is a diagram illustrating the repulsive force vector after correction. The correction performed by the prediction unit 12 will be described with reference to FIG. 3. The prediction unit 12 recognizes a pedestrian and an obstacle based on image data, and calculates a velocity vector calculated from the pedestrian's speed and an obstacle position vector indicating the direction from the pedestrian to the obstacle. The prediction unit 12 also calculates a destination direction vector indicating the direction from the pedestrian to the destination and a normal vector determined based on the shape of the obstacle. Before calculating the destination direction vector, the prediction unit 12 may acquire information about the destination or estimate the pedestrian's destination. For example, if the pedestrian is moving straight from right to left in the image before calculating the destination direction vector, the prediction unit 12 can estimate that the destination is on an extension of the straight path. The prediction unit 12 may, for example, identify a long side of the obstacle based on its shape in a two-dimensional image and determine a normal vector perpendicular to the long side. For example, when LiDAR is used, the position and shape of the object may be represented by a set of detected points (point cloud). In this case, the normal vector may be calculated based on a plurality of detection points including a detection point on the obstacle and other detection points existing around the detection point. As a specific example, the normal vector may be calculated by principal component analysis using the plurality of detection points. Here, it is not necessary to use all of the detection points on the obstacle in calculating the normal vector, and the calculation may be performed using, for example, a representative portion of the detection points.
[0035] The prediction unit 12 performs prediction using f1', which is a corrected repulsive force vector calculated by the following equation (1), as the first repulsive force.
[0036]
number
[0037] Here, f1 is a repulsive force vector from the obstacle determined according to the distance between the pedestrian and the obstacle, and is a repulsive force vector used as the first repulsive force itself in conventional technology using a social force model. θ is the angle between the velocity vector and the obstacle position vector. Also, ω θis a weight based on θ.
[0038] ω θ The value of ω is set to decrease as θ increases. Here, it is assumed that the predetermined values θ1 and θ2 satisfy 0<θ1≦θ2. In this case, ω θ The value of ω may be 1 when θ satisfies 0≦θ<θ1, and may be 0 when θ satisfies θ2≦θ. θ By introducing θ1 and θ2, if an obstacle exists in the pedestrian's immediate destination (θ is small), the repulsive force vector from the obstacle is taken into consideration, but if an obstacle exists behind the pedestrian (θ is large), for example, no avoidance action is taken. For example, θ1 and θ2 can be "π / 2" (rad) or a value close to it. ω as shown in the graph in Figure 3 θ As the function having the above characteristic, a monotonically decreasing function may be used, such as a step function, a sigmoid function, or a linear function.
[0039] φ is the angle between the destination direction vector and the normal vector. φ is a weight based on φ.
[0040] ω φ The value of ω is set to decrease as φ increases. Here, it is assumed that the predetermined values φ1 and φ2 satisfy 0<φ1≦φ2. In this case, ω φ The value of ω may be 1 when φ satisfies 0≦φ<φ1, and may be 0 when φ satisfies φ2≦φ. φ By introducing φ, if an obstacle exists on the path to the destination (φ is small), the repulsive force vector from the obstacle is taken into consideration, but if an obstacle exists parallel to the path (φ is large), it can be expressed that no avoidance action is taken. For example, φ1 and φ2 can be "π / 2" (rad) or a value close to it. ω as shown in the graph of Figure 3 φ As the function having the above characteristic, a monotonically decreasing function may be used, such as a step function, a sigmoid function, or a linear function.
[0041] The prediction unit 12 predicts the future position of the pedestrian using a social force model, with f1', the corrected repulsive force vector calculated by the above formula (1), as the first repulsive force. The prediction unit 12 may generate a prediction result image by, for example, superimposing the trajectory of the predicted position of the pedestrian on an image of the image data. Then, the prediction unit 12 may display the prediction result image on the display unit.
[0042] Here, if there are multiple obstacles, the corrected repulsive force vector for each obstacle can be calculated using the above formula (1), and the sum of these vectors can be used in predictions using the social force model. If the corrected repulsive force vector for i obstacles is F, F is calculated using the following formula (2):
[0043]
number
[0044] Here, the prediction unit 12 calculates ω θ or ω φ For example, when there is no wall as an obstacle in the direction of the destination, the prediction unit 12 calculates ω φ is a predetermined weight (e.g., a predetermined constant), and then ω θ Alternatively, when, for example, importance is placed on the influence of a wall present in the direction of the destination, the prediction unit 12 may calculate ω θ is a predetermined weight (e.g., a predetermined constant), and then ω φ may be calculated exactly.
[0045] 4 is a flowchart showing the process of the information processing method executed by the electronic device 10 according to this embodiment. By executing the information processing method described below, the electronic device 10 can improve the accuracy of predicting the position to which a pedestrian will move.
[0046] The acquisition unit 11 acquires image data from the camera (step S1, acquisition step).
[0047] The prediction unit 12 recognizes pedestrians and obstacles based on the acquired image data (step S2).
[0048] The prediction unit 12 predicts the repulsive force vector from the obstacle (f in Equation (1)). 1、 f in equation (2) i ) is calculated (step S3).
[0049] The prediction unit 12 calculates a speed vector, an obstacle position vector, a destination direction vector, and a normal vector (step S4).
[0050] The prediction unit 12 calculates θ, which is the angle formed between the velocity vector and the obstacle position vector (step S5).
[0051] The prediction unit 12 calculates a weight ω based on θ. θ is calculated (step S6).
[0052] The prediction unit 12 calculates φ, which is the angle formed between the destination direction vector and the normal vector (step S7).
[0053] The prediction unit 12 calculates a weight ω based on φ. φ is calculated (step S8).
[0054] The prediction unit 12 calculates the corrected repulsive force vector (f1' in equation (1) and F in equation (2)) (step S9).
[0055] The prediction unit 12 predicts the future position of the pedestrian using the social force model, with the corrected repulsive force vector as the first repulsive force (step S10). Here, steps S2 to S10 correspond to prediction steps.
[0056] The prediction unit 12 outputs a prediction result image showing the predicted trajectory of the pedestrian's movement position (step S11).
[0057] FIG. 5 is a diagram comparing an example of the conventional technology with an example of this embodiment. The elements shown in FIG. 5 are the same as those in FIGS. 2A and 2B. In the example of this embodiment, the repulsive force from the parked vehicle is corrected, particularly for the first repulsive force, and the excessive effect of the first repulsive force seen in the conventional technology can be suppressed. Furthermore, as shown in the prediction result image, in the example of this embodiment, a prediction close to the actual trajectory (reference trajectory) is made.
[0058] As described above, the electronic device 10, program, and information processing method according to this embodiment, by virtue of the above configuration, can suppress excessive repulsive forces that attempt to avoid obstacles in the social force model, thereby improving the accuracy of predicting the location where a pedestrian will move.
[0059] Although the embodiments of the present disclosure have been described based on the drawings and examples, it should be noted that those skilled in the art would easily be able to make various modifications or alterations based on the present disclosure. Therefore, it should be noted that these modifications and alterations are included within the scope of the present disclosure. For example, the functions included in each component or step may be rearranged so as not to cause logical inconsistencies, and multiple components or steps may be combined or divided into one. The embodiments of the present disclosure may also be realized as a storage medium on which a program executed by a processor included in an apparatus is recorded. It should be understood that these modifications and alterations are also included within the scope of the present disclosure. [Explanation of symbols]
[0060] 10 Electronic equipment 11 Acquisition Department 12 Prediction Department 13 Storage section
Claims
1. an acquisition unit that acquires image data of a pedestrian and the surroundings of the pedestrian; a prediction unit that predicts a position where the pedestrian will move using a social force model that calculates a resultant force of an attractive force that moves the pedestrian toward a destination, a first repulsive force that is a repulsive force that the pedestrian tries to avoid an obstacle, and a second repulsive force that is a repulsive force that the pedestrian tries to avoid other pedestrians; The prediction unit Recognizing the pedestrian and the obstacle based on the image data, and calculating a velocity vector calculated from the velocity of the pedestrian and an obstacle position vector directed from the pedestrian to the obstacle; The weight based on θ, which is the angle between the velocity vector and the obstacle position vector, is set as ω θ and the repulsive force vector from the obstacle determined according to the distance between the pedestrian and the obstacle is f １ Let the predetermined weight be ω φ and f is the corrected repulsive force vector calculated by the following formula: １ The electronic device performs the prediction using the first repulsive force as the first repulsive force. [Equation 1]
2. Said ω θ The electronic device according to claim 1 , wherein the value of decreases as the angle θ increases.
3. A predetermined value θ １ and θ ２ is 0 < θ １ ≦θ ２ Fulfilling Said ω θ The value of θ is 0≦θ<θ １ is 1 when the above equation is satisfied, and the above equation is θ ２ 3. The electronic device according to claim 1, wherein the value is 0 if .ltoreq..theta..ltoreq..theta..times ...
4. an acquisition unit that acquires image data of a pedestrian and the surroundings of the pedestrian; a prediction unit that predicts a position where the pedestrian will move using a social force model that calculates a resultant force of an attractive force that moves the pedestrian toward a destination, a first repulsive force that is a repulsive force that the pedestrian tries to avoid an obstacle, and a second repulsive force that is a repulsive force that the pedestrian tries to avoid other pedestrians; The prediction unit Recognizing the pedestrian and the obstacle based on the image data, and calculating a destination direction vector from the pedestrian to the destination and a normal vector determined based on the shape of the obstacle; A weight based on φ, which is the angle between the destination direction vector and the normal vector, is set as ω φ and the repulsive force vector from the obstacle determined according to the distance between the pedestrian and the obstacle is f １ Let the predetermined weight be ω θ and f is the corrected repulsive force vector calculated by the following formula: １ The electronic device performs the prediction using the first repulsive force as the first repulsive force. [Equation 2]
5. The electronic device according to claim 4 , wherein the normal vector is calculated based on a plurality of detection points including a detection point on the obstacle and other detection points present around the detection point.
6. The electronic device according to claim 5 , wherein the normal vector is calculated by principal component analysis using the plurality of detection points.
7. Said ω φ The electronic device according to claim 4 , wherein the value of decreases as the angle φ increases.
8. A predetermined value φ １ and φ ２ is 0 < φ １ ≦φ ２ Fulfilling Said ω φ The value of φ is 0≦φ<φ １ is 1 when ２ The electronic device according to claim 4 , wherein the value is 0 if ≦φ is satisfied.
9. The electronic device according to claim 1 , wherein the prediction unit causes a display unit to display a prediction result image indicating a predicted trajectory of the position of the pedestrian.
10. 5 . The electronic device according to claim 1 , wherein i is an integer equal to or greater than 1, and the corrected repulsive force vector targeting the i obstacles is F, where F is calculated by the following formula: [Equation 3]
11. Computer, an acquisition unit that acquires image data of a pedestrian and the surroundings of the pedestrian; a social force model that calculates a resultant force of an attractive force that moves the pedestrian toward the destination, a first repulsive force that is a repulsive force that the pedestrian tries to avoid an obstacle, and a second repulsive force that is a repulsive force that the pedestrian tries to avoid other pedestrians, and the social force model functions as a prediction unit that predicts the position where the pedestrian will move, The prediction unit Recognizing the pedestrian and the obstacle based on the image data, and calculating a velocity vector calculated from the velocity of the pedestrian and an obstacle position vector directed from the pedestrian to the obstacle; The weight based on θ, which is the angle between the velocity vector and the obstacle position vector, is set as ω θ and the repulsive force vector from the obstacle determined according to the distance between the pedestrian and the obstacle is f １ Let the predetermined weight be ω φ and f is the corrected repulsive force vector calculated by the following formula: １ The program performs the prediction using the first repulsive force as the first repulsive force. [Equation 4]
12. Computer, an acquisition unit that acquires image data of a pedestrian and the surroundings of the pedestrian; a social force model that calculates a resultant force of an attractive force that moves the pedestrian toward the destination, a first repulsive force that is a repulsive force that the pedestrian tries to avoid an obstacle, and a second repulsive force that is a repulsive force that the pedestrian tries to avoid other pedestrians, and the social force model functions as a prediction unit that predicts the position where the pedestrian will move, The prediction unit Recognizing the pedestrian and the obstacle based on the image data, and calculating a destination direction vector from the pedestrian to the destination and a normal vector determined based on the shape of the obstacle; A weight based on φ, which is the angle between the destination direction vector and the normal vector, is set as ω φ and the repulsive force vector from the obstacle determined according to the distance between the pedestrian and the obstacle is f １ Let the predetermined weight be ω θ and f is the corrected repulsive force vector calculated by the following formula: １ The program performs the prediction using the first repulsive force as the first repulsive force. [Equation 5]
13. An information processing method executed by an electronic device, an acquisition step of acquiring image data of a pedestrian and the surroundings of the pedestrian; a prediction step of predicting a position where the pedestrian will move using a social force model that calculates a resultant force of an attractive force that moves the pedestrian toward the destination, a first repulsive force that is a repulsive force that the pedestrian tries to avoid an obstacle, and a second repulsive force that is a repulsive force that the pedestrian tries to avoid other pedestrians; The prediction step Recognizing the pedestrian and the obstacle based on the image data, and calculating a velocity vector calculated from the velocity of the pedestrian and an obstacle position vector directed from the pedestrian to the obstacle; The weight based on θ, which is the angle between the velocity vector and the obstacle position vector, is set as ω θ and the repulsive force vector from the obstacle determined according to the distance between the pedestrian and the obstacle is f １ Let the predetermined weight be ω φ and f is the corrected repulsive force vector calculated by the following formula: １ The information processing method, wherein the prediction is performed using the first repulsive force as the first repulsive force. [Equation 6]
14. An information processing method executed by an electronic device, an acquisition step of acquiring image data of a pedestrian and the surroundings of the pedestrian; a prediction step of predicting a position where the pedestrian will move using a social force model that calculates a resultant force of an attractive force that moves the pedestrian toward the destination, a first repulsive force that is a repulsive force that the pedestrian tries to avoid an obstacle, and a second repulsive force that is a repulsive force that the pedestrian tries to avoid other pedestrians; The prediction step Recognizing the pedestrian and the obstacle based on the image data, and calculating a destination direction vector from the pedestrian to the destination and a normal vector determined based on the shape of the obstacle; A weight based on φ, which is the angle between the destination direction vector and the normal vector, is set as ω φ and the repulsive force vector from the obstacle determined according to the distance between the pedestrian and the obstacle is f １ Let the predetermined weight be ω θ and f is the corrected repulsive force vector calculated by the following formula: １ The information processing method, wherein the prediction is performed using the first repulsive force as the first repulsive force. [Equation 7]
Citation Information
Patent Citations
Mobile object predictor
JP2018124663A