A simulation method and system applied to pedestrian travel avoidance
By introducing critical path point navigation and dynamic obstacle avoidance mechanisms into the traditional social force model, identifying nearby obstacles and constructing an active obstacle avoidance zone, and calculating active avoidance forces, the obstacle avoidance problem of the traditional model in complex environments is solved, and the rationality and realism of pedestrian simulation are improved.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- SHENZHEN URBAN TRANSPORT PLANNING CENT CO LTD
- Filing Date
- 2026-03-20
- Publication Date
- 2026-06-19
AI Technical Summary
Traditional social force models are passive in obstacle avoidance in complex geometric scenarios, prone to global and local navigation conflicts, and have poor adaptability to special geometric scenarios such as concave obstacles, which makes pedestrians prone to oscillation, stagnation, or getting trapped in concave traps during pedestrian simulation.
By combining critical path point navigation with a dynamic active obstacle avoidance mechanism, the active obstacle avoidance area is constructed by identifying nearby obstacle points and calculating active avoidance forces. The total force of the pedestrian is synthesized to update the motion state and guide the pedestrian to avoid obstacles in advance.
It enables precise obstacle avoidance for pedestrians in complex environments, preventing tremors and stagnation, and improving the rationality and realism of pedestrian simulation.
Smart Images
Figure CN122242187A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of pedestrian simulation technology, specifically to a simulation method and system for pedestrian movement and avoidance. Background Technology
[0002] The social force model, proposed by Helbing et al., is a multi-particle self-driven model, where pedestrians are represented by self-driven particles. The concept of social force does not refer to physically existing forces, but rather represents the social psychology of pedestrians and the interactions between pedestrians and between pedestrians and the environment in a virtual way. A pedestrian of mass *m* always desires to move in a desired direction at a desired speed; therefore, during walking, pedestrians expect to adjust their current actual speed to the desired speed in a short period. The interactions between pedestrians and walls, and between pedestrians and other pedestrians, include repulsion and friction. The closer the distance, the greater the interaction force. In the social force model, *m* and *m* are often used to represent the interaction forces between pedestrians and between pedestrians and walls, respectively. However, traditional social force models suffer from passive obstacle avoidance and susceptibility to local optima in complex structured environments.
[0003] In scenarios with complex geometries such as concave obstacles and narrow passages, traditional social force models have significant shortcomings: their obstacle avoidance is a reactive short-range repulsive force that only takes effect when pedestrians approach the obstacle, making it difficult to simulate human forward-looking obstacle avoidance behavior; global navigation based on waypoints and local repulsive force obstacle avoidance are prone to conflict, causing pedestrians to oscillate at the obstacle entrance or be pushed into the concave area; at the same time, they are poorly adaptable to special geometries such as U-shaped and L-shaped structures, and symmetrical repulsive force fields are prone to causing congestion at the entrance and pedestrians to be stuck inside. Summary of the Invention
[0004] To address the technical problems of traditional social force models, such as passive obstacle avoidance in complex geometric scenarios, susceptibility to global and local navigation conflicts, and poor adaptability to special geometric scenarios like concave obstacles, leading to pedestrian oscillations, stagnation, or getting trapped in concave obstacles during simulation, this invention provides a simulation method for pedestrian movement and obstacle avoidance, comprising:
[0005] S1. The pedestrian navigates according to a predefined sequence of key path points, determines the current target path point and the next target path point, and determines the target driving force for the pedestrian to move towards the next target path point.
[0006] S2. Based on the next target path point, identify nearby obstacle points that pose a direct threat to the pedestrian's progress;
[0007] S3. Taking the nearby obstacle point as the starting point and the direction pointing to the next target path point as the reference direction, construct an active obstacle avoidance area next to the path close to the current side of the pedestrian.
[0008] S4. When a pedestrian enters the active obstacle avoidance area, the active avoidance force acting on the pedestrian is calculated, and the direction of the active avoidance force is configured to guide the pedestrian to avoid the nearby obstacle points in advance.
[0009] S5. Combine the active avoidance force with the target driving force, the force between pedestrians and the force between pedestrians and obstacles to obtain the total force of the pedestrian, so as to update the pedestrian's motion state.
[0010] Furthermore, in S1, the sequence of key path points is as follows:
[0011]
[0012] Where W is the set of critical path points for pedestrian movement. These are the 1st, 2nd to nth critical path points arranged in the order of pedestrian movement, where n is the total number of critical path points;
[0013] The current target path point is The next target path point is ;
[0014] The target driving force is:
[0015]
[0016] Among them, among them, For pedestrians The driving force of the goal For pedestrians The size parameters, For a moment point to pedestrians The composite vector of desired velocity and desired direction. For pedestrians At any moment The current actual speed, Adjust the time parameters for pedestrian speed.
[0017] Furthermore, in S2, the nearby obstacle point is:
[0018]
[0019] in, Let 'd' be the nearest obstacle point, arg min be the parameter minimum function, and d be the distance function. For the vertices of the obstacle, For the set of obstacle vertices, For pedestrians Current location Indicates from pedestrians Current location Point to the next target path point The line segment.
[0020] Furthermore, in S3, the unit reference vector for the reference direction is:
[0021]
[0022] Among them, among them, As a unit reference vector, The Euclidean distance between the next target path point and the adjacent obstacle point is given.
[0023] Furthermore, in S4, the active avoidance force is:
[0024]
[0025] in, In order to actively avoid forces, The force intensity coefficient, The vertical distance between the pedestrian and the virtual wall. The force field attenuation coefficient, Let θ be the direction sensitivity factor, and θ be the angle between the pedestrian's velocity direction and n. This is the unit normal vector that is far away from the obstacle.
[0026] Furthermore, in S5, the total force exerted by the pedestrian is:
[0027]
[0028] in, For pedestrians The total force at time t, The target driving force at time t, For other pedestrians at time t For pedestrians The sum of the interaction forces, Let t be the time of obstacle i for pedestrian The sum of the repulsive forces, Let be the active avoidance force at time t.
[0029] It also includes a simulation system for pedestrian avoidance, comprising:
[0030] The waypoint navigation module is used for pedestrians to navigate according to a predefined sequence of key waypoints, determine the current target waypoint and the next target waypoint, and determine the target driving force for the pedestrian to move towards the next target waypoint.
[0031] The obstacle identification module is used to identify nearby obstacle points that pose a direct threat to the pedestrian's progress based on the next target path point;
[0032] The obstacle avoidance area construction module is used to construct an active obstacle avoidance area next to the path close to the current side of the pedestrian, with the nearby obstacle point as the starting point and the direction pointing to the next target path point as the reference direction.
[0033] The obstacle avoidance force calculation module is used to calculate the active obstacle avoidance force acting on the pedestrian when the pedestrian enters the active obstacle avoidance area. The direction of the active obstacle avoidance force is configured to guide the pedestrian to avoid the nearby obstacle points in advance.
[0034] The total force synthesis drive module is used to synthesize the active avoidance force with the target driving force, the force between pedestrians and the force between pedestrians and obstacles to obtain the total force of the pedestrian, so as to update the pedestrian's motion state.
[0035] The beneficial effects of this invention are:
[0036] This invention improves the traditional social force model by integrating critical path point navigation and dynamic active obstacle avoidance mechanism. It achieves accurate identification of nearby obstacle points, directional construction of active obstacle avoidance area, and calculation of active avoidance force with distance attenuation and direction sensitivity characteristics. The active avoidance force is integrated with the target driving force, the interaction force between pedestrians, and the repulsive force between pedestrians and obstacles into the total force of pedestrians. This effectively solves the problems of passive obstacle avoidance, conflict between global and local navigation, and poor adaptability to special geometries in traditional models in complex geometric scenarios. Attached Figure Description
[0037] Figure 1 This is a flowchart of a simulation method for pedestrian avoidance.
[0038] Figure 2 A schematic diagram illustrating a scenario where a pedestrian becomes trapped and stranded by a concave obstacle, as shown in a traditional social force model.
[0039] Figure 3 This is a schematic diagram illustrating the nearby obstacle point recognition and active obstacle avoidance area construction of the present invention;
[0040] Figure 4 This is a schematic diagram illustrating a scenario where the present invention actively guides pedestrians away from concave obstacles. Detailed Implementation
[0041] The technical solution of the present invention will be further described below with reference to embodiments, but it is not limited thereto. Any modifications or equivalent substitutions to the technical solution of the present invention without departing from the spirit and scope of the technical solution of the present invention should be covered within the protection scope of the present invention. In the following embodiments, process equipment or devices not specifically specified are all conventional equipment or devices in the art. Unless specifically specified, the technical means used in the embodiments of the present invention are all conventional means well known to those skilled in the art.
[0042] Example 1, combined with Figure 1 This embodiment describes a simulation method for pedestrian avoidance, comprising:
[0043] S1. The pedestrian navigates according to a predefined sequence of key path points, determines the current target path point and the next target path point, and determines the target driving force for the pedestrian to move towards the next target path point.
[0044] S2. Based on the next target path point, identify nearby obstacle points that pose a direct threat to the pedestrian's progress;
[0045] S3. Taking the nearby obstacle point as the starting point and the direction pointing to the next target path point as the reference direction, construct an active obstacle avoidance area next to the path close to the current side of the pedestrian.
[0046] S4. When a pedestrian enters the active obstacle avoidance area, the active avoidance force acting on the pedestrian is calculated, and the direction of the active avoidance force is configured to guide the pedestrian to avoid the nearby obstacle points in advance.
[0047] S5. Combine the active avoidance force with the target driving force, the force between pedestrians and the force between pedestrians and obstacles to obtain the total force of the pedestrian, so as to update the pedestrian's motion state.
[0048] Specifically, such as Figure 2 As shown, under the traditional social force model, when a pedestrian starts from point P0 and travels through P1 and P2 towards point P3, the presence of concave obstacles can easily cause pedestrians to fall into local optima, resulting in crowding and stagnation in the obstacle area, making it difficult to escape the trap effectively. This embodiment addresses this problem by proposing a simulation method for pedestrian obstacle avoidance. This invention introduces critical path point navigation to determine the pedestrian's target direction of travel, accurately identifies nearby obstacle points that pose a direct threat to the pedestrian's movement, and constructs an active obstacle avoidance zone along the path. When a pedestrian enters this zone, a directional active avoidance force is calculated and applied. This force is then combined with the target driving force, inter-pedestrian forces, and obstacle repulsion to obtain the total force, thereby updating the pedestrian's motion state and guiding them to avoid obstacles in advance. This prevents oscillations, stagnation, or falling into concave traps, improving the rationality and realism of pedestrian simulation in complex environments.
[0049] Furthermore, in S1, the sequence of key path points is as follows:
[0050]
[0051] Where W is the set of critical path points for pedestrian movement. These are the 1st, 2nd to nth critical path points arranged in the order of pedestrian movement, where n is the total number of critical path points;
[0052] The current target path point is The next target path point is ;
[0053] The target driving force is:
[0054]
[0055] Among them, among them, For pedestrians The driving force of the goal For pedestrians The size parameters, For a moment point to pedestrians The composite vector of desired velocity and desired direction. For pedestrians At any moment The current actual speed, Adjust the time parameters for pedestrian speed.
[0056] Specifically, this step clarifies the pedestrian's global travel route by predefining a sequence of critical path points, determines the current target path point and the next target path point, and calculates the target driving force based on a formula. Its core principle is to transform the pedestrian's global navigation intention into a quantifiable driving vector, providing a clear directional reference for subsequent local obstacle avoidance decisions.
[0057] Furthermore, in S2, the nearby obstacle point is:
[0058]
[0059] in, Let 'd' be the nearest obstacle point, arg min be the parameter minimum function, and d be the distance function. For the vertices of the obstacle, For the set of obstacle vertices, For pedestrians Current location Indicates from pedestrians Current location Point to the next target path point The line segment.
[0060] Furthermore, in S3, the unit reference vector for the reference direction is:
[0061]
[0062] Among them, among them, As a unit reference vector, The Euclidean distance between the next target path point and the adjacent obstacle point is given.
[0063] Specifically, such as Figure 3 As shown, when the pedestrian is located at the current target path point P2 (i.e., the current target path point is...), The next target path point is P3 (i.e., the next target path point is...). When this method is used, it identifies nearby obstacle points by calculating nearby obstacle points, and uses these points as the starting point to point towards... Using the direction as a reference, the active obstacle avoidance area shown in the figure is constructed, providing a spatial reference for the subsequent application of active avoidance forces.
[0064] The construction of the active obstacle avoidance zone consists of two steps: First, starting from the nearest obstacle point, a unit reference vector is established pointing towards the next target path point, and a virtual wall of length Lwall is generated along this direction; Second, on the side of the virtual wall closest to the pedestrian's current path point, a rectangular obstacle avoidance zone is constructed. The centerline of this zone is parallel to or coincides with the virtual wall. The width is used to control the lateral range of the obstacle avoidance zone, and the height is equal to the length of the virtual wall. At the same time, a normal n perpendicular to the unit reference vector and pointing towards the safe side is defined. This zone serves as the domain of the active avoidance force and can be regarded as a guiding force field of "soft constraint".
[0065] Furthermore, in S4, the active avoidance force is:
[0066]
[0067] in, In order to actively avoid forces, The force intensity coefficient, The vertical distance between the pedestrian and the virtual wall. The force field attenuation coefficient, Let θ be the direction sensitivity factor, and θ be the angle between the pedestrian's velocity direction and n. This is the unit normal vector that is far away from the obstacle.
[0068] Furthermore, in S5, the total force exerted by the pedestrian is:
[0069]
[0070] in, For pedestrians The total force at time t, The target driving force at time t, For other pedestrians at time t For pedestrians The sum of the interaction forces, Let t be the time of obstacle i for pedestrian The sum of the repulsive forces, Let be the active avoidance force at time t.
[0071] Specifically, such as Figure 4 As shown, when a pedestrian enters the active obstacle avoidance area, this method calculates the active avoidance force and combines it with the target driving force, the interaction force between pedestrians and the repulsive force of the obstacle to obtain the total force, which guides the pedestrian to leave the concave obstacle area along a reasonable path, thus avoiding the stagnation and oscillation problems that occur in traditional models.
[0072] It also includes a simulation system for pedestrian avoidance, comprising:
[0073] The waypoint navigation module is used for pedestrians to navigate according to a predefined sequence of key waypoints, determine the current target waypoint and the next target waypoint, and determine the target driving force for the pedestrian to move towards the next target waypoint.
[0074] The obstacle identification module is used to identify nearby obstacle points that pose a direct threat to the pedestrian's progress based on the next target path point;
[0075] The obstacle avoidance area construction module is used to construct an active obstacle avoidance area next to the path close to the current side of the pedestrian, with the nearby obstacle point as the starting point and the direction pointing to the next target path point as the reference direction.
[0076] The obstacle avoidance force calculation module is used to calculate the active obstacle avoidance force acting on the pedestrian when the pedestrian enters the active obstacle avoidance area. The direction of the active obstacle avoidance force is configured to guide the pedestrian to avoid the nearby obstacle points in advance.
[0077] The total force synthesis drive module is used to synthesize the active avoidance force with the target driving force, the force between pedestrians and the force between pedestrians and obstacles to obtain the total force of the pedestrian, so as to update the pedestrian's motion state.
Claims
1. A simulation method for pedestrian avoidance, characterized in that, include: S1. The pedestrian navigates according to a predefined sequence of key path points, determines the current target path point and the next target path point, and determines the target driving force for the pedestrian to move towards the next target path point. S2. Based on the next target path point, identify nearby obstacle points that pose a direct threat to the pedestrian's progress; S3. Taking the nearby obstacle point as the starting point and the direction pointing to the next target path point as the reference direction, construct an active obstacle avoidance area next to the path close to the current side of the pedestrian. S4. When a pedestrian enters the active obstacle avoidance area, the active avoidance force acting on the pedestrian is calculated, and the direction of the active avoidance force is configured to guide the pedestrian to avoid the nearby obstacle points in advance. S5. Combine the active avoidance force with the target driving force, the force between pedestrians and the force between pedestrians and obstacles to obtain the total force of the pedestrian, so as to update the pedestrian's motion state.
2. The simulation method for pedestrian avoidance according to claim 1, characterized in that, In S1, the sequence of key path points is as follows: Where W is the set of critical path points for pedestrian movement. These are the 1st, 2nd to nth critical path points arranged in the order of pedestrian movement, where n is the total number of critical path points; The current target path point is The next target path point is ; The target driving force is: Among them, among them, For pedestrians The driving force of the goal For pedestrians The size parameters, For a moment point to pedestrians The composite vector of desired velocity and desired direction. For pedestrians At any moment The current actual speed, Adjust the time parameters for pedestrian speed.
3. The simulation method for pedestrian avoidance according to claim 1, characterized in that, In S2, the nearby obstacle points are: in, Let 'd' be the nearest obstacle point, arg min be the parameter minimum function, and d be the distance function. For the vertices of the obstacle, For the set of obstacle vertices, For pedestrians Current location Indicates from pedestrians Current location Point to the next target path point The line segment.
4. The simulation method for pedestrian avoidance according to claim 1, characterized in that, In S3, the unit reference vector for the reference direction is: Among them, among them, As a unit reference vector, The Euclidean distance between the next target path point and the adjacent obstacle point is given.
5. The simulation method for pedestrian avoidance according to claim 1, characterized in that, In S4, the active avoidance force is: in, In order to actively avoid forces, The force intensity coefficient, The vertical distance between the pedestrian and the virtual wall. The force field attenuation coefficient, Let θ be the direction sensitivity factor, and θ be the angle between the pedestrian's velocity direction and n. This is the unit normal vector that is far away from the obstacle.
6. The simulation method for pedestrian avoidance according to claim 1, characterized in that, In S5, the total force exerted by the pedestrian is: in, For pedestrians The total force at time t, The target driving force at time t, For other pedestrians at time t For pedestrians The sum of the interaction forces, Let t be the time of obstacle i for pedestrian The sum of the repulsive forces, Let be the active avoidance force at time t.
7. A simulation system for pedestrian avoidance, characterized in that, include: The waypoint navigation module is used for pedestrians to navigate according to a predefined sequence of key waypoints, determine the current target waypoint and the next target waypoint, and determine the target driving force for the pedestrian to move towards the next target waypoint. The obstacle identification module is used to identify nearby obstacle points that pose a direct threat to the pedestrian's progress based on the next target path point; The obstacle avoidance area construction module is used to construct an active obstacle avoidance area next to the path close to the current side of the pedestrian, with the nearby obstacle point as the starting point and the direction pointing to the next target path point as the reference direction. The obstacle avoidance force calculation module is used to calculate the active obstacle avoidance force acting on the pedestrian when the pedestrian enters the active obstacle avoidance area. The direction of the active obstacle avoidance force is configured to guide the pedestrian to avoid the nearby obstacle points in advance. The total force synthesis drive module is used to synthesize the active avoidance force with the target driving force, the force between pedestrians and the force between pedestrians and obstacles to obtain the total force of the pedestrian, so as to update the pedestrian's motion state.