A mobile intelligent cabinet path planning method and system fusing passenger flow data

Abstract

The application belongs to the technical field of data processing, and particularly relates to a mobile intelligent cabinet path planning method and system fusing passenger flow data, and the method comprises the following steps: based on multi-source sensor data, a local passenger flow field containing flow direction information is constructed by using DeepSort tracking and Kalman filtering; an environmental fluctuation index is calculated by analyzing density mutation and flow direction change, and a self-adaptive damping coefficient is generated in combination with power state; a virtual mass concept is introduced, artificial potential field force is dynamically smoothed by using the self-adaptive damping coefficient, and a smooth driving force is generated to control the motor. The application can effectively suppress high-frequency environmental disturbance, and realize smooth, efficient and energy-saving navigation of the intelligent cabinet.

Description

Technical Field

[0001] This invention relates to the field of data processing technology. More specifically, this invention relates to a method and system for route planning of mobile smart lockers that integrates passenger flow data. Background Technology

[0002] Mobile smart lockers, as an emerging service terminal, are being used in high-dynamic, high-traffic areas such as shopping malls and airports to provide users with convenient following and mobility services. In such complex scenarios, achieving safe, efficient, and smooth autonomous navigation is crucial to ensuring service quality.

[0003] Existing technologies typically utilize sensors such as LiDAR and cameras, employing algorithms like DeepSort for multi-target tracking to detect surrounding pedestrians, and combining this with artificial potential field methods for local obstacle avoidance and path planning. While the DeepSort algorithm can identify and track pedestrian trajectories, traditional navigation strategies often rely solely on discrete, instantaneous obstacle location information to construct environmental models, lacking a holistic consideration of macroscopic pedestrian flow trends and dynamic environmental changes.

[0004] Meanwhile, traditional artificial potential field methods are typically based on quasi-static environment assumptions, directly guiding the robot based on the resultant force of the repulsive force generated by obstacles and the attractive force of the target. In highly dynamic, densely populated environments, the repulsive field often experiences high-frequency and violent fluctuations due to sensor noise and the uncertainty of pedestrian movement. Because traditional algorithms lack a quantitative assessment mechanism for the spatiotemporal instability of the environment and do not consider the robot's own inertia and energy state, directly responding to the resultant force of such violent fluctuations can lead to severe path oscillations, frequent sudden stops, or stationary rotations in mobile smart cabinets, making it difficult to achieve smooth and energy-efficient motion control. Summary of the Invention

[0005] To address the technical problem of poor motion control performance in the aforementioned mobile intelligent cabinets, the present invention provides solutions in the following aspects.

[0006] In a first aspect, the present invention provides a method for route planning of a mobile smart locker that integrates passenger flow data, comprising:

[0007] A local passenger flow field including location, speed, and direction is obtained. Based on the local passenger flow field, an environmental fluctuation index is calculated. The environmental fluctuation index is positively correlated with the difference between the passenger flow density of all grids in the local passenger flow field at the current moment and the passenger flow density at the previous moment, and the cosine of the angle between the local average passenger flow velocity vector and the local average passenger flow velocity vector at the previous moment. It is negatively correlated with the number of grids in the local passenger flow field. Based on the power status of the mobile smart cabinet and the environmental fluctuation index, an adaptive damping coefficient is obtained. The adaptive damping coefficient is positively correlated with the sum of the products of the basic damping constant, the environmental fluctuation index, and the power penalty term. The power penalty term is positively correlated with the power difference at full charge and has an exponential relationship with the power sensitivity coefficient. Based on the adaptive damping coefficient and virtual mass, the artificial potential field force is smoothed to obtain a smooth driving force, and the movement of the mobile smart cabinet is controlled according to the smooth driving force.

[0008] This invention macroscopically perceives the environment by constructing a local passenger flow field that includes location, speed, and direction. It then calculates an adaptive damping coefficient by combining the environmental fluctuation index with its own power status, and introduces virtual mass to dynamically smooth the artificial potential field force. This enables the smart locker to sense the overall flow trend of the environment and proactively increase inertia and reduce sensitivity to high-frequency interference when environmental fluctuations are severe or power is insufficient, thereby effectively suppressing path oscillations and achieving smooth, safe, and energy-efficient autonomous navigation in highly dynamic pedestrian flow environments.

[0009] Preferably, obtaining the local passenger flow field including location, speed, and direction includes:

[0010] The mobile smart cabinet is equipped with LiDAR, depth camera and cloud interface to acquire point cloud and image information of the surrounding environment in real time.

[0011] A local sensing range is defined, which is the area that moves in real time along with the movement of the mobile smart cabinet, with the geometric center of the mobile smart cabinet as the origin.

[0012] The environment within the local perception range is divided into a grid map;

[0013] The DeepSort multi-object tracking algorithm is used to identify static obstacles and dynamic pedestrians in the environment. Based on the tracking results, the real-time passenger flow density distribution and average flow direction data of the area are statistically analyzed.

[0014] The raw data is cleaned by Kalman filtering and spatiotemporally aligned to construct a local passenger flow field that includes information on location, velocity magnitude, and flow direction.

[0015] This invention acquires data using lidar and depth cameras, sets a dynamic local sensing range and rasterizes the environment, and constructs a continuous local passenger flow field using Kalman filtering and spatiotemporal alignment. This can filter out sensor noise and transform discrete pedestrian individuals into a continuous field description containing flow direction information, providing an accurate and macroscopic data foundation for subsequent analysis of environmental instability and solving the problem that instantaneous location cannot reflect the trend of crowd flow.

[0016] Preferably, the DeepSort multi-target tracking algorithm uses the MobileNetV2 network structure as a feature extractor to map the detected pedestrian image blocks into feature vectors and perform L2 normalization processing; it determines the same target by calculating the feature distance and combining it with the maximum cosine distance threshold.

[0017] Preferably, the environmental fluctuation index satisfies the expression:

[0018] ;

[0019] In the formula, Indicates the first The environmental fluctuation index at any given time; Indicates the number of sampling grids within the local sensing range; , Indicates the first The grid in the first Time, Number Passenger flow density at any given time; This represents the preset maximum passenger flow density constant; , Indicates the first Time, Number The local average passenger flow velocity vector at time t; express and The cosine of the angle between them; , This represents the density change weighting coefficient and the direction change weighting coefficient.

[0020] This invention integrates the mutation rate of density with the difference in flow direction, and gives higher weight to changes in flow direction. It can capture the spatiotemporal instability in complex situations such as intersections or reversed pedestrian flow, measure the degree of environmental disorder, and provide the control system with a more valuable dynamic adjustment basis than density alone, thereby more accurately predicting potential congestion and collision risks.

[0021] Preferably, the adaptive damping coefficient satisfies the expression:

[0022] ;

[0023] In the formula, Indicates the first The adaptive damping coefficient at time t; Represents the basic damping constant; Indicates the mobile smart cabinet The remaining battery power at any given time; This indicates the full charge level of the mobile smart cabinet; This refers to the power sensitivity coefficient. Indicates the first The environmental fluctuation index at any given time; Represents the natural logarithm function; This represents the minimum value function.

[0024] This invention constructs an adaptive damping coefficient formula that integrates environmental fluctuations and remaining battery power. It smooths the impact of environmental fluctuations using a natural logarithmic function and constructs a battery power penalty term using a nonlinear power function. This allows the system to remain sensitive when battery power is sufficient and the environment is stable, while forcing it into a high-damping state when the environment is harsh or battery power is critically low. This dual adjustment mechanism ensures obstacle avoidance safety while maximizing the battery life of the smart cabinet.

[0025] Preferably, the artificial potential field force satisfies the expression:

[0026] ;

[0027] In the formula, Indicates the first The primal combined force of moments; Indicates the gravitational gain coefficient; Indicates the target location coordinates of the mobile smart cabinet; Indicates the mobile smart cabinet Position vector at any given time; This represents the total number of obstacles at time t; Indicates the repulsive force gain coefficient; Let represent the distance vector between the mobile smart cabinet and the j-th obstacle at time t; Let represent the unit direction vector from the j-th obstacle to the mobile smart cabinet at time t; Symbol for modulus calculation; It represents a tiny positive value.

[0028] This invention combines target attraction with total environmental repulsion based on the inverse square law, so that the smart cabinet is mainly attracted by the target when it is far away from the obstacle, while the repulsion increases exponentially when it is close to the obstacle. This ensures the safety of basic obstacle avoidance and provides the original perception driving force that can provide real-time feedback on environmental changes for subsequent smooth processing.

[0029] Preferably, the smooth driving force satisfies the expression:

[0030] ;

[0031] In the formula, This represents the smooth driving force at time t; This represents the adaptive damping coefficient at time t; Let represent the original net force at time t; Represents the virtual mass constant; Let represent the velocity vector of the mobile intelligent cabinet at time t-1; This indicates the time interval between adjacent moments.

[0032] Preferably, controlling the movement of the mobile intelligent cabinet according to the smooth driving force specifically includes:

[0033] The smooth driving force is converted into acceleration commands according to Newton's second law, and then integrated to obtain linear velocity and angular velocity commands.

[0034] The command is sent to the underlying motion controller of the mobile smart cabinet. Based on the kinematic model of the mobile smart cabinet, the target speed command of each hub motor is calculated, and the hub motor is driven to perform the action.

[0035] Preferably, the density change weighting coefficient The weighting factor for directional changes is 0.4. It is 0.6.

[0036] Secondly, the present invention provides a mobile smart locker path planning system that integrates passenger flow data, including a processor and a memory. The memory stores computer program instructions, and when the computer program instructions are executed by the processor, the above-mentioned mobile smart locker path planning method integrating passenger flow data is implemented.

[0037] By adopting the above technical solution, a mobile smart cabinet path planning method that integrates passenger flow data is generated into a computer program and stored in a memory for loading and execution by a processor. This allows for the creation of terminal devices based on the memory and processor, making them convenient to use.

[0038] The beneficial effects of this invention are as follows: This invention constructs a passenger flow field including flow direction using DeepSort and Kalman filtering, calculates the spatiotemporal instability of the environment as a fluctuation index, calculates an adaptive damping coefficient in conjunction with the battery status, and introduces the concept of virtual mass to dynamically adjust the weights of perception-driven and inertia-maintaining mechanisms. This mechanism enables the smart locker to proactively increase inertia and filter high-frequency interference when passenger flow is turbulent or battery levels are low, thus avoiding path oscillation problems in highly dynamic environments. Attached Figure Description

[0039] Figure 1This is a flowchart illustrating a mobile smart locker path planning method that integrates passenger flow data according to the present invention.

[0040] Figure 2 This is a schematic diagram illustrating the perception of a local passenger flow field. Detailed Implementation

[0041] This invention discloses a method for path planning of mobile smart lockers that integrates passenger flow data, referring to... Figure 1 This includes steps S1-S5:

[0042] S1: Based on multi-source sensor data, DeepSort multi-target tracking and Kalman filtering are performed to obtain local passenger flow field.

[0043] It should be noted that mobile smart lockers typically operate in highly dynamic, densely populated environments such as shopping malls and airports. In such scenarios, relying solely on the instantaneous location of obstacles for obstacle avoidance has significant limitations, as the instantaneous obstacle location cannot reflect the overall flow trend of the surrounding crowd. This can easily lead to the mobile smart locker getting surrounded by people or moving against the mainstream flow, resulting in frequent sudden stops, stationary turns, and detours, severely impacting operational efficiency and safety. Therefore, this invention perceives the environment from a macroscopic perspective, transforming discrete individual pedestrians into a continuous local passenger flow field, obtaining a comprehensive environmental description including location, speed, and flow direction.

[0044] Specifically, the mobile smart cabinet uses LiDAR, depth cameras, and cloud interfaces to acquire point cloud and image information of the surrounding environment in real time.

[0045] It should be noted that, in order to define the boundaries of data processing and ensure the real-time performance of computation, it is first necessary to set the local sensing range.

[0046] The local sensing range is defined as the area covering the surrounding physical space, with the geometric center of the mobile smart cabinet as the origin. This local sensing range moves in real-time along with the movement of the mobile smart cabinet. It should be noted that the size of the local sensing range depends on the effective detection range of the mounted LiDAR. In this embodiment, the local sensing range is defined as a circular area with a radius of R, where R is 10 meters in an exemplary manner.

[0047] Preferably, the environment within the local sensing range is divided into sections with a resolution of [resolution value missing]. A raster map. The choice of resolution requires a balance between accuracy and computational power: if the raster is too large, for example, exceeding... If the grid is too small, it cannot accurately describe the tiny gaps between pedestrians in a dense crowd; if the grid is too small, for example, smaller than... The computational load increases exponentially, which cannot meet the real-time requirements of mobile embedded systems.

[0048] The DeepSort multi-object tracking algorithm is used to identify static obstacles and dynamic pedestrians in the environment. The DeepSort algorithm employs Kalman filtering for state prediction and the Hungarian algorithm for data association. In this embodiment, considering the computational resource limitations of the mobile smart cabinet, the lightweight MobileNetV2 network structure is preferably used as the feature extractor for the convolutional neural network used to extract appearance features in the DeepSort multi-object tracking algorithm. MobileNetV2 maps the detected pedestrian image patches to... 1. Dimensional feature vectors, and perform eigenvector processing on the feature vectors. Normalization process.

[0049] For example, to maintain tracking stability in dense crowds, a maximum cosine distance threshold is set to... When the calculated feature distance is less than the threshold, they are determined to be the same target; the maximum lifetime of the trajectory is set to... Frame, that is, when the target disappears for more than [time period missing] The target's trajectory information is deleted only after a few seconds; the minimum number of hits required to confirm the trajectory is set to [value missing]. Secondly, it filters out occasional false detections caused by sensor noise.

[0050] Based on the tracking results, real-time passenger flow density distribution and average flow direction data within the statistical area are analyzed. The raw data is cleaned using Kalman filtering to remove high-frequency fluctuations and then spatiotemporally aligned to construct a local passenger flow field containing location, speed, and flow direction information. It should be noted that, as... Figure 2 This is a schematic diagram of local passenger flow field perception, showing the passenger flow density distribution and average flow direction vector within the local perception range.

[0051] At this point, a local passenger flow field containing location, speed, and direction has been obtained.

[0052] S2: Based on the local passenger flow field, perform density mutation and flow direction change analysis to obtain the environmental fluctuation index.

[0053] It should be noted that in areas such as intersections, promotional areas, or boarding gates, passenger density and flow direction often fluctuate drastically within a short period. Traditional path planning algorithms typically assume a quasi-static environment, directly responding to the current repulsive force and ignoring the continuity of environmental changes and noise interference. This short-term response mechanism can cause mobile smart lockers to experience severe path oscillations when faced with chaotic crowds. Therefore, this invention constructs an environmental fluctuation index describing the spatiotemporal instability of the environment, serving as the basis for subsequent adjustment and control strategies.

[0054] Specifically, based on local passenger flow patterns, analysis of density abrupt changes and the degree of flow direction changes is performed to obtain an environmental fluctuation index, including:

[0055] It should be noted that, in order to comprehensively measure environmental instability, this invention decomposes environmental fluctuations into changes in two core physical dimensions: first, the abrupt change in density, i.e., the drastic increase or decrease in the number of people per unit time and per unit area, which reflects rapid changes in environmental congestion and foreshadows potential congestion risks; second, the degree of change in flow direction, i.e., the consistency of the direction of crowd movement. When the flow direction changes drastically between two points in time, the uncertainty of the environment increases significantly. Therefore, this invention uses normalized density differences to calculate density abrupt changes and the cosine distance between the angles of the velocity vectors between two points in time to calculate the degree of change in flow direction. Considering that in obstacle avoidance scenarios for mobile robots, a sudden reversal of the flow direction is more disruptive to path planning than a simple increase or decrease in density, this invention gives a higher weight to the degree of change in flow direction, and finally constructs a dimensionless comprehensive evaluation index through weighted summation.

[0056] Preferably, the environmental fluctuation index satisfies the expression:

[0057] ;

[0058] In the formula, Indicates the first The environmental fluctuation index at any given time; Indicates the number of sampling grids within the local sensing range; , Indicates the first The grid in the first Time, Number Passenger flow density at any given time; This represents the preset maximum passenger flow density constant; , Indicates the first Time, Number The local average passenger flow velocity vector at time t; express and The cosine of the angle between them; , This represents the density change weighting coefficient and the direction change weighting coefficient. For example, setting... , To enhance the system's sensitivity to sudden changes in flow direction, the following settings are implemented. for people .

[0059] In the formula, Indicates the first The normalized rate of change of passenger flow density in the i-th grid is the value of the grid. A larger value indicates a greater change in passenger flow density in the i-th grid, which in turn affects the rate of change in passenger flow density in the ith grid. The greater the contribution of the environmental fluctuation index at any given moment; This represents a directional difference measure based on cosine similarity, and its value range is... When the flow direction remains consistent, this item is This term reaches its maximum value when the flow direction reverses. This allows them to keenly detect directional disturbances; This indicates the degree of spatiotemporal instability of the i-th grid within the current control cycle. The larger this value, the greater the environmental fluctuation index of the i-th grid at time t. This represents the average environmental fluctuation index of all grids at time t, reflecting the overall environmental fluctuation within the local sensing range at time t.

[0060] Thus, the environmental fluctuation index, which represents the dynamic instability of the environment, was obtained.

[0061] S3: Based on the power status of the mobile smart cabinet and the environmental fluctuation index, perform correlation calculations to obtain the adaptive damping coefficient.

[0062] It should be noted that the response strategy of the mobile smart cabinet should also take into account its own power status. To suppress path oscillations, virtual damping needs to be introduced into the control system to make the mobile smart cabinet less sensitive to high-frequency disturbances. Especially when the power is low, frequent acceleration, deceleration, and turning will accelerate power depletion and shorten service time. Therefore, this invention constructs an adaptive damping coefficient that integrates environmental characteristics and its own power level. In harsh environments or when the power is critical, it actively increases the system's inertia, sacrificing a slight obstacle avoidance sensitivity in exchange for smooth movement and energy efficiency.

[0063] Specifically, based on the power status of the mobile smart cabinet and the environmental fluctuation index, a correlation calculation is performed to obtain the adaptive damping coefficient, including:

[0064] It's important to note that the core of the adaptive damping coefficient design lies in establishing a dual penalty mechanism for both the environment and energy. First, a base damping mechanism needs to be set to ensure the system has minimal smoothness under any ideal operating conditions, preventing control divergence. Second, for environmental factors, a natural logarithmic function is used to smooth and compress the environmental fluctuation index, making the adaptive damping coefficient sensitive to drastic fluctuations without overreacting. Finally, for energy factors, a nonlinear power function is used to construct energy sensitivity. As the power consumption decreases, the system's tolerance for energy consumption should decrease rapidly and nonlinearly, forcing the system to quickly enter a high-damping energy-saving mode. Therefore, the expression for the adaptive damping coefficient is established.

[0065] Preferably, the adaptive damping coefficient satisfies the following expression:

[0066] ;

[0067] In the formula, Indicates the first The adaptive damping coefficient at time t; Represents the basic damping constant; Indicates the mobile smart cabinet The remaining battery power at any given time; This indicates the full charge level of the mobile smart cabinet; This refers to the power sensitivity coefficient. Indicates the first Environmental fluctuation index at any given time; Represents the natural logarithm function; This represents the minimum value function. For example, the basic damping constant is... The power sensitivity coefficient is 2.

[0068] In the formula, This constitutes an energy penalty term, which approaches a certain value when the remaining charge is close to full. The damping coefficient is mainly determined by environmental fluctuations; as the power decreases, the energy penalty term gradually increases, meaning that in low-power mode, even with small environmental fluctuations, the system will forcibly maintain a high adaptive damping coefficient, thereby limiting the motor's violent movements; through The adaptive damping coefficient is limited to a value less than or equal to 1.

[0069] Thus, the adaptive damping coefficient that integrates the environment and its own state was obtained.

[0070] S4: Based on the adaptive damping coefficient and virtual mass, the artificial potential field force is smoothed to obtain a smooth driving force.

[0071] It should be noted that traditional artificial potential field methods directly utilize the resultant force of attraction and repulsion as the control command. When the repulsive force field fluctuates drastically due to environmental noise or dynamic pedestrian interference, the direction of the resultant force will also change accordingly, causing oscillations in the movement of the mobile smart cabinet. To fundamentally eliminate path oscillations, the original resultant force cannot be used directly; instead, an adaptive damping coefficient should be used to smooth the resultant force. This invention introduces the concept of virtual mass, transforming the adaptive damping coefficient into a specific dynamic constraint. Through the principle of inertia in physics, high-frequency jitter is smoothed, generating a final smooth driving force.

[0072] Specifically, based on the adaptive damping coefficient and virtual mass, the artificial potential field force is smoothed to obtain a smooth driving force, including:

[0073] It should be noted that the resultant force calculated using the traditional artificial potential field method based on current sensor data can provide real-time feedback on environmental changes. Therefore, this invention utilizes this resultant force as a sensing drive to ensure the system's sensitivity to environmental changes. Considering that the motion state should have the ability to resist sudden changes in order to eliminate oscillations, and the inertial force that maintains the motion state at the previous moment can well reflect this resistance capability, this invention combines the sensing drive force representing environmental sensitivity and the inertial drive force representing motion continuity, using an adaptive damping coefficient as an adjustment weight to establish a dynamic smoothing model. This allows for seamless switching between sensing drive and inertial drive when the adaptive damping coefficient changes.

[0074] The two-dimensional coordinates of the mobile smart cabinet in the global coordinate system are obtained, forming its position vector. It should be noted that the coordinates of the mobile smart cabinet can be calculated in real time using the LiDAR SLAM positioning module mounted on the cabinet.

[0075] Obtain the target location coordinates. It should be noted that the target location coordinates are known coordinates. For example, they may be issued by the upper-level global path planning system of the mobile smart cabinet, or the target point location vector may be formed by the task endpoint coordinates specified by the user.

[0076] Obstacle information acquisition: Based on the output of the DeepSort multi-target tracking algorithm, identify all obstacles within the local perception range at the current moment, and calculate the distance vector and direction vector of each obstacle relative to the smart cabinet.

[0077] Preferably, the original resultant force satisfies the expression:

[0078] ;

[0079] In the formula, Indicates the first The primal combined force of moments; Indicates the gravitational gain coefficient; Indicates the target location coordinates of the mobile smart cabinet; Indicates the mobile smart cabinet Position vector at any given time; Indicates the repulsive force gain coefficient; Let represent the distance vector between the mobile smart cabinet and the j-th obstacle at time t; Let represent the unit direction vector from the j-th obstacle to the mobile smart cabinet at time t; Symbol for modulus calculation; This represents a small positive value to avoid a denominator of 0, for example. It should be noted that, and These are preset constants stored in the mobile smart cabinet navigation system, used to adjust the force weights of the smart cabinet towards the target and away from obstacles. For example, to ensure obstacle avoidance safety is prioritized, the gravity gain coefficient is set to 2 and the repulsion gain coefficient is set to 15.

[0080] In the formula, This represents the position deviation vector of the smart cabinet's current position relative to the target point; This represents the gravitational driving term, whose direction always points to the target point and whose magnitude is proportional to the distance, used to pull the smart cabinet to the endpoint. This represents the unit repulsive force intensity based on the inverse square law, meaning that the closer the distance, the more exponentially the repulsive force increases. This represents the individual repulsive force generated by the j-th obstacle alone; This represents the total environmental repulsive force obtained by vector superposition of the individual repulsive forces of all obstacles detected within the local sensing range; It represents the resultant force of the original artificial potential field, which combines the attraction of the target and the total repulsion of the environment.

[0081] Preferably, the smooth driving force satisfies the expression:

[0082] ;

[0083] In the formula, Indicates the first The smooth driving force of time; Indicates the first The adaptive damping coefficient at time t; Indicates the first The primal combined force of moments; Represents the virtual mass constant; Indicates the mobile smart cabinet The velocity vector at any given moment; This represents the time interval between adjacent moments. For example, It weighs 10 kilograms.

[0084] In the formula, Indicates based on The equivalent acceleration calculated from the state of motion at any given moment to maintain the current trend of motion. It represents the inertial driving force, which is the virtual internal maintaining force generated by the system in order to maintain its original motion trend and resist external disturbances; This represents a dynamic weighted fusion of perception-driven and inertial-driven forces. The formula indicates that the smooth driving force is jointly determined by environmental perception and motion inertia; when the environment is open and the battery is sufficiently charged... The force is relatively small, and the smooth driving force is mainly dominated by the original resultant force, exhibiting high sensitivity; however, when environmental fluctuations are severe or the battery is low... As the system increases, it retains only a small amount of current perception, with most of the driving force coming from the inertial force that maintains the original state of motion. This allows it to ignore the chaotic and rapidly changing repulsive forces around it and smoothly transition along the original trajectory.

[0085] Thus, a smooth driving force capable of suppressing oscillations was obtained.

[0086] S5: Based on smooth driving force, it performs kinematic calculations and motor control to obtain the smooth motion trajectory of the mobile smart cabinet.

[0087] Specifically, the smooth driving force is converted into acceleration commands according to Newton's second law, and then integrated to obtain linear velocity and angular velocity commands, which are sent to the underlying motion controller of the mobile smart cabinet via the CAN bus. The underlying motion controller, based on the kinematic model of the mobile smart cabinet (e.g., a differential drive model or a Mecanum wheel omnidirectional drive model), calculates the target speed commands for each hub motor, drives each hub motor to perform actions, and completes the smooth movement from the current position to the dynamic passenger flow target point.

[0088] This completes the smooth path planning and motion control of the mobile intelligent cabinet in dynamic and complex environments.

[0089] This invention also discloses a mobile smart locker route planning system that integrates passenger flow data, including a processor and a memory. The memory stores computer program instructions, and when the computer program instructions are executed by the processor, a mobile smart locker route planning method integrating passenger flow data according to the present invention is implemented.

[0090] The system also includes other components well known to those skilled in the art, such as communication buses and communication interfaces, the settings and functions of which are known in the art and will not be described in detail here.

[0091] While this specification has shown and described numerous embodiments of the invention, it will be apparent to those skilled in the art that such embodiments are provided by way of example only. Many modifications, alterations, and alternatives will occur to those skilled in the art without departing from the spirit and essence of the invention. It should be understood that various alternatives to the embodiments of the invention described herein may be employed in the practice of this invention.

Claims

1. A method for route planning of mobile smart lockers that integrates passenger flow data, characterized in that, include: Obtain local passenger flow fields including location, speed, and direction; Based on the local passenger flow field, the environmental fluctuation index is calculated. The environmental fluctuation index is positively correlated with the difference between the passenger flow density of all grids in the local passenger flow field at the current time and the passenger flow density at the previous time, and the cosine of the angle between the local average passenger flow velocity vector and the local average passenger flow velocity vector at the previous time. It is negatively correlated with the number of grids in the local passenger flow field. Based on the power status and environmental fluctuation index of the mobile smart cabinet, an adaptive damping coefficient is obtained; the adaptive damping coefficient is positively correlated with the sum of the products of the basic damping constant, the environmental fluctuation index, and the power penalty term; the power penalty term is positively correlated with the power difference under full charge and has an exponential relationship with the power sensitivity coefficient. Based on the adaptive damping coefficient and virtual mass, the artificial potential field force is smoothed to obtain a smooth driving force, and the movement of the mobile intelligent cabinet is controlled according to the smooth driving force. The environmental fluctuation index satisfies the following expression: In the formula, Indicates the first The environmental fluctuation index at any given time; Indicates the number of sampling grids within the local sensing range; , Indicates the first The grid in the first Time, Number Passenger flow density at any given time; This represents the preset maximum passenger flow density constant; , Indicates the first Time, Number The local average passenger flow velocity vector at time t; express and The cosine of the angle between them; , This represents the weighting coefficients for density changes and direction changes; The adaptive damping coefficient satisfies the following expression: In the formula, Indicates the first The adaptive damping coefficient at time t; Represents the basic damping constant; Indicates the mobile smart cabinet The remaining battery power at any given time; This indicates the full charge level of the mobile smart cabinet; This refers to the power sensitivity coefficient. Indicates the first The environmental fluctuation index at any given time; Represents the natural logarithm function; This represents the minimum value function.

2. The method for route planning of a mobile smart locker integrating passenger flow data according to claim 1, characterized in that, The acquisition of the local passenger flow field, including location, speed, and direction, includes: The mobile smart cabinet is equipped with LiDAR, depth camera and cloud interface to acquire point cloud and image information of the surrounding environment in real time. A local sensing range is defined, which is the area that moves in real time along with the movement of the mobile smart cabinet, with the geometric center of the mobile smart cabinet as the origin. The environment within the local perception range is divided into a grid map; The DeepSort multi-object tracking algorithm is used to identify static obstacles and dynamic pedestrians in the environment. Based on the tracking results, the real-time passenger flow density distribution and average flow direction data of the area are statistically analyzed. The raw data is cleaned by Kalman filtering and spatiotemporally aligned to construct a local passenger flow field that includes information on location, velocity magnitude, and flow direction.

3. The method for route planning of a mobile smart locker integrating passenger flow data according to claim 2, characterized in that, The DeepSort multi-target tracking algorithm uses the MobileNetV2 network structure as a feature extractor, maps the detected pedestrian image blocks into feature vectors and performs L2 normalization processing; it determines the same target by calculating the feature distance and combining it with the maximum cosine distance threshold.

4. The method for route planning of a mobile smart locker integrating passenger flow data according to claim 1, characterized in that, The artificial potential field force satisfies the expression: ； In the formula, Indicates the first The primal combined force of moments; Indicates the gravitational gain coefficient; Indicates the target location coordinates of the mobile smart cabinet; Indicates the mobile smart cabinet Position vector at any given time; This represents the total number of obstacles at time t; Indicates the repulsive force gain coefficient; Let represent the distance vector between the mobile smart cabinet and the j-th obstacle at time t; Let represent the unit direction vector from the j-th obstacle to the mobile smart cabinet at time t; Symbol for modulus calculation; It represents a tiny positive value.

5. The method for route planning of a mobile smart locker integrating passenger flow data according to claim 4, characterized in that, The smooth driving force satisfies the expression: ； In the formula, Indicates the first The smooth driving force of time; Indicates the first The adaptive damping coefficient at time t; Indicates the first The primal combined force of moments; Represents the virtual mass constant; Indicates the mobile smart cabinet The velocity vector at any given moment; This indicates the time interval between adjacent moments.

6. The method for route planning of a mobile smart locker integrating passenger flow data according to claim 1, characterized in that, The control of the movement of the mobile intelligent cabinet based on the smooth driving force specifically includes: The smooth driving force is converted into acceleration commands according to Newton's second law, and then integrated to obtain linear velocity and angular velocity commands. The command is sent to the underlying motion controller of the mobile smart cabinet. Based on the kinematic model of the mobile smart cabinet, the target speed command of each hub motor is calculated, and the hub motor is driven to perform the action.

7. The method for route planning of a mobile smart locker integrating passenger flow data according to claim 1, characterized in that, The density change weighting coefficient The weighting factor for directional changes is 0.

4. It is 0.

6.

8. A mobile intelligent locker route planning system integrating passenger flow data, characterized in that, include: A processor and a memory, wherein the memory stores computer program instructions that, when executed by the processor, implement a mobile smart locker path planning method that integrates passenger flow data according to any one of claims 1-7.

Images