Navigation method and system of autonomous mobile robot in dense people stream environment

Abstract

The invention discloses a navigation method and system of an autonomous mobile robot in a dense people stream environment. The method comprises the following steps: acquiring environmental data by using a sensor system on the robot; recognizing pedestrians, obstacles and ground states based on the collected data to obtain real-time sensing information; judging whether the environment is a dense people flow environment according to the real-time sensing information, if not, executing a normal navigation mode, and if yes, executing a people flow passing mode. The people flow passing mode comprises the following steps: dynamically planning a passing path according to a preset global path and real-time sensing information; judging whether the running direction of a pedestrian is consistent with the passing path direction or not, selecting a matched pedestrian as a following object, and keeping a safe distance; and if no matched pedestrian exists, the passing possibility is evaluated, and when the possibility is higher than a preset threshold value, the robot is controlled to perform real-time passing action. Through data fusion, dynamic path planning and an intelligent following strategy, the navigation efficiency, the success rate and the safety of the robot in a dense people stream environment are improved.

Description

Technical Field

[0001] The present invention relates to the field of robot control technology, and in particular to a navigation method and system for an autonomous mobile robot in a dense human flow environment. Background Art

[0002] With the development of autonomous mobile robot technology, more and more robots are being used in various scenarios, such as shopping malls, airports, hospitals and other public places. These places usually have a large number of dynamic people flows, which brings great challenges to the navigation of robots. Traditional navigation algorithms, such as global path planning algorithms (A*, Dijkstra, JPS, etc.) and local planning algorithms (DWA, TEB, etc.), although they perform well in static or low-dynamic environments, are incapable of handling dense crowd environments. These algorithms usually assume that the environment is static and cannot cope with dynamically changing crowd flows in real time. When encountering sudden pedestrians or obstacles, robots often need to replan their paths frequently, resulting in inefficient navigation. In addition, although local planning algorithms can handle a certain degree of dynamic environments, they are difficult to make optimal decisions in highly dynamic and unpredictable crowd flows, which can easily cause robots to frequently pause, collide or get stuck in the crowd flow.

[0003] In dense crowds, the risk of collision between robots and pedestrians increases significantly, affecting user experience and safety. Robots frequently replan their paths in dense crowds, resulting in longer task execution time and increased energy consumption. In complex dynamic environments, it is difficult for robots to choose the optimal path, especially when they need to cross the crowd horizontally, and there is a lack of effective strategy and algorithm support. Therefore, the existing autonomous mobile robot navigation technology has many shortcomings in dense crowd environments, and a new method is urgently needed to solve these problems and improve the navigation performance of robots in complex dynamic environments.

[0004] In the prior art, patent CN110941279A discloses a control method for preventing a robot from crushing its feet in a dense crowd scene. The method uses different detection information of sensor components to establish regional divisions of obstacles, including obstacle areas, human foot estimation areas, and safety areas. Based on these regional divisions, the robot is controlled to prevent the robot from crushing pedestrians' feet in dense crowds. This patent only involves preventing the robot from crushing its feet, and does not provide any revelations related to robot navigation. Patent CN115542902A discloses a robot path planning method, device, medium, and equipment based on crowd estimation. The main goal of this patent is to predict the future movement trajectory of pedestrians in a dense crowd environment and plan a path for the robot to avoid them in advance. The main technical route is to avoid pedestrians. In actual implementation, the planned route is complex and the robot's operating efficiency is low. Summary of the invention

[0005] Purpose of the invention: In order to overcome the deficiencies in the prior art, the present invention provides an efficient, safe and adaptable navigation method and system for an autonomous mobile robot in a dense human flow environment.

[0006] Technical solution: To achieve the above-mentioned purpose, the navigation method of the autonomous mobile robot in a dense human flow environment of the present invention comprises:

[0007] The sensor system on the robot is used to collect environmental data; the sensor system includes RGBD cameras, RGB cameras, TOF cameras, ultrasonic sensors, lidar sensors and other sensors.

[0008] Based on the collected data, pedestrians, obstacles and ground status data are identified to obtain real-time perception information;

[0009] Based on the real-time perception information, determine whether it is a dense crowd flow environment, if not, execute the normal navigation mode, if yes, execute the crowd flow passing mode;

[0010] The pedestrian flow mode specifically includes the following steps:

[0011] Dynamically plan the travel path based on the preset global path and real-time perception information;

[0012] According to the travel path, determining whether there is a pedestrian whose running direction is consistent with the direction of the travel path, and obtaining a first determination result;

[0013] When the first judgment result is yes, a pedestrian matching the direction of the travel path is selected as a follow object, and a safe distance is maintained;

[0014] This step is applicable when the robot moves along the flow of people or when the crowd moves in multiple directions and there are pedestrians moving in the same direction as the robot.

[0015] When the first judgment result is no, the possibility of walking is evaluated, and when the possibility is higher than a preset threshold, the robot is controlled to move in real time.

[0016] This step is applicable to situations where the robot is crossing or moving against the flow of people and there are no pedestrians moving in the same direction as the robot.

[0017] Furthermore, based on the collected data, pedestrians, obstacles and ground status data are identified to obtain real-time perception information, including:

[0018] Preprocess the data collected by each sensor, including noise filtering and outlier processing;

[0019] Extract useful feature information from the processed data, including the location, speed, direction of pedestrians, the size and location of obstacles, and other specific object information;

[0020] Using Kalman filtering or particle filtering algorithm to fuse information from different sensors to obtain the real-time perception information;

[0021] The information from different sensors is fused using algorithms such as Kalman filtering or particle filtering to obtain the real-time perception information.

[0022] Furthermore, the dynamically planning of the travel path according to the preset global path and the real-time perception information includes:

[0023] Based on the information of pedestrians, obstacles and other specific objects, the feasible paths are calculated and scored using the A* algorithm, Dijkstra algorithm or fast search random tree algorithm;

[0024] In this step, when calculating the feasible path, the control system extracts and comprehensively calculates the movement directions of multiple pedestrians in front to determine the overall movement direction of the crowd, and then plans the path along the overall movement direction based on the robot's position and destination, so that the robot's next movement direction is consistent with the overall movement direction. In this way, conflicts between the robot and pedestrians can be effectively avoided.

[0025] In combination with a pedestrian following strategy, a genetic algorithm or a particle swarm optimization algorithm is used to optimize the feasible path;

[0026] The feasible path is made smoother by a smoothing algorithm to obtain the travel path.

[0027] Further, the selecting a pedestrian whose direction matches the travel path as a follow-up object and maintaining a safe distance includes:

[0028] Based on the moving direction and speed information of each pedestrian ahead, a pedestrian with a high matching degree with the travel path is selected as a follow target;

[0029] Dynamically adjust the safety distance according to the speed and direction of the target being followed;

[0030] Based on the PID controller, the proportional, integral and differential parameters are adjusted to maintain stable following of the following target.

[0031] A navigation system for an autonomous mobile robot in a dense human flow environment, the navigation system may include or be divided into one or more program modules, one or more program modules are stored in a storage medium and executed by one or more processors to complete the present invention and implement the above-mentioned navigation method. The program module referred to in the scheme of the present invention refers to a series of computer program instruction segments that can complete specific functions, which is more suitable for describing the execution process of the navigation method in the storage medium than the program itself. The following description will specifically introduce the functions of each program module of this scheme: the system includes:

[0032] A collection module, which is used to collect environmental data using a sensor system on the robot;

[0033] A recognition module, which is used to recognize pedestrians, obstacles and ground state data based on the collected data to obtain real-time perception information;

[0034] A determination implementation module, which is used to determine whether it is a dense crowd flow environment based on the real-time perception information, if not, execute the normal navigation mode, if yes, execute the crowd flow passing mode;

[0035] The pedestrian flow mode specifically includes the following steps:

[0036] Dynamically plan the travel path based on the preset global path and real-time perception information;

[0037] According to the travel path, determining whether there is a pedestrian whose running direction is consistent with the direction of the travel path, and obtaining a first determination result;

[0038] When the first judgment result is yes, a pedestrian matching the direction of the travel path is selected as a follow object, and a safe distance is maintained;

[0039] When the first judgment result is no, the possibility of walking is evaluated, and when the possibility is higher than a preset threshold, the robot is controlled to move in real time.

[0040] Other contents of implementing the above-mentioned navigation method based on the above-mentioned navigation system have been introduced in detail in the previous scheme. Please refer to the corresponding contents in the previous scheme and will not be repeated here.

[0041] Beneficial effects: The navigation method and system of the autonomous mobile robot in a dense human flow environment of the present invention have the following beneficial effects:

[0042] The robot can switch between normal navigation mode and pedestrian walking mode as needed to improve walking efficiency. When the robot is in pedestrian walking mode, it can follow pedestrians as needed and intelligently evaluate the possibility of walking when there are no pedestrians to follow to ensure the safety of the robot and pedestrians. Based on the above navigation method, the robot's navigation walking efficiency and its ability to cope with complex scenarios can be significantly improved, while ensuring the success rate and safety, reducing collisions between robots and pedestrians, optimizing path planning, and reducing travel time and energy consumption. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] Figure 1 A flowchart of a navigation method for an autonomous mobile robot in a dense crowd environment;

[0044] Figure 2 Schematic diagram of the flow of people passing through the pattern. DETAILED DESCRIPTION

[0045] The present invention will be further described below in conjunction with the accompanying drawings.

[0046] A navigation method for an autonomous mobile robot in a dense crowd environment, the method comprising the following steps S101-S103:

[0047] Step S101, using the sensor system on the robot to collect environmental data; the sensor system includes sensors such as RGBD camera, RGB camera, TOF camera, ultrasonic sensor, lidar sensor, etc.

[0048] Step S102, based on the collected data, identifying pedestrians, obstacles and ground state data to obtain real-time perception information;

[0049] Step S103, judging whether it is a dense crowd flow environment based on the real-time perception information, if not, executing the normal navigation mode, if yes, executing the crowd flow passing mode;

[0050] The pedestrian flow mode specifically includes the following steps S201-S204:

[0051] Step S201, dynamically planning a travel path according to a preset global path and real-time perception information;

[0052] Step S202, judging whether there is a pedestrian whose running direction is consistent with the direction of the walking path according to the walking path, and obtaining a first judgment result;

[0053] Step S203: when the first judgment result is yes, a pedestrian matching the direction of the travel path is selected as a follow object, and a safe distance is maintained;

[0054] This step is applicable when the robot moves along the flow of people or when the crowd moves in multiple directions and there are pedestrians moving in the same direction as the robot.

[0055] Step S204, when the first judgment result is no, the possibility of walking is evaluated, and when the possibility is higher than a preset threshold, the robot is controlled to perform a real-time walking action.

[0056] This step is applicable to situations where the robot is crossing or moving against the flow of people and there are no pedestrians moving in the same direction as the robot.

[0057] In the above method, the robot can switch between normal navigation mode and crowd flow mode as needed to improve walking efficiency. When the robot is in crowd flow mode, it can follow pedestrians as needed and intelligently evaluate the possibility of passage when there are no pedestrians to follow to ensure the safety of the robot and pedestrians. Based on the above navigation method, the robot's navigation walking efficiency and its ability to cope with complex scenarios can be significantly improved.

[0058] Preferably, in the above step S102, pedestrians, obstacles and ground state data are identified based on the collected data to obtain real-time perception information, including the following steps S301-S304:

[0059] Step S301, preprocessing the data collected by each sensor, including noise filtering and outlier processing;

[0060] Step S302, extracting useful feature information from the processed data, the feature information including the position, speed, direction of pedestrians, the size and position of obstacles, and information of other specific objects;

[0061] Step S303, using a Kalman filter or a particle filter algorithm to fuse information from different sensors to obtain the real-time perception information;

[0062] Step S304: Use Kalman filtering or particle filtering algorithms to fuse information from different sensors to obtain the real-time perception information.

[0063] The above method is used to process data from multiple sensors, which can effectively aggregate the data information of each sensor, realize data merging, and prevent data duplication.

[0064] Preferably, the above step S201 dynamically plans the travel path according to the preset global path and the real-time perception information, including the following steps S401-S403:

[0065] Step S401, based on the information of pedestrians, obstacles and other specific objects, the feasible paths are calculated using the A* algorithm, the Dijkstra algorithm or the fast search random tree algorithm, and are scored;

[0066] In this step, when calculating the feasible path, the control system extracts and comprehensively calculates the movement directions of multiple pedestrians in front to determine the overall movement direction of the crowd, and then plans the path along the overall movement direction based on the robot's position and destination, so that the robot's next movement direction is consistent with the overall movement direction. In this way, conflicts between the robot and pedestrians can be effectively avoided.

[0067] Step S402, optimizing the feasible path by using a genetic algorithm or a particle swarm optimization algorithm in combination with a pedestrian following strategy;

[0068] Step S403: using a smoothing algorithm, the feasible path is made smoother to obtain the travel path.

[0069] Preferably, the step S202 of selecting a pedestrian matching the direction of the travel path as a follow-up object and maintaining a safe distance comprises the following steps S501-S503:

[0070] Step S501, based on the moving direction and speed information of each pedestrian ahead, selecting a pedestrian with a high matching degree with the travel path as a follow target;

[0071] Step S502, dynamically adjusting the safety distance according to the speed and direction of the target being followed;

[0072] Step S503: Based on the PID controller, adjust the proportional, integral and differential parameters to maintain stable following of the following target.

[0073] When the followed pedestrian deviates from the travel path or remains stationary for more than a set time, a pedestrian matching the travel path direction is searched again as a new follow object. When there is no suitable follow object, the travel operation is continued based on the method of step S204.

[0074] Preferably, the step S204 of evaluating the possibility of movement and controlling the robot to move in real time when the possibility is higher than a preset threshold value comprises the following steps S601-S602:

[0075] Step S601, calculating the possibility of traveling based on the following formula:

[0076]

[0077] Among them, S is the size of the gap in the detected pedestrian team, V r is the maximum speed of the robot, V p is the average speed of the pedestrian team, T is the expected crossing time, and W is the width of the robot itself plus a safety margin;

[0078] Step S602, determine the size relationship between P and 1, if P>1, execute the horizontal crossing action, otherwise continue to wait for the crossing opportunity.

[0079] By adopting the above method, when performing a crossing action without pedestrians to follow, the robot can maintain an appropriate distance from surrounding pedestrians, thereby ensuring the safety of the robot and the pedestrians.

[0080] In the above steps, the travel path is optimized based on the following method: define the path cost function J = w1L + w2C + w3T, where J is the path cost function, L is the path length, C is the path safety, T is the expected crossing time, and w1, w2, and w3 are weight coefficients. By adjusting these weights, a comprehensive optimal crossing path is found in the emotion where there are no pedestrians to follow.

[0081] Preferably, in the above step S203 of selecting a pedestrian matching the direction of the travel path as a follow-up object, when there are multiple pedestrians that can be followed, the method further includes the following steps S701-S704:

[0082] Step S701, calculate the angle θ between the direction of each pedestrian and the direction of the dynamically planned path i ;

[0083] Step S702, calculate the relative speed v of each pedestrian i and distance d i ;

[0084] Step S703, calculate the comprehensive score S i for:

[0085]

[0086] Among them, w4, w5, w6 are weight coefficients, which respectively represent the importance of directional consistency, relative speed and distance; v max is the maximum speed of the robot, d max is the set maximum following distance;

[0087] Step S704: Select the pedestrian with the highest comprehensive score as the following object.

[0088] A navigation system for an autonomous mobile robot in a dense human flow environment, the navigation system may include or be divided into one or more program modules, one or more program modules are stored in a storage medium and executed by one or more processors to complete the present invention and implement the above-mentioned navigation method. The program module referred to in the embodiment of the present invention refers to a series of computer program instruction segments that can complete specific functions, which is more suitable for describing the execution process of the navigation method in the storage medium than the program itself. The following description will specifically introduce the functions of each program module of this embodiment: the system includes:

[0089] A collection module 801 is used to collect environmental data using a sensor system on the robot;

[0090] An identification module 802 is used to identify pedestrians, obstacles and ground state data based on the collected data to obtain real-time perception information;

[0091] A determination implementation module 803 is used to determine whether it is a dense crowd flow environment based on the real-time perception information, and if not, execute a normal navigation mode, and if yes, execute a crowd flow passing mode;

[0092] The pedestrian flow mode specifically includes the following steps:

[0093] Dynamically plan the travel path based on the preset global path and real-time perception information;

[0094] According to the travel path, determining whether there is a pedestrian whose running direction is consistent with the direction of the travel path, and obtaining a first determination result;

[0095] When the first judgment result is yes, a pedestrian matching the direction of the travel path is selected as a follow object, and a safe distance is maintained;

[0096] When the first judgment result is no, the possibility of walking is evaluated, and when the possibility is higher than a preset threshold, the robot is controlled to move in real time.

[0097] Other contents of implementing the above-mentioned navigation method based on the above-mentioned navigation system have been introduced in detail in the previous embodiments. Please refer to the corresponding contents in the previous embodiments and will not be repeated here.

[0098] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principle of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A navigation method for an autonomous mobile robot in a dense crowd environment, characterized in that: The method comprises: Collect environmental data using the sensor system on the robot; Based on the collected data, pedestrians, obstacles and ground status data are identified to obtain real-time perception information; Based on the real-time perception information, determine whether it is a dense crowd flow environment, if not, execute the normal navigation mode, if yes, execute the crowd flow passing mode; The pedestrian flow mode specifically includes the following steps: Dynamically plan the travel path based on the preset global path and real-time perception information; According to the travel path, determining whether there is a pedestrian whose running direction is consistent with the direction of the travel path, and obtaining a first determination result; When the first judgment result is yes, a pedestrian matching the direction of the travel path is selected as a follow object, and a safe distance is maintained; When the first judgment result is no, the possibility of walking is evaluated, and when the possibility is higher than a preset threshold, the robot is controlled to move in real time.

2. The navigation method of the autonomous mobile robot in a dense crowd environment according to claim 1, characterized in that: Based on the collected data, pedestrians, obstacles, and ground status data are identified to obtain real-time perception information, including: Preprocess the data collected by each sensor, including noise filtering and outlier processing; Extract useful feature information from the processed data, including the location, speed, direction of pedestrians, the size and location of obstacles, and other specific object information; Using Kalman filtering or particle filtering algorithm to fuse information from different sensors to obtain the real-time perception information; The information from different sensors is fused using algorithms such as Kalman filtering or particle filtering to obtain the real-time perception information.

3. The navigation method of the autonomous mobile robot in a dense human flow environment according to claim 1, characterized in that: The dynamically planning of the travel path according to the preset global path and the real-time perception information includes: Based on the information of pedestrians, obstacles and other specific objects, the feasible paths are calculated and scored using the A* algorithm, Dijkstra algorithm or fast search random tree algorithm; In combination with a pedestrian following strategy, a genetic algorithm or a particle swarm optimization algorithm is used to optimize the feasible path; The feasible path is made smoother by a smoothing algorithm to obtain the travel path.

4. The method for navigating an autonomous mobile robot in a dense human flow environment according to claim 1, characterized in that: The selecting a pedestrian whose direction matches the travel path as a follow-up object and maintaining a safe distance includes: Based on the moving direction and speed information of each pedestrian ahead, a pedestrian with a high matching degree with the travel path is selected as a follow target; Dynamically adjust the safety distance according to the speed and direction of the target being followed; Based on the PID controller, the proportional, integral and differential parameters are adjusted to maintain stable following of the following target.

5. A navigation system for an autonomous mobile robot in a dense human flow environment, characterized in that: The system comprises: A collection module, which is used to collect environmental data using a sensor system on the robot; A recognition module, which is used to recognize pedestrians, obstacles and ground state data based on the collected data to obtain real-time perception information; A determination implementation module, which is used to determine whether it is a dense crowd flow environment based on the real-time perception information, if not, execute the normal navigation mode, if yes, execute the crowd flow passing mode; The pedestrian flow mode specifically includes the following steps: Dynamically plan the travel path based on the preset global path and real-time perception information; According to the travel path, determining whether there is a pedestrian whose running direction is consistent with the direction of the travel path, and obtaining a first determination result; When the first judgment result is yes, a pedestrian matching the direction of the travel path is selected as a follow object, and a safe distance is maintained; When the first judgment result is no, the possibility of walking is evaluated, and when the possibility is higher than a preset threshold, the robot is controlled to move in real time.

Images