Path navigation methods, devices and media for mobile robots in densely populated environments

Abstract

This invention discloses a path navigation method, device, and medium for mobile robots in densely populated environments. The method includes identifying objects in the densely populated environment and constructing semantic map information based on these objects; performing semantic information analysis on the objects and using a unified interaction space method to describe the human-machine-object relationship to construct an interaction space model; constructing a sampling path tree by grafting several sampling paths to provide sampling paths with a reasonable density based on the needs of the densely populated environment; using different sampling path time allocations based on different navigation modes to simulate different pedestrian navigation behaviors; performing collision detection and comfort assessment on the sampling paths; setting evaluation indicators and comprehensively evaluating and filtering the sampling paths based on these indicators to obtain the sampling path corresponding to the best evaluation indicator as the optimal navigation path. This invention predicts pedestrian movement trajectories in real time and quickly updates path planning, ensuring safe navigation of the mobile robot in dynamic environments.

Description

Technical Field

[0001] This invention relates to the field of robot application technology, and in particular to a path navigation method, device and medium for mobile robots in densely populated environments. Background Technology

[0002] With the rapid development of robotics technology, mobile robots are being used more and more widely in various scenarios, such as shopping malls, airports, train stations, hospitals, and other public places. These scenarios are often characterized by dense crowds, requiring mobile robots to navigate autonomously in complex environments to complete tasks such as goods transportation, service guidance, and environmental monitoring. However, densely populated environments bring many challenges to the path planning of mobile robots.

[0003] In traditional path planning methods, mobile robots typically treat obstacles in the environment as static, using static maps for path planning. This approach limits the adaptability of mobile robots, especially in dynamic environments. These methods often neglect social order among pedestrians, such as following, polite stopping, and yielding. In dynamic, dense crowds, pedestrians act as dynamic obstacles, and their trajectories are difficult to predict, making traditional path planning methods ineffective. Furthermore, most existing path planning methods focus only on the feasibility and safety of the path, lacking the quantification and consideration of social norms. This may result in mobile robots, while avoiding collisions, behaving in ways that do not meet human expectations, impacting the user experience. Moreover, most existing path planning methods are only applicable to a single navigation mode, lacking support for multiple navigation modes, which limits the adaptability and flexibility of mobile robots in complex environments. Summary of the Invention

[0004] This invention provides a path navigation method, device, and medium for mobile robots in densely populated environments, aiming to solve at least one of the technical problems existing in the prior art.

[0005] The technical solution of this invention is a path navigation method for a mobile robot in a densely populated environment, comprising:

[0006] Identify objects in densely populated environments and construct semantic map information based on these objects;

[0007] Semantic information analysis is performed on the objects, and a unified interaction space method is used to describe the human-machine-object relationship in order to construct an interaction space model;

[0008] A sampling path tree is constructed by grafting several sampling paths, and sampling paths of reasonable density are provided based on the needs of the dense population environment.

[0009] Different sampling path time allocations are used based on different navigation modes to simulate different pedestrian navigation behaviors;

[0010] Collision detection and comfort assessment are performed on the sampling path;

[0011] Set evaluation indicators, and use these indicators to comprehensively evaluate and filter sampling paths, obtaining the sampling path corresponding to the best evaluation indicator as the best navigation path.

[0012] According to some embodiments of the present invention, the navigation mode includes a follow mode, and the use of different sampling path time allocations based on different navigation modes to simulate different pedestrian navigation behaviors includes:

[0013] In follow mode, the mobile robot is configured to allocate sampling path time according to a trapezoidal speed, and the upper limit of the speed allocation is set to dynamically change according to the pedestrian's speed. The time allocation formula is as follows:

[0014]

[0015] in, This represents the timestamp after the trapezoidal velocity allocation. Indicates the distance from the starting point to the path point. The length at that point, This indicates the set acceleration. Indicates the initial velocity. Indicates reference speed;

[0016] After completing the time allocation for each sampling path in follow mode, the first time allocation map on the sampling path is obtained.

[0017] According to some embodiments of the present invention, the path navigation method for a mobile robot in a densely populated environment further includes:

[0018] In the following mode, the mobile robot identifies pedestrians with similar directions of movement within a crowd as leader pedestrians, provided that the leader pedestrians meet the following conditions:

[0019]

[0020] in, Indicates the direction of movement of the leader pedestrian. Indicates the target direction of movement for the mobile robot. Indicates the current direction of movement of the mobile robot. This represents the angle between the direction of movement of the leader pedestrian and the target direction of movement of the mobile robot. The angle between the current direction of motion of the mobile robot and the target direction of motion of the mobile robot is represented by the symbol. Indicates the calculation of the angle between vectors;

[0021] The angle between the leader pedestrian's movement direction and the target movement direction of the mobile robot, and the angle between the mobile robot's current movement direction and the mobile robot's target movement direction, respectively satisfy the following conditions:

[0022]

[0023] in, This represents the angle between the direction of movement of the leader pedestrian and the target direction of movement of the mobile robot. This represents the maximum angle between the direction of movement of the leader pedestrian and the target direction of movement of the mobile robot. This represents the angle between the current direction of motion of the mobile robot and the target direction of motion of the mobile robot. This represents the maximum angle between the current direction of motion of the mobile robot and the target direction of motion of the mobile robot.

[0024] According to some embodiments of the present invention, the navigation mode further includes a solo mode and a stopped mode, and the use of different sampling path time allocations based on different navigation modes to simulate different pedestrian navigation behaviors includes:

[0025] In solo mode, the mobile robot is configured to allocate sampling path time according to trapezoidal velocity, and the maximum navigation speed for the trapezoidal velocity allocation is set. The time allocation formula is as follows:

[0026]

[0027] in, This represents the timestamp after the trapezoidal velocity allocation. Indicates the distance from the starting point to the path point. The length at that point, Indicates the maximum acceleration. This indicates the set acceleration. Indicates the initial velocity. Indicates maximum speed;

[0028] After completing the time allocation for each sampling path in the solo mode, a second time allocation map on the sampling path is obtained;

[0029] In stop mode, the mobile robot is configured to allocate sampling path time according to a trapezoidal velocity, and a deceleration requirement is set for the allocation. The time allocation formula is as follows:

[0030]

[0031] in, This represents the timestamp after the trapezoidal velocity allocation. Indicates the distance from the starting point to the path point. The length at that point, Indicates the maximum acceleration. This indicates the set acceleration. Indicates the initial velocity. Indicates the maximum speed. Indicates invalid;

[0032] After completing the time allocation for each sampling path in stop mode, a third time allocation map on the sampling path is obtained.

[0033] According to some embodiments of the present invention, the step of performing semantic information analysis on the things and using a unified interaction space method to describe human-machine-thing to construct an interaction space model includes:

[0034] Based on sensor-based object recognition to complete semantic information analysis, a unified interaction space method is used to describe human-machine-object interactions, including interactions between pedestrians, between robots and pedestrians, and between pedestrians and objects, in order to construct an interaction space model. The methods for constructing the interaction space model include:

[0035] A pedestrian interaction space model is constructed, and the formula for the pedestrian interaction space model is as follows:

[0036]

[0037] in, Represents pedestrian interaction space, Represents any point in space. Indicates time, Indicates the weighting coefficient. It is an asymmetric Gaussian distribution function, representing the interaction space range of pedestrians or objects. The specific formula for the interaction space range is described as follows:

[0038]

[0039] in, Represents any point in space. It represents the pose information of pedestrians or objects. Indicates parameters, Represents the backward covariance matrix. Represents the forward covariance matrix;

[0040] A transformation operation is performed on the asymmetric Gaussian distribution function to obtain the transformation formula for the asymmetric Gaussian distribution function, which is as follows:

[0041]

[0042]

[0043] in, Represents any point in space. It represents the pose information of pedestrians or objects. Indicates parameters, The rotation matrix representing the pedestrian. Indicates the location of the pedestrian. , Indicates the pedestrian's position and posture. Represents the backward covariance matrix. Denotes the forward covariance matrix. This represents the variance value. , , These are parameters used to describe pedestrian space. This represents the transverse covariance of an asymmetric Gaussian function. This represents the forward covariance of an asymmetric Gaussian function. This represents the backward covariance of an asymmetric Gaussian function. This indicates the normalized orientation angle.

[0044] According to some embodiments of the present invention, the path navigation method for a mobile robot in a densely populated environment further includes:

[0045] The weighting coefficients are set to achieve numerical normalization of the pedestrian interaction space, thereby obtaining the numerical distribution of the pedestrian interaction space. The weighting coefficients are calculated as follows:

[0046]

[0047]

[0048] in, i=j This represents the pedestrian's own interaction space, at this time =1, therefore, the expression for the numerical distribution of the pedestrian interaction space is as follows:

[0049]

[0050] The paths with numerical distributions in the pedestrian interaction space within [0.6, 1) are classified as non-sampled paths, the paths with numerical distributions in the pedestrian interaction space within [0.4, 0.6) are classified as undetermined sampling paths, and the paths with numerical distributions in the pedestrian interaction space within [0, 0.4) are classified as target sampling paths.

[0051] According to some embodiments of the present invention, the collision detection and comfort assessment of the sampling path includes:

[0052] An evaluation model is established for collision detection and comfort assessment of the sampling path. The expression of the evaluation model is as follows:

[0053]

[0054] in, Indicates the evaluation model, Indicates the sampling path, Indicates sampling path The point on, Indicates time, Indicates the pose point The values ​​of the numerical distribution in the pedestrian interaction space. This represents the time difference between the timestamp of a waypoint and the predicted time of a pedestrian's arrival at that waypoint. Indicates the time threshold;

[0055] The expression for the time threshold is:

[0056]

[0057] in, Indicates the time threshold. The diameter of the circle representing the pedestrian. Indicates the pedestrian's current speed. This represents the coefficient of thermal expansion, which is usually greater than 1.0;

[0058] When the mobile robot enters the pedestrian interaction space, the closer the mobile robot gets to the pedestrian's sampling path... The stronger the interference of the mobile robot with pedestrians, the greater the value of the interference, which is taken as the comfort level of the sampling path. The expression for the comfort level of the sampling path is:

[0059]

[0060] in, Indicates the comfort level of the sampling path. This represents the interference value of the mobile robot on the pedestrian sampling path. Indicates the sampling path, Indicates time;

[0061] Collision detection is performed by obtaining the numerical distribution of the pedestrian interaction space. Paths with numerical distribution in [0.6, 1) and paths with numerical distribution in [0, 0.4) are divided into collision-prone detection regions.

[0062] According to some embodiments of the present invention, the step of setting evaluation indicators and comprehensively evaluating and screening sampling paths through the evaluation indicators to obtain the sampling path corresponding to the best evaluation indicator as the best navigation path includes:

[0063] Efficiency is set as the core indicator for mobile robot sampling path navigation. The formulas for efficiency evaluation indicators under different motion mode conditions are as follows:

[0064]

[0065] in, Indicates efficiency indicators. This indicates path length calculation. This represents the vector between the end of the sampling path and the local target point. This represents the vector from the starting point to the local target point. This indicates the maximum speed of the mobile robot. Indicates the pedestrian's current speed. The diameter of the circle representing the pedestrian. This represents the diameter of the circle representing the mobile robot. This indicates the angle between the direction of movement of the leader pedestrian and the target direction of movement of the mobile robot;

[0066] The social norm indicators are quantified and applied to mobile robots. The formula for these social norm indicators is as follows:

[0067]

[0068] in, Indicators representing social norms and All are constants and satisfy the following conditions: ;

[0069] A consistency metric is set to measure the degree of oscillation in the planned sampling path of the mobile robot at different time points. The formula for the consistency metric is as follows:

[0070]

[0071] in, Indicators of consistency This represents the distance between two sampling paths. The smaller the value, the higher the consistency between the two sampling paths;

[0072] A smoothness metric is set up to measure the smoothness of the sampling path planned by the mobile robot. The formula for the smoothness metric is as follows:

[0073]

[0074] in, Indicates smoothness index, This represents the curvature calculation function. Indicates the sampling path;

[0075] Based on the efficiency index, social norm index, consistency index, and smoothness index, an evaluation equation for a multi-dimensional index is obtained. Multiple sampling paths are then comprehensively evaluated and screened according to this evaluation equation to obtain the sampling path corresponding to the optimal evaluation index as the best navigation path. The formula for the evaluation equation of the multi-dimensional index is as follows:

[0076]

[0077] in, Indicates the best navigation path. This represents the weighting coefficient of each evaluation indicator. Indicates the evaluation indicators.

[0078] The present invention also relates to a computer device, including a memory and a processor, wherein the processor performs the above-described method when executing a computer program stored in the memory.

[0079] The present invention also relates to a computer-readable storage medium storing computer program instructions thereon, which, when executed by a processor, implement the above-described method.

[0080] The path navigation method, device, and medium for mobile robots in densely populated environments provided in this invention have at least one of the following advantages or beneficial effects: The path navigation method for mobile robots in densely populated environments provides an interactive space model to describe complex human interaction relationships, preventing mobile robots from intruding into pedestrians' personal interaction spaces. Using a universal set of sampled paths, only the corresponding time map needs to be updated to represent different motion modes using the same sampled paths. Simultaneously, by updating the time map under different modes, the same path set can represent different navigation behaviors, and efficient dynamic obstacle collision detection can be achieved. Furthermore, a series of evaluation indicators are set and mapped to the sampled path score, making mobile robot navigation more in line with human social norms, effectively solving the problems of safe, efficient, and comfortable navigation for mobile robots in dynamic environments, and enabling the movement of mobile robots to deeply integrate into human society.

[0081] Furthermore, additional aspects and advantages of the invention will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention. Attached Figure Description

[0082] Figure 1 This is a general flowchart of the path navigation method for mobile robots in densely populated environments provided in the embodiments of the present invention;

[0083] Figure 2This is a graph of sampling path generation implemented by the grafting strategy based on sampling trees provided in this embodiment of the invention;

[0084] Figure 3 This is a time planning diagram of the trapezoidal velocity in the following mode provided in the embodiments of the present invention;

[0085] Figure 4 This is the first time allocation diagram on the sampling path in the follow mode provided in the embodiment of the present invention;

[0086] Figure 5 This is the first time allocation map on the sampling tree in the follow mode provided in the embodiments of the present invention;

[0087] Figure 6 This is a first schematic diagram of a leader's pedestrian path provided in an embodiment of the present invention;

[0088] Figure 7 This is a second schematic diagram of the leader's pedestrian path provided in an embodiment of the present invention;

[0089] Figure 8 This is a schematic diagram of the path provided by an embodiment of the present invention for a mobile robot following a leader pedestrian;

[0090] Figure 9 This is a time planning diagram of the trapezoidal velocity in the solo mode provided in the embodiments of the present invention;

[0091] Figure 10 is a time planning diagram of the trapezoidal velocity in the stopping mode provided by an embodiment of the present invention;

[0092] Figure 11 is a third time allocation diagram on the sampling path in the stop mode provided by an embodiment of the present invention;

[0093] Figure 12 This is the third time allocation diagram on the sampling tree in the stop mode provided in the embodiments of the present invention;

[0094] Figure 13 This is a schematic diagram of the pedestrian interaction space based on the unified interaction space model provided in the embodiments of the present invention;

[0095] Figure 14 is an interactive space diagram composed of multiple pedestrians provided in an embodiment of the present invention;

[0096] Figure 15 is an interactive space diagram of a single pedestrian and object provided in an embodiment of the present invention;

[0097] Figure 16 This is the semantically segmented environment map provided in this embodiment of the invention;

[0098] Figure 17 This is a simulation experiment diagram of the first optimal navigation path provided in the embodiments of the present invention;

[0099] Figure 18 This is a simulation experiment diagram of the second optimal navigation path provided in the embodiments of the present invention. Detailed Implementation

[0100] The following will provide a clear and complete description of the concept, specific structure, and technical effects of the present invention in conjunction with the embodiments and accompanying drawings, so as to fully understand the purpose, solution, and effects of the present invention.

[0101] It should be noted that, unless otherwise specified, when a feature is referred to as "fixed" or "connected" to another feature, it can be directly fixed or connected to the other feature, or indirectly fixed or connected to the other feature. The singular forms "a," "described," and "the" used herein are also intended to include the plural forms, unless the context clearly indicates otherwise. Furthermore, unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art. The terminology used in this specification is for the purpose of describing particular embodiments only and not for limiting the invention. The term "and / or" as used herein includes any combination of one or more of the associated listed items.

[0102] It should be understood that although the terms first, second, third, etc., may be used to describe various elements in this invention, these elements should not be limited to these terms. These terms are only used to distinguish elements of the same type from each other. For example, a first element may also be referred to as a second element without departing from the scope of the invention, and similarly, a second element may also be referred to as a first element. Any and all instances or exemplary language (“e.g.,” “such as,” etc.) provided herein are intended only to better illustrate embodiments of the invention and, unless otherwise required, do not impose a limitation on the scope of the invention.

[0103] In traditional path planning methods, mobile robots typically treat obstacles in the environment as static, using static maps for path planning. This approach limits the adaptability of mobile robots, especially in dynamic environments. These methods often neglect social order among pedestrians, such as following, polite stopping, and yielding. In dynamic, dense crowds, pedestrians act as dynamic obstacles, and their trajectories are difficult to predict, making traditional path planning methods ineffective. Furthermore, most existing path planning methods focus only on the feasibility and safety of the path, lacking the quantification and consideration of social norms. This may result in mobile robots, while avoiding collisions, behaving in ways that do not meet human expectations, impacting the user experience. Moreover, most existing path planning methods are only applicable to a single navigation mode, lacking support for multiple navigation modes, which limits the adaptability and flexibility of mobile robots in complex environments.

[0104] Based on this, embodiments of the present invention provide a path navigation method, device, and medium for mobile robots in densely populated environments. The mobile robot can perceive and predict the movement trajectory of pedestrians in real time and quickly update the path planning, ensuring safe navigation of the mobile robot in dynamic environments. Social norms are quantified and integrated into the path planning, making the behavior of the mobile robot conform to human expectations and improving the user experience.

[0105] Please refer to the following. Figures 1 to 18 The path navigation method, device, and medium for mobile robots in densely populated environments provided in the embodiments of this application are further described.

[0106] Reference Figure 1 As shown, Figure 1 This is a general flowchart of a path navigation method for a mobile robot in a dense crowd environment provided by an embodiment of the present invention. The path navigation method for a mobile robot in a dense crowd environment includes, but is not limited to, steps S100 to S600. Specifically,

[0107] S100: Identify objects in densely populated environments and construct semantic map information based on these objects;

[0108] S200: Perform semantic information analysis on things and use the unified interaction space method to describe human-machine-thing in order to construct an interaction space model;

[0109] S300: Constructs a sampling path tree by grafting several sampling paths, and provides a reasonable density of sampling paths based on the needs of densely populated environments;

[0110] S400: Different sampling path time allocations are used based on different navigation modes to simulate different pedestrian navigation behaviors;

[0111] S500: Performs collision detection and comfort assessment on the sampling path;

[0112] S600: Set evaluation indicators, and use the evaluation indicators to comprehensively evaluate and screen the sampling paths, and obtain the sampling path corresponding to the best evaluation indicator as the best navigation path.

[0113] In some embodiments of the present invention, a path navigation method for a mobile robot in densely populated environments includes: identifying pedestrians, objects (such as railings, seats, etc.), and other important elements in a densely populated scene (such as a shopping mall, a train station, etc.); assigning semantic information (such as "this is a chair" or "this is a staircase") to the identified objects; and integrating their location and semantic information into a map to form semantic map information. This semantic map not only contains location information but also semantic descriptions of objects, which helps with subsequent path planning and interaction.

[0114] Semantic information analysis is performed on objects to further analyze their semantic information, such as determining their function (e.g., "This door is an exit," "This area is closed to traffic"). A unified framework or method is adopted to describe the interaction relationships between pedestrians, machines (such as mobile robots and smart devices), and objects in the environment. For example, the interaction methods between people and doors (e.g., "People pass through doors") and between machines and obstacles (e.g., "Machines avoid obstacles"). Based on the above description, a model is built to simulate and predict the interaction behaviors between people, machines, and objects in order to better plan paths and behaviors.

[0115] By connecting multiple sampled paths (i.e., possible walking paths of a mobile robot), a more complex path tree structure is formed. This method can increase path diversity, improve the flexibility of path planning, and formulate sampled paths of appropriate density based on the characteristics of densely populated environments (such as high crowd density and rapid dynamic changes) to characterize various navigation behaviors. For example, increasing the density of sampled paths in densely populated areas can better reflect environmental changes.

[0116] Different time slots are allocated to the sampled paths based on different navigation modes to simulate different pedestrian navigation behaviors. For example, in solo mode, the time allocation for the path will tend to favor shorter paths; while in follow mode or stop mode, the time allocation for the path will tend to favor longer and safer paths.

[0117] Check for collision risks along the sampling path, such as collisions with pedestrians or obstacles. Assess the comfort of the sampling path, such as whether the path is too winding or requires frequent turns; comfort assessment can improve the user experience of the path.

[0118] A series of evaluation indicators are set up, and all sampled paths are comprehensively evaluated based on these indicators to select the optimal path. Finally, the path with the highest comprehensive score of the evaluation indicators is selected as the best navigation path.

[0119] In summary, this invention provides an interactive space model to describe complex human interaction relationships, preventing mobile robots from intruding into pedestrians' personal interaction spaces. Using a universal set of sampled paths, only the corresponding time map needs to be updated to represent different motion patterns using the same sampled paths. Furthermore, by updating the time maps for different patterns, the same path set can represent different navigation behaviors, while also enabling efficient dynamic obstacle collision detection. In addition, a series of evaluation indicators are set and mapped to the sampled path score, making mobile robot navigation more consistent with human social norms. This effectively solves the problems of safe, efficient, and comfortable navigation for mobile robots in dynamic environments, allowing the movement of mobile robots to deeply integrate into human society.

[0120] In some embodiments of the present invention, step S100 involves identifying objects in a densely populated environment and constructing semantic map information based on these objects, including but not limited to:

[0121] To better understand the human social environment and help mobile robots accurately identify semantic information in the environment during navigation, objects in the environment, such as tables, chairs, doors, and pedestrians, are labeled during the mapping and localization process using lasers or cameras. The position information of each object in the environment is marked for subsequent interactive space modeling.

[0122] In laser-based mapping and localization, LiDAR (Light Detection and Ranging) acquires 3D point cloud data of the environment by emitting laser beams and measuring the time it takes for the reflected light to travel. It provides high-precision distance information, making it suitable for mapping and localization. In camera-based mapping and localization, the camera acquires visual information about the environment. RGB-D cameras (such as Kinect) can directly provide depth information, while monocular cameras require computer vision algorithms (such as stereo vision or monocular depth estimation) to obtain depth information. The acquired data undergoes preprocessing, such as noise reduction and filtering.

[0123] Object detection and recognition: Target detection algorithms are used to detect and classify objects in the environment; semantic segmentation algorithms are used to perform semantic segmentation of the environment to obtain more detailed object information.

[0124] Object position calibration: For LiDAR data, point cloud processing algorithms are used to extract the bounding box and position information of the object; for camera data, depth information and target detection results are combined to calculate the position of the object in three-dimensional space.

[0125] Labeling and semantic information annotation: Define a labeling system to assign a unique label to each object; map the detected objects to the labeling system and annotate semantic information.

[0126] Interactive space modeling: Based on the labeled object positions and semantic information, an interactive space model is constructed to describe the relationships and interaction methods between objects.

[0127] In some embodiments of the present invention, in step S300, a sampling path tree is constructed by grafting several sampling paths, and sampling paths of a reasonable density are provided based on the needs of densely populated environments, such as... Figure 2 As shown, sampling paths with a reasonable density are formulated according to the requirements of the navigation environment to characterize various navigation behaviors.

[0128] In some embodiments of the present invention, the navigation modes include follow mode, solo mode, and stop mode. In step S400, different sampling path time allocations are used based on different navigation modes to simulate different navigation behaviors of pedestrians, including:

[0129] In follow mode, the mobile robot is configured to allocate sampling path time according to a trapezoidal speed, and the upper limit of the speed allocation is set to dynamically change according to the pedestrian's speed. The time allocation formula is as follows:

[0130]

[0131] in, This represents the timestamp after the trapezoidal velocity allocation. Indicates the distance from the starting point to the path point. The length at that point, This indicates the set acceleration. Indicates the initial velocity. Indicates reference speed;

[0132] After completing the time allocation for each sampling path in follow mode, the first time allocation map on the sampling path is obtained.

[0133] See the detailed allocation ladder diagram. Figure 3 As shown, after time allocation is completed for each sampling path in follow mode, a first time allocation map on the sampling path can be obtained. The first time allocation map represents the time allocation of each path segment, as shown in the figure. Figure 4 As shown. Considering the shape of the mobile robot, the first-time assignment map on the specific sampling tree is as follows. Figure 5 As shown.

[0134] The mobile robot is configured to allocate sampling path time according to a trapezoidal speed, and the upper limit of the speed allocation is set to dynamically change with the pedestrian speed. This sampling path time allocation method in the following mode can ensure that the mobile robot can maintain stable movement while adapting to changes in pedestrian speed during the following process, thereby improving the robot's following performance and user experience.

[0135] In some embodiments of the present invention, in densely populated environments, treating pedestrians merely as dynamic obstacles may disrupt their movement order. When there are pedestrians with similar movement targets to the mobile robot, the mobile robot should select one of them as the leader pedestrian. This achieves navigation as efficiently as possible while avoiding interference with the crowd. This step of determining the leader pedestrian is used for navigation in follow mode. The selection of the leader pedestrian needs to meet the following two conditions.

[0136] First, no other pedestrians obstruct the mobile robot's view, such as... Figure 6As shown;

[0137] Secondly, as a leader, the pedestrian needs to have a similar direction of movement to the mobile robot, such as... Figure 7 As shown;

[0138] Understandably, specifically, the leading pedestrians or groups being followed need to meet the following conditions:

[0139] In the following mode, the mobile robot identifies pedestrians with similar directions of movement within a crowd as leader pedestrians, provided that the leader pedestrians meet the following conditions:

[0140]

[0141] in, Indicates the direction of movement of the leader pedestrian. Indicates the target direction of movement for the mobile robot. Indicates the current direction of movement of the mobile robot. This represents the angle between the direction of movement of the leader pedestrian and the target direction of movement of the mobile robot. The angle between the current direction of motion of the mobile robot and the target direction of motion of the mobile robot is represented by the symbol. Indicates the calculation of the angle between vectors;

[0142] The angle between the leader pedestrian's movement direction and the target movement direction of the mobile robot, and the angle between the mobile robot's current movement direction and the mobile robot's target movement direction, respectively satisfy the following conditions:

[0143]

[0144] in, This represents the angle between the direction of movement of the leader pedestrian and the target direction of movement of the mobile robot. This represents the maximum angle between the direction of movement of the leader pedestrian and the target direction of movement of the mobile robot. This represents the angle between the current direction of motion of the mobile robot and the target direction of motion of the mobile robot. This represents the maximum angle between the current direction of motion of the mobile robot and the target direction of motion of the mobile robot.

[0145] symbol The calculation of the angle between vectors is as follows:

[0146] ;

[0147] The leader pedestrian is selected by the above method, and other pedestrians are treated as dynamic obstacles.

[0148] The similarity between the mobile robot's current direction of movement and the direction of movement of each pedestrian can be calculated using the angle between the vectors; the smaller the angle, the more similar the directions. From all tracked pedestrians, those whose angle with the mobile robot's direction of movement is less than a preset threshold are selected as the leader pedestrians.

[0149] If multiple eligible pedestrians exist, other factors, such as pedestrian speed, distance, and behavioral patterns, can be further evaluated to determine the final leader pedestrian.

[0150] Speed ​​similarity: Select pedestrians whose speed is similar to that of the mobile robot.

[0151] Distance priority: Select pedestrians who are closer to the mobile robot for better tracking and interaction.

[0152] Behavioral patterns: Consider pedestrian behavior patterns, such as whether they lead the way in a crowd.

[0153] In some embodiments of the present invention, reference is made to... Figure 8 As shown, when the mobile robot detects a discrepancy between its target and that of a leader pedestrian, it gradually disengages from the following state, rather than continuously following a particular leader pedestrian. The mobile robot replans its path based on the current environment and task objectives, adjusting its route to avoid obstacles or congested areas, and attempts to identify a new leader pedestrian before resuming the following task. This ability of the mobile robot to safely and effectively adjust its behavior when detecting a discrepancy with the leader pedestrian's target ensures the successful completion of the task and a safe environment.

[0154] Using the above method, the mobile robot can effectively identify pedestrians with similar movement directions in crowds and follow them as leaders. This not only improves the navigation performance of the mobile robot, but also enhances its adaptability in complex environments.

[0155] In some embodiments of the present invention, in solo mode, the mobile robot is configured to allocate sampling path time according to trapezoidal velocity, and a maximum navigation speed for the trapezoidal velocity allocation is set. The time allocation formula is as follows:

[0156]

[0157] in, This represents the timestamp after the trapezoidal velocity allocation. Indicates the distance from the starting point to the path point. The length at that point, Indicates the maximum acceleration. This indicates the set acceleration. Indicates the initial velocity. Indicates maximum speed;

[0158] After completing the time allocation for each sampling path in solo mode, a second time allocation map on the sampling path is obtained.

[0159] It should be noted that the initial time allocation diagram along the sampling path of the mobile robot in follow mode, and the specific time allocation diagram, are similar to those in solo mode. See the specific allocation ladder diagram below. Figure 9 As shown, after time allocation for each sampling path, a second time allocation map on the sampling path can be obtained, such as... Figure 4 As shown, considering the shape of the mobile robot, the second time allocation map on the specific sampling tree is as follows. Figure 5 As shown.

[0160] Using a trapezoidal velocity curve for time allocation of the sampling path and setting a maximum navigation speed for the trapezoidal velocity allocation is an effective method to ensure that the robot can move efficiently while maintaining smoothness and safety during autonomous navigation. This trapezoidal velocity curve ensures smooth transitions during robot movement, reducing abrupt acceleration changes and thus improving comfort and safety.

[0161] In solo mode, the robot's maximum navigation speed is an important parameter, as it determines the robot's speed during the constant-speed phase. The following factors need to be considered when setting the maximum navigation speed:

[0162] Environmental complexity: In complex environments (such as densely populated areas), the maximum speed should be lower to ensure safety.

[0163] Task requirements: The maximum speed can be adjusted appropriately depending on the urgency of the task.

[0164] Robot performance: The robot's hardware performance (such as motor power, sensor accuracy, etc.) will also affect the setting of the maximum speed.

[0165] Through the above steps and techniques, the mobile robot can effectively allocate time for sampling paths and set the maximum navigation speed for trapezoidal velocity allocation in solo mode, thereby improving the robot's autonomous navigation performance and user experience.

[0166] In some embodiments of the present invention, in the stopped mode, the mobile robot is configured to allocate sampling path time according to a trapezoidal speed, and a speed deceleration requirement is set for the allocation. The time allocation formula is as follows:

[0167]

[0168] in, This represents the timestamp after the trapezoidal velocity allocation. Indicates the distance from the starting point to the path point. The length at that point, Indicates the maximum acceleration. This indicates the set acceleration. Indicates the initial velocity. Indicates the maximum speed. Indicates invalid;

[0169] After completing the time allocation for each sampling path in stop mode, a third time allocation map on the sampling path is obtained.

[0170] See the detailed allocation ladder diagram. Figure 10 As shown, after time allocation for each sampling path, a third time map on the sampling path can be obtained, such as... Figure 11 As shown, considering the shape of the mobile robot, the specific time allocation diagram is as follows: Figure 12 As shown.

[0171] In the stopping mode of a mobile robot, although the robot's ultimate goal is to stop, it may need to move along a specific path to the stopping point before stopping. In this case, using a trapezoidal velocity curve for the time allocation of the sampling path and setting deceleration requirements for the allocation is a reasonable approach. This ensures that the robot reaches the stopping point smoothly and safely during deceleration, thereby improving the robot's stopping performance and user experience.

[0172] In some embodiments of the present invention, semantic information analysis is performed on objects, and a unified interaction space method is used to describe human-machine-object interactions to construct an interaction space model. This includes: identifying objects based on sensors, completing semantic information analysis, and using a unified interaction space method to describe human-machine-object interactions, including the interaction space description between pedestrians, between robots and pedestrians, and between pedestrians and objects. (See also...) Figure 13 As shown, Figure 13 This is a schematic diagram of a pedestrian interaction space based on a unified interaction space model provided in an embodiment of the present invention, wherein... Figure 14 It is an interactive space diagram composed of multiple pedestrians. Figure 15 A diagram representing the interactive space between a single pedestrian and objects;

[0173] The interaction space modeling method is as follows:

[0174] Construct a pedestrian interaction space model. The formula for the pedestrian interaction space model is:

[0175]

[0176] in, Represents pedestrian interaction space, Represents any point in space. Indicates time, Indicates the weighting coefficient. It is an asymmetric Gaussian distribution function, representing the interaction space range of pedestrians or objects. The specific formula for the interaction space range is described as follows:

[0177] in, Represents any point in space. It represents the pose information of pedestrians or objects. Indicates parameters, Represents the backward covariance matrix. Represents the forward covariance matrix;

[0178] By performing a transformation operation on the asymmetric Gaussian distribution function, we obtain the transformation formula for the asymmetric Gaussian distribution function, which is as follows:

[0179]

[0180]

[0181]

[0182] in, Represents any point in space. It represents the pose information of pedestrians or objects. Indicates parameters, The rotation matrix representing the pedestrian. Indicates the location of the pedestrian. , Indicates the pedestrian's position and posture. Represents the backward covariance matrix. Denotes the forward covariance matrix. This represents the variance value. , , These are parameters used to describe pedestrian space. This represents the transverse covariance of an asymmetric Gaussian function. This represents the forward covariance of an asymmetric Gaussian function. This represents the backward covariance of an asymmetric Gaussian function. This indicates the normalized orientation angle.

[0183] Understandable Indicates time Midpoint in space pedestrian interaction space This represents a weighting coefficient, used to indicate the relative importance or influence of different interaction spaces. It indicates the pose information of pedestrians or objects, specifically including position and orientation. The parameter is the covariance matrix, which describes the shape and extent of the Gaussian distribution and reflects the uncertainty or variability of the interaction space. , , It is a parameter used to describe pedestrian space, and its size determines the size of the interaction space. This allows us to obtain pedestrian interaction space such as Figure 13 As shown.

[0184] This interaction space model, by representing the interaction space of each pedestrian or object as a weighted asymmetric Gaussian distribution, can more flexibly describe the distribution and interaction behavior of pedestrians in space. Weighting coefficients This allows the model to adjust the impact of different interaction spaces based on actual conditions, while the covariance matrix... This provides control over the shape and size of the interactive space, enabling it to adapt to different environments and behavioral patterns.

[0185] In addition, weighting coefficients were set for normalization. The specific calculation process is as follows:

[0186] By setting weighting coefficients to normalize the values ​​of the pedestrian interaction space, the numerical distribution of the pedestrian interaction space is obtained. The weighting coefficients are calculated as follows:

[0187]

[0188]

[0189] in, i=j This represents the pedestrian's own interaction space, at this time =1, therefore, the expression for the numerical distribution of the pedestrian interaction space is as follows:

[0190]

[0191] Therefore, the above equation is a matrix problem, which involves a system of linear equations, and its purpose is to solve for the weight coefficients. This process achieves numerical normalization of the pedestrian interaction space, ensuring that the sum of all interaction spaces is 1, thus allowing the model's output to be interpreted as a probability distribution. This allows for numerical normalization of any number of interactions, preventing excessive encroachment on the feasible space as the number of interactions increases.

[0192] Furthermore, the pedestrian space can be divided into private space, social space, and public space from near to far. Understandably, the target sampling path is not allowed to enter the private space to avoid potential collisions, and in order to ensure the comfort of pedestrians, the target sampling path should not intrude into the social space as much as possible, but the target sampling path can fall into the public space.

[0193] The scope of private space, social space, and public space is described using the numerical distribution in the aforementioned unified interaction space as follows:

[0194] The region where the pedestrian interaction space has a numerical distribution in [0.6, 1) is defined as private space, and the paths distributed in private space are classified as unsampled paths, meaning that these paths are not suitable for interaction between mobile robots and pedestrians. The region where the pedestrian interaction space has a numerical distribution in [0.4, 0.6) is defined as social space, and the paths distributed in social space are classified as undetermined sampling paths, which may be suitable for interaction but require further evaluation. The region where the pedestrian interaction space has a numerical distribution in [0, 0.4) is defined as public space, and the paths distributed in public space are classified as target sampling paths, which are most suitable for interaction between robots and pedestrians.

[0195] This interactive space model and normalization method help mobile robots select optimal paths in complex environments, avoiding private spaces and prioritizing public spaces. It analyzes pedestrian behavior patterns in different spaces to optimize the design and management of public spaces. By considering pedestrian interaction spaces in path planning, it improves traffic flow efficiency and safety. This approach allows for more effective management and optimization of pedestrian and robot interactions in different spaces, enhancing overall safety and efficiency.

[0196] In some embodiments of the present invention, collision detection and comfort assessment of the sampling path includes:

[0197] An evaluation model is established for collision detection and comfort assessment of the sampling path. The expression of the evaluation model is as follows:

[0198]

[0199] in, Indicates the evaluation model, Indicates the sampling path, Indicates sampling path The point on, Indicates time, Indicates the pose point The values ​​of the numerical distribution in the pedestrian interaction space. This represents the time difference between the timestamp of a waypoint and the predicted time of a pedestrian's arrival at that waypoint. Indicates the time threshold;

[0200] The expression for the time threshold is:

[0201]

[0202] in, Indicates the time threshold. The diameter of the circle representing the pedestrian. Indicates the pedestrian's current speed. This represents the coefficient of thermal expansion, which is usually greater than 1.0;

[0203] When the mobile robot enters the pedestrian interaction space, the closer the mobile robot gets to the pedestrian's sampling path... The stronger the interference of the mobile robot with pedestrians, the greater the value of the interference, which is taken as the comfort level of the sampling path. The expression for the comfort level of the sampling path is:

[0204]

[0205] in, Indicates the comfort level of the sampling path. This represents the interference value of the mobile robot on the pedestrian sampling path. Indicates the sampling path, Indicates time;

[0206] Understandable This represents the interference value of the mobile robot on the pedestrian's sampling path ξ at time t. This value is calculated based on the numerical distribution of the pedestrian interaction space. The comfort level of the sampling path reflects the degree of interference the mobile robot causes to pedestrians on a specific path. Considering that the mobile robot should avoid entering the pedestrian's interaction space as much as possible, when the mobile robot enters the pedestrian's interaction space, the sampling path closer to the pedestrian's center... The stronger the interference to pedestrians, the greater the value of the interference to pedestrians, and the more comfortable the path is.

[0207] Collision detection is performed by obtaining the numerical distribution of the pedestrian interaction space. Paths with numerical distributions in [0.6, 1) and [0, 0.4) are divided into collision-prone areas.

[0208] If the sampling path is a path where the numerical distribution of the pedestrian interaction space is in the range [0.6, 1), then this path needs to be deleted and cannot participate in the planning process. If the numerical distribution of the pedestrian interaction space is in the range [0, 0.4), it may lead to a lower overall score for the sampling path, so the probability of this path being selected will be reduced.

[0209] Comfort assessment considers the spatial distribution of pedestrian interactions along the path, avoiding interactions in private areas. By setting different sampling path time allocations, the time sensitivity of the assessment model can be adjusted, thereby optimizing path selection to improve comfort. In dynamic environments, mobile robots can use the assessment model to predict and avoid collisions with pedestrians while selecting paths with higher comfort levels. In path planning, it can help optimize the paths of pedestrians and mobile robots, reduce congestion, and improve safety. It can more effectively manage and optimize the interactions of pedestrians and mobile robots in different spaces, improving overall safety and comfort.

[0210] In some embodiments of the present invention, in step S600 of the path navigation method for a mobile robot in a densely populated environment, an evaluation index is set, and the sampled paths are comprehensively evaluated and screened using the evaluation index to obtain the sampled path corresponding to the best evaluation index as the best navigation path, including:

[0211] Efficiency is set as the core indicator for mobile robot sampling path navigation. The formulas for efficiency evaluation indicators under different motion mode conditions are as follows:

[0212]

[0213] in, Indicates efficiency indicators. This indicates path length calculation. This represents the vector between the end of the sampling path and the local target point. This represents the vector from the starting point to the local target point. This indicates the maximum speed of the mobile robot. Indicates the pedestrian's current speed. The diameter of the circle representing the pedestrian. This represents the diameter of the circle representing the mobile robot. This indicates the angle between the direction of movement of the leader pedestrian and the target direction of movement of the mobile robot;

[0214] Considering that pedestrians will follow underlying social rules in real-world scenarios, such as keeping to the side of the road, we have set up social norm indicators for quantitative application in mobile robots. The formula for these social norm indicators is as follows:

[0215]

[0216] in, Indicators representing social norms and All are constants and satisfy the following conditions: ;

[0217] A consistency metric is set to measure the degree of oscillation in the planned sampling path of the mobile robot at different time points. The formula for the consistency metric is as follows:

[0218]

[0219] in, Indicators of consistency This represents the distance between two sampling paths. A smaller value indicates higher consistency between the two sampling paths. The consistency index measures the degree of oscillation between the planned paths at different time points. In practice, it is required that the selected paths remain as stable and consistent as possible to reduce oscillations and achieve better predictability. This represents the Fréchet distance between two trajectories; a smaller value indicates higher consistency.

[0220] A smoothness metric is set up to measure the smoothness of the sampling path planned by the mobile robot. The formula for the smoothness metric is as follows:

[0221]

[0222] in, Indicates smoothness index, This represents the curvature calculation function. Indicates the sampling path;

[0223] The planned path should be as smooth as possible, with minimal changes in curvature and minimal jitter, in order to achieve efficient navigation as much as possible.

[0224] Based on efficiency, social norms, consistency, and smoothness indicators, a multi-dimensional evaluation equation is derived. Multiple sampling paths are then comprehensively evaluated and selected according to this equation to obtain the optimal navigation path corresponding to the best evaluation indicator. The formula for the multi-dimensional evaluation equation is as follows:

[0225]

[0226] in, Indicates the best navigation path. This represents the weighting coefficient of each evaluation indicator. Indicates evaluation indicators, , , , , In different environments Different values ​​need to be set. (Refer to...) Figure 16 The semantic information shown is used to select and adjust parameters for the navigation process based on the robot's environment in the semantic map.

[0227] Based on the navigation behavior of different sampled paths obtained from the evaluation, the sampled path with the minimum value is selected as the optimal navigation path, such as... Figure 17 and Figure 18 As shown in the image. Next, the optimal navigation route is input to the vehicle controller, which then tracks the optimal route to achieve a safe and efficient navigation task.

[0228] It should be understood that the method steps in the embodiments of the present invention can be implemented or carried out by computer hardware, a combination of hardware and software, or by computer instructions stored in a non-transitory computer-readable storage medium. The method can use standard programming techniques. Each program can be implemented in a high-level procedural or object-oriented programming language to communicate with the computer system. However, if necessary, the program can be implemented in assembly or machine language. In any case, the language can be a compiled or interpreted language. Furthermore, for this purpose, the program can run on a programmed application-specific integrated circuit (ASIC).

[0229] Furthermore, the procedures described herein may be performed in any suitable order unless otherwise indicated herein or otherwise clearly contradicted by the context. The procedures described herein (or variations and / or combinations thereof) may be executed under the control of one or more computer systems configured with executable instructions, and may be implemented by hardware or a combination thereof as code (e.g., executable instructions, one or more computer programs, or one or more applications) that commonly executes on one or more processors. The computer program comprises a plurality of instructions executable by one or more processors.

[0230] Furthermore, the method can be implemented in any suitable type of computing platform, including but not limited to personal computers, minicomputers, mainframes, workstations, networked or distributed computing environments, standalone or integrated computer platforms, or in communication with charged particle tools or other imaging devices, etc. Aspects of the invention can be implemented as machine-readable code stored on a non-transitory storage medium or device, whether removable or integrated into a computing platform, such as a hard disk, optical read and / or write storage medium, RAM, ROM, etc., such that it is readable by a programmable computer, and when the storage medium or device is read by the computer, it can be used to configure and operate the computer to perform the processes described herein. Furthermore, the machine-readable code, or portions thereof, can be transmitted via wired or wireless networks. The invention described herein includes these and other different types of non-transitory computer-readable storage media when such media comprises instructions or programs that implement the steps described above in conjunction with a microprocessor or other data processor. When programmed according to the methods and techniques described in the invention, the invention may also include the computer itself.

[0231] A computer program can be applied to input data to perform the functions described herein, thereby transforming the input data to generate output data stored in non-volatile memory. The output information can also be applied to one or more output devices, such as a display. In a preferred embodiment of the invention, the transformed data represents physical and tangible objects, including specific visual depictions of physical and tangible objects generated on the display.

[0232] The above description is merely a preferred embodiment of the present invention. The present invention is not limited to the above-described embodiments. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention, as long as they achieve the technical effects of the present invention by the same means, should be included within the scope of protection of the present invention. Within the scope of protection of the present invention, the technical solutions and / or implementation methods can have various modifications and variations.

Claims

1. A path navigation method for a mobile robot in a dense crowd environment, characterized by, The method comprises the following steps: Identifying objects in a dense crowd environment, and constructing semantic map information based on the objects; Performing semantic information analysis on the objects, and describing human-machine-object using a unified interaction space method to construct an interaction space model; Using a plurality of sampling path grafting methods to construct a sampling path tree, and providing a reasonable density of sampling paths based on the requirements of the dense crowd environment; Using different sampling path time allocations based on different navigation modes to simulate different navigation behaviors of pedestrians; Performing collision detection and comfort evaluation on the sampling paths; Setting evaluation indexes, and performing comprehensive evaluation and screening on the sampling paths through the evaluation indexes to obtain a sampling path corresponding to the best evaluation index as the best navigation path; The semantic information analysis on the objects and the description of human-machine-object using the unified interaction space method to construct the interaction space model comprises the following steps: Identifying objects based on sensors to complete semantic information analysis, and describing human-machine-object using the unified interaction space method, including the interaction space description between pedestrians, robots and objects, to construct the interaction space model, and the method for constructing the interaction space model comprises the following steps: Constructing a pedestrian interaction space model, and the formula of the pedestrian interaction space model is as follows: wherein, represents the interaction space of a pedestrian, represents an arbitrary point in the space, represents time, represents a weight coefficient, is an asymmetric Gaussian distribution function, representing the interaction space range of a pedestrian or object, the specific formula of which is described as follows: wherein, represents an arbitrary point in space, represents pose information of a pedestrian or object, represents a parameter, represents a backward covariance matrix, represents a forward covariance matrix; Performing transformation operation on the asymmetric Gaussian distribution function to obtain a transformation formula of the asymmetric Gaussian distribution function, and the transformation formula of the asymmetric Gaussian distribution function is as follows: wherein, represents an arbitrary point in space, represents pose information of a pedestrian or object, represents a parameter, represents a rotation matrix of a pedestrian, represents a position of a pedestrian, , represents a pedestrian pose, represents a backward covariance matrix, represents a forward covariance matrix, represents a variance value, , , is a parameter for describing a pedestrian space, represents a lateral covariance of an asymmetric Gaussian function, represents a forward covariance of an asymmetric Gaussian function, represents a backward covariance of an asymmetric Gaussian function, represents a normalized orientation angle. 2.The path navigation method for a mobile robot in a dense crowd environment according to claim 1, wherein, The navigation modes include a following mode, and the different sampling path time allocations based on different navigation modes to simulate different navigation behaviors of pedestrians comprise the following steps: In the following mode, the mobile robot is set to perform sampling path time allocation according to trapezoidal velocity, and the upper limit of the velocity allocation is set to dynamically change to follow the pedestrian velocity, and the time allocation formula is as follows: wherein, denotes the time stamp after trapezoidal velocity distribution, denotes the length from the start point to the path point at the point, denotes the set acceleration, denotes the initial velocity, denotes the reference velocity; After completing the time allocation of each sampling path in the following mode, a first time allocation diagram on the sampling path is obtained. 3.The path navigation method of the mobile robot in the dense crowd environment according to claim 2, wherein, The method further comprises the following steps: In the following mode, the mobile robot determines a pedestrian having a similar motion direction as the mobile robot in the crowd flow as a leader pedestrian, and the leader pedestrian satisfies the following conditions: wherein, denotes the direction of motion of the leader pedestrian, denotes the target direction of motion of the mobile robot, denotes the current direction of motion of the mobile robot, denotes the angle between the direction of motion of the leader pedestrian and the target direction of motion of the mobile robot, denotes the angle between the current direction of motion of the mobile robot and the target direction of motion of the mobile robot, the symbol denotes the angle calculation of vectors; The included angle between the motion direction of the leader pedestrian and the target motion direction of the mobile robot and the included angle between the current motion direction of the mobile robot and the target motion direction of the mobile robot respectively satisfy the following conditions: wherein denotes an angle between the moving direction of the leader pedestrian and the target moving direction of the mobile robot, denotes a maximum value of the angle between the moving direction of the leader pedestrian and the target moving direction of the mobile robot, denotes an angle between the current moving direction of the mobile robot and the target moving direction of the mobile robot, denotes a maximum value of the angle between the current moving direction of the mobile robot and the target moving direction of the mobile robot. 4.The path navigation method of a mobile robot in a dense crowd environment according to claim 1, wherein, The navigation modes further include a solo mode and a stop mode, and the different sampling path time allocations based on different navigation modes to simulate different navigation behaviors of pedestrians comprise the following steps: In the solo mode, the mobile robot is set to perform sampling path time allocation according to trapezoidal velocity, and the maximum navigation velocity of the trapezoidal velocity allocation is set, and the time allocation formula is as follows: wherein, denotes the time stamp after trapezoidal velocity distribution, denotes the length from the start to the path point at the path point, denotes the maximum acceleration, denotes the set acceleration, denotes the initial velocity, denotes the maximum velocity; After completing the time allocation of each sampling path in the solo mode, a second time allocation diagram on the sampling path is obtained. In the stop mode, the mobile robot is set to perform sampling path time allocation according to trapezoidal velocity, and the velocity is set to be allocated according to the deceleration requirement, and the time allocation formula is as follows: wherein, represents the time stamp after trapezoidal velocity distribution, represents the length from the start point to the path point at the path point, represents the maximum acceleration, represents the set acceleration, represents the initial velocity, represents the maximum velocity, represents invalid; After completing the time allocation of each sampling path in the stop mode, a third time allocation diagram on the sampling path is obtained. 5.The path navigation method of a mobile robot in a dense crowd environment according to claim 1, wherein, The method further comprises the following steps: The weight coefficient is set to realize numerical normalization processing of the pedestrian interaction space, to obtain a numerical distribution of the pedestrian interaction space, and the weight coefficient is calculated in the following manner: where i = j represents the interaction space of the pedestrian himself, in which case = 1, so the expression of the numerical distribution of the pedestrian interaction space is as follows: The path with the numerical distribution of the pedestrian interaction space in [0.6, 1] is divided into a non-sampling path, the path with the numerical distribution of the pedestrian interaction space in [0.4, 0.6] is divided into a to-be-determined sampling path, and the path with the numerical distribution of the pedestrian interaction space in [0, 0.4] is divided into a target sampling path. 6.The path navigation method of the mobile robot in the dense crowd environment according to claim 5, wherein, The collision detection and comfort evaluation of the sampling path include: An evaluation model for collision detection and comfort evaluation of the sampling path is established, and the expression of the evaluation model is as follows: wherein, represents an evaluation model, represents a sampling path, represents a point on a sampling path , represents a time, represents a pose point at a value of a numerical distribution of a pedestrian interaction space, represents a difference between a timestamp of a path point and a time when the pedestrian is predicted to arrive at the path point, represents a time threshold; The expression of the time threshold is as follows: wherein, represents a time threshold, represents a circle diameter of a pedestrian, represents a current speed of a pedestrian, represents an expansion coefficient, typically greater than 1.0; When the mobile robot enters the pedestrian interaction space, the closer the mobile robot is to the sampling path of the pedestrian , the stronger the interference of the mobile robot on the pedestrian, and the maximum value of the interference of the mobile robot on the pedestrian is obtained as the comfort degree of the sampling path, and an expression of the comfort degree of the sampling path is: wherein, represents a comfort of the sampled path, represents a sampled path interference value of the mobile robot to the pedestrian, represents a sampled path, represents time; The numerical distribution of the pedestrian interaction space is obtained for collision detection, and the path with the numerical distribution of the pedestrian interaction space in [0.6, 1] and the path with the numerical distribution of the pedestrian interaction space in [0, 0.4] are divided into an easy collision detection area. 7.The path navigation method of a mobile robot in a dense crowd environment according to claim 1, wherein, The evaluation index is set, and the sampling path is comprehensively evaluated and screened through the evaluation index, and the sampling path corresponding to the best evaluation index is obtained as the best navigation path, which includes: The efficiency index is set as the core index of the sampling path navigation of the mobile robot, and the formula of the efficiency evaluation index under different motion mode conditions is as follows: wherein, denotes the efficiency indicator, denotes the path length calculation, denotes the vector between the end of the sample path and the local target point, denotes the vector from the start point to the local target point, denotes the maximum velocity of the mobile robot, denotes the current velocity of the pedestrian, denotes the circle diameter of the pedestrian, denotes the circle diameter of the mobile robot, denotes the angle between the motion direction of the leader pedestrian and the target motion direction of the mobile robot; The social norm index is set to be quantitatively used in the mobile robot, and the formula of the social norm index is as follows: wherein, represents a social norm indicator, and are constants and satisfy ; The consistency index is set to measure the oscillation degree of the sampling path planned by the mobile robot at different time points, and the formula of the consistency index is as follows: wherein, represents a consistency index, represents a distance between two sampling paths, the smaller the value of the consistency index represents the higher the consistency of the two sampling paths; The smoothness index is set to measure the smoothness of the sampling path planned by the mobile robot, and the formula of the smoothness index is as follows: wherein, denotes a smoothness index, denotes a curvature calculation function, denotes a sampling path; Based on the efficiency index, the social norm index, the consistency index, and the smoothness index, an evaluation equation of a multi-dimensional index is obtained, a plurality of sampling paths are comprehensively evaluated and screened according to the evaluation equation of the multi-dimensional index, and the sampling path corresponding to the best evaluation index is obtained as the best navigation path, and the formula of the evaluation equation of the multi-dimensional index is as follows: wherein, represents the optimal navigation path, represents the weight coefficient of each evaluation index, represents the evaluation index.

8. A computer apparatus comprising a memory and a processor, characterised in that, The processor executes the computer program stored in the memory to implement the method of any one of claims 1 to 7.

9. A computer readable storage medium having stored thereon program instructions, wherein, The program instructions are executed by the processor to implement the method of any one of claims 1 to 7.

Images