A logistics trolley navigation control method, system, device and medium

Abstract

A logistics trolley navigation control method, system, device and medium, wherein the method comprises: obtaining position data of the logistics trolley at the current position, task parameters, environment data corresponding to the position data, and historical environment evolution data of the workshop where the logistics trolley is located; constructing a navigation intention vector of the logistics trolley according to the task parameters; constructing a navigation behavior graph based on the navigation intention vector and the historical environment evolution data; mapping the position data of the logistics trolley to the navigation behavior graph to obtain graph context information of the target navigation behavior node corresponding to the current position; generating a first navigation behavior sequence according to the historical passable passage, the historical passing success rate and the environment data; adjusting the initial control parameters based on the passage weight in the navigation behavior graph, the environment data and the behavior risk level to obtain a second navigation behavior sequence; and performing navigation control on the logistics trolley based on the second navigation behavior sequence. The present application can improve the navigation efficiency of the logistics trolley.

Description

Technical Field

[0001] This application relates to the field of navigation control technology, specifically to a navigation control method, system, device, and medium for a logistics vehicle. Background Technology

[0002] With the rapid development of smart manufacturing, logistics vehicles are being used more and more widely in industrial settings. These vehicles can autonomously complete material transportation tasks, improving production efficiency and have become an indispensable part of smart factories.

[0003] In related technologies, navigation control for logistics vehicles primarily employs preset path planning. Specifically, the system creates a static navigation map based on the factory layout and plans fixed navigation paths on the map. When performing tasks, the logistics vehicle obtains its own location information through a positioning system and then travels along the preset path. To ensure navigation safety, the system also collects environmental data in real time for obstacle avoidance control.

[0004] However, the environmental conditions in industrial settings are often highly dynamic and uncertain. For example, factors such as production line layout adjustments, equipment relocation, and personnel movement can all cause changes in the environment. This dynamic and changing environment makes navigation methods based on static path planning difficult to adapt to complex and ever-changing real-world scenarios, reducing the navigation efficiency of logistics vehicles. Summary of the Invention

[0005] This application provides a navigation control method, system, device, and medium for logistics vehicles, which can improve the navigation efficiency of logistics vehicles.

[0006] The first aspect of this application provides a navigation control method for a logistics vehicle, applied to a server. The method includes: acquiring location data of the logistics vehicle at its current location, task parameters, environmental data corresponding to the location data, and historical environmental evolution data of the workshop where the logistics vehicle is located; constructing a navigation intent vector for the logistics vehicle based on the task parameters; constructing a navigation behavior graph based on the navigation intent vector and historical environmental evolution data, the navigation behavior graph including navigation behavior nodes, behavior paths, and path weights of the behavior paths; mapping the location data of the logistics vehicle to the navigation behavior graph to obtain graph context information of the target navigation behavior node corresponding to the current location, the graph context information including historical traversable paths, historical traversal success rates, and behavior risk levels of the target navigation behavior node; generating a first navigation behavior sequence based on historical traversable paths, historical traversal success rates, and environmental data, the first navigation behavior sequence containing multiple navigation control actions, each navigation control action corresponding to a set of initial control parameters; calculating adjustment changes based on the path weights, environmental data, and behavior risk levels in the navigation behavior graph, and adjusting the initial control parameters according to the adjustment changes to obtain a second navigation behavior sequence; and performing navigation control on the logistics vehicle based on the second navigation behavior sequence.

[0007] Optionally, constructing a navigation behavior map based on the navigation intent vector and the historical environmental evolution data specifically includes: determining environmental change events in the workshop based on the historical environmental evolution data; determining navigation behavior nodes based on the environmental change events, and determining the environmental feature vector of each navigation behavior node, wherein the navigation behavior nodes include main navigation behavior nodes and transitional navigation behavior nodes; analyzing the environmental feature vector in conjunction with the navigation intent vector, calculating the environmental stability value of each navigation behavior node within each first preset time period, and determining the first preset time period in which the environmental stability value is greater than a preset environmental stability threshold and satisfies the constraint condition of the navigation intent vector as the target passage period; and determining the path weight and the behavior path based on the target passage period and the environmental change events.

[0008] Optionally, the environmental feature vector includes a first feature sub-vector and a second feature sub-vector. The step of determining the navigation behavior node based on the environmental change event and determining the environmental feature vector of each navigation behavior node specifically includes: marking the location where the environmental change event occurs as a primary navigation behavior node; setting multiple transitional navigation behavior nodes between adjacent primary navigation behavior nodes based on a preset spatial division rule, wherein the navigation behavior nodes include multiple transitional navigation behavior nodes between the primary navigation behavior node and adjacent primary navigation behavior nodes; dividing the historical environmental evolution data according to a first preset time period, and aggregating the historical environmental evolution data for each first preset time period based on the spatial location of the navigation behavior node; extracting features from the aggregated historical environmental evolution data to obtain a first feature sub-vector of the navigation behavior node within each first preset time period; analyzing the fluctuation amplitude of the first feature sub-vector within a second preset time period to obtain a second feature sub-vector, wherein the second preset time period consists of multiple consecutive first preset time periods; and concatenating the first feature sub-vector with the first feature sub-vector to obtain the environmental feature vector.

[0009] Optionally, determining the path weight and the behavioral path based on the target travel time period and the environmental change event specifically includes: when there is a passable physical path between the target travel time periods of adjacent target navigation behavior nodes in physical space, a candidate behavioral path is established between the target navigation behavior nodes, wherein the target navigation behavior nodes are the navigation behavior nodes whose environmental change events meet preset conditions; the second feature sub-vectors of all target navigation behavior nodes in the candidate behavioral paths within the overlapping recommended travel time period are concatenated to obtain a target second feature sub-vector; the feature indicators in the target second feature sub-vector are weighted to obtain the travel difficulty of each candidate behavioral path, the reciprocal of the travel difficulty is determined as the path weight, and the candidate behavioral paths whose travel difficulty is less than a preset travel difficulty threshold are determined as the behavioral paths.

[0010] Optionally, a first navigation behavior sequence is generated based on the historical passable routes, the historical passability success rate, and the environmental data. Specifically, this includes: determining the current passability status of the historical passable routes based on the environmental data; identifying passable routes where the historical passability success rate is greater than a preset success rate threshold and the current passability status meets preset passability conditions as target passable routes; determining target navigation control points for the target passable routes, the target navigation control points including turning points, obstacle avoidance points, and speed change points; and dividing the target passable routes into multiple passable sub-segments based on the target navigation control points; and for each... The navigation control actions for each traffic sub-segment are determined based on its starting and ending positions and the target navigation control point corresponding to the starting position. These actions include straight-line driving, obstacle avoidance, and speed adjustment. Initial control parameters are determined based on the length and width of the traffic sub-segment corresponding to each navigation control action and the positional distribution feature vector of the target navigation control point in the target traffic path. The navigation control actions and the initial control parameters corresponding to them are connected according to the temporal relationship of the traffic sub-segments to obtain the first navigation behavior sequence.

[0011] Optionally, the adjustment change amount is calculated based on the path weights, environmental data, and behavioral risk levels in the navigation behavior map, and the initial control parameters are adjusted according to the adjustment change amount to obtain a second navigation behavior sequence. Specifically, this includes: analyzing the navigation stability feature vector of each navigation control action based on the path weights, environmental data, and behavioral risk levels in the navigation behavior map; inputting the navigation stability feature vector into a preset adjustment function to calculate the adjustment change amount, and adjusting the initial control parameters corresponding to each navigation control action according to the navigation stability feature vector to obtain a second navigation behavior sequence.

[0012] Optionally, the navigation stability feature vector of each navigation control action is analyzed based on the path weights, environmental data, and behavioral risk levels in the navigation behavior map. Specifically, this includes: determining the risk level scores of the environmental data across multiple preset dimensions based on preset rules; combining the risk level scores into a traffic environment feature vector, where the environmental data is the environmental data of the traffic sub-segment corresponding to the navigation control action; performing a matching analysis between the traffic environment feature vector and the behavioral risk level to obtain a risk impact factor for each navigation control action, where the risk impact factor characterizes the degree of influence of the behavioral risk level on the navigation control action; and performing a weighted calculation on the path weights and the risk impact factor to obtain a navigation stability feature vector for each navigation control action.

[0013] A second aspect of this application provides a navigation control system for a logistics vehicle, comprising: an acquisition module for acquiring location data of the logistics vehicle at its current location, task parameters, environmental data corresponding to the location data, and historical environmental evolution data of the workshop where the logistics vehicle is located; a first construction module for constructing a navigation intent vector of the logistics vehicle based on the task parameters; a second construction module for constructing a navigation behavior graph based on the navigation intent vector and historical environmental evolution data, the navigation behavior graph including navigation behavior nodes, behavior paths, and path weights of the behavior paths; and a mapping module for mapping the location data of the logistics vehicle to the navigation behavior graph to obtain the target navigation behavior node corresponding to the current location. The system includes a navigation behavior graph context information, which includes the historical passable paths, historical success rates, and behavior risk levels of the target navigation behavior nodes; a generation module, which generates a first navigation behavior sequence based on the historical passable paths, historical success rates, and environmental data. The first navigation behavior sequence contains multiple navigation control actions, each corresponding to a set of initial control parameters; an adjustment module, which calculates the adjustment change based on the path weights, environmental data, and behavior risk levels in the navigation behavior graph, and adjusts the initial control parameters according to the adjustment change to obtain a second navigation behavior sequence; and a control module, which performs navigation control on the logistics vehicle based on the second navigation behavior sequence.

[0014] A third aspect of this application provides an electronic device including a processor, a memory, a user interface, and a network interface. The memory is used to store instructions, the user interface and the network interface are both used to communicate with other devices, and the processor is used to execute the instructions stored in the memory to cause the electronic device to perform any of the methods described above.

[0015] In a fourth aspect, this application provides a computer-readable storage medium storing instructions that, when executed, perform the method described in any of the foregoing descriptions.

[0016] In summary, one or more technical solutions provided in this application have at least the following technical effects or advantages:

[0017] 1. By acquiring the location data, task parameters, environmental data, and historical environmental evolution data of the workshop where the logistics vehicle is located, a navigation intent vector is constructed based on the task parameters. Then, combined with the historical environmental evolution data, a navigation behavior graph containing navigation behavior nodes, behavior paths, and path weights is constructed, thereby establishing a deep understanding of the dynamic changes in the industrial environment. By mapping the location data of the logistics vehicle to the navigation behavior graph, the graph context information of the target navigation behavior node corresponding to the current location can be obtained, including historical traversable paths, historical success rates, and behavior risk levels, achieving full utilization of historical navigation experience. Based on historical traversable paths, historical success rates, and environmental data, a first navigation behavior sequence containing multiple navigation control actions is generated. The initial control parameters are then adjusted based on the path weights, environmental data, and behavior risk levels in the navigation behavior graph to obtain a second navigation behavior sequence, thereby improving the logistics vehicle's adaptability to dynamic environments while ensuring navigation safety. Finally, navigation control of the logistics vehicle is achieved based on the second navigation behavior sequence, improving the navigation efficiency of the logistics vehicle in complex and changing environments.

[0018] 2. By analyzing historical environmental evolution data, environmental change events are identified, and the locations of these events are marked as primary navigation behavior nodes. Transitional navigation behavior nodes are then set between adjacent primary navigation behavior nodes based on preset spatial division rules, achieving a reasonable division of the industrial scene space. Environmental memory mapping is performed on each navigation behavior node to obtain its environmental memory characteristics over multiple first preset time periods and its environmental change characteristics over second preset time periods, thereby establishing a deep understanding of the dynamic changes in the environment. By combining navigation intent vector analysis with environmental characteristics, the environmental stability values ​​of the navigation behavior nodes in each time period are calculated. Time periods with high environmental stability values ​​and that satisfy navigation intent constraints are selected as target passage periods, ensuring the stability of the navigation process. Based on this, we further analyze the overlap of target passage time periods and physical accessibility between adjacent nodes in physical space, establish candidate behavioral paths, and analyze the target environment change characteristics of candidate behavioral paths based on environmental change events. By using the reciprocal of the passage difficulty as the path weight, we select the candidate behavioral path with lower passage difficulty as the final behavioral path, thereby constructing a navigation behavior map that considers both dynamic environmental changes and navigation safety, providing a reliable guarantee for the efficient navigation of logistics vehicles in complex industrial environments.

[0019] 3. By combining environmental data analysis with the corresponding traffic sub-segments for navigation control actions, traffic environment characteristics, including obstacle distribution, traffic width changes, and ground conditions, are obtained, achieving precise perception of the local traffic environment. Through matching analysis of traffic environment characteristics with behavioral risk levels, risk impact factors characterizing the degree of influence of behavioral risk levels on navigation control actions are obtained. Based on path weights, the risk impact factors are weighted to obtain the navigation stability characteristics of each navigation control action, thus establishing the correlation between navigation control actions and environmental risks and generating a second navigation behavior sequence. This enables the logistics vehicle to maintain navigation stability while adapting to environmental changes in industrial scenarios, effectively improving the navigation safety and reliability of the logistics vehicle in complex and changing environments. Attached Figure Description

[0020] Figure 1 This is a schematic diagram of the system architecture of an embodiment of a logistics vehicle navigation control method or a logistics vehicle navigation control system applied in this application.

[0021] Figure 2 This is a flowchart illustrating a navigation control method for a logistics vehicle according to an embodiment of this application;

[0022] Figure 3 This is a schematic diagram of the structure of a logistics vehicle navigation control system according to an embodiment of this application;

[0023] Figure 4 This is a schematic diagram of the structure of the electronic device in the embodiments of this application.

[0024] Explanation of reference numerals in the attached drawings: 301, Acquisition module; 302, First construction module; 303, Second construction module; 304, Mapping module; 305, Generation module; 306, Adjustment module; 307, Control module; 401, Processor; 402, Communication bus; 403, User interface; 404, Network interface; 405, Memory. Detailed Implementation

[0025] To enable those skilled in the art to better understand the technical solutions in this specification, the technical solutions in the embodiments of this specification will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments.

[0026] Figure 1 An exemplary system architecture 100 is shown, which can be applied to an embodiment of the logistics vehicle navigation control method or logistics vehicle navigation control system of this application.

[0027] like Figure 1As shown, system architecture 100 may include terminal devices 101, 102, and 103, a network 104, and a server 105. Network 104 serves as the medium for providing communication links between terminal devices 101, 102, and 103 and server 105. Network 104 may include various connection types, such as wired or wireless communication links, or fiber optic cables, etc.

[0028] Terminal devices 101, 102, and 103 are logistics vehicles deployed at logistics operation sites. Each vehicle is an independent terminal in the system, integrating a positioning module (such as a Global Positioning System), an environmental perception module (such as LiDAR, cameras, and ultrasonic sensors), a navigation control module, and a communication module. Their main functions include: real-time acquisition of their own location data; acquisition or reception of task parameter information; perception of the surrounding environment data; collaboration with the server to obtain navigation control strategies generated based on navigation behavior maps; execution of autonomous navigation and obstacle avoidance control based on navigation control action sequences; and uploading environmental evolution data and traffic feedback information collected during operation to the server for subsequent map updates and model optimization.

[0029] Network 104 is used to connect multiple vehicle terminal devices and the server for data communication. It can include a local wireless network (such as Wi-Fi, 5G industrial private network) or a wired network, supporting functions such as task assignment, data synchronization, map updates, and control policy downlink. In some scenarios, the vehicle has edge computing capabilities and can use local caching strategies to perform navigation tasks when the network is down.

[0030] Server 105 is the centralized control and decision-making platform of the system, deployed with navigation control algorithms, a map generation module, and a task scheduling module. Its main functions include: constructing a navigation behavior map based on historical environmental evolution data; analyzing task parameters to generate the vehicle's navigation intent vector; generating navigation behavior sequences based on the map and real-time environmental data; evaluating behavior risk levels and path weights, and optimizing initial parameters; distributing the generated navigation control strategies (such as the second navigation behavior sequence) to each logistics vehicle; and receiving operational data from the vehicles for continuous model learning and map updates. Terminal devices 101, 102, and 103 can interact with server 105 via network 104 to receive or send messages. Various communication client applications, such as model training applications and video recognition applications, can be installed on terminal devices 101, 102, and 103. Server 105 can be a server providing various services, such as a backend server for processing data displayed on terminal devices 101, 102, and 103. The backend server can analyze and process the received data and feed the processing results back to the terminal devices.

[0031] Figure 2This is a flowchart illustrating a navigation control method for a logistics vehicle according to an embodiment of this application.

[0032] Please see Figure 2 This application provides a navigation control method for a logistics vehicle, applied to a server. The method includes:

[0033] S201. Obtain the location data, task parameters, environmental data corresponding to the location data, and historical environmental evolution data of the workshop where the logistics vehicle is located at its current location.

[0034] The logistics vehicle uses a high-precision positioning module to collect real-time two-dimensional or three-dimensional coordinate information of its current location, forming location data. This location data not only provides the basis for the vehicle's positioning within a map, but also serves as a crucial input for evaluating the feasibility of subsequent route planning.

[0035] Simultaneously, the task parameters that the logistics vehicle needs to execute are parsed from the scheduling instructions issued by the task scheduling system. These task parameters include, but are not limited to, the workshop number where the logistics vehicle is located, the type of material, the urgency level, and the expected arrival time, and are used to construct a navigation intent vector. The navigation intent vector is a vectorized abstract representation of the task, employing a multi-dimensional feature embedding method to map task-related semantic information into a vector space. Its function is to convey the task objective tendency that the current navigation behavior must satisfy to the navigation decision module.

[0036] To comprehensively perceive the current environmental conditions and improve adaptability to environmental changes, the logistics vehicle also needs to acquire environmental data corresponding to its current location. This environmental data originates from a multimodal perception system deployed in the workshop, including LiDAR, industrial cameras, and ultrasonic sensors, which can acquire real-time information such as obstacle distribution, aisle congestion, and the trajectories of dynamic objects (e.g., personnel, other vehicles). After data fusion and spatiotemporal normalization, the environmental data generates a standardized environmental state vector, serving as a crucial input for navigation map construction and behavior sequence generation.

[0037] Furthermore, to improve the navigation system's ability to predict environmental evolution trends, step S201 also requires acquiring historical environmental evolution data corresponding to the workshop where the logistics vehicle is located. This historical environmental evolution data is collected by an environmental monitoring system deployed long-term in the workshop, recording information such as changes in spatial traffic conditions, obstacle frequency, and congestion distribution patterns within the workshop at different time periods. By processing the historical environmental data using time series analysis methods (such as sliding window averaging, periodic pattern recognition, and change point detection), representative environmental evolution characteristics are extracted. This not only provides a basis for assessing path weights and risk levels in the subsequent navigation behavior map but also helps to avoid potentially high-risk paths in advance, improving the safety and stability of navigation behavior.

[0038] S202. Construct the navigation intent vector of the logistics vehicle based on the task parameters;

[0039] In this embodiment, to enable the logistics vehicle to understand the target task and adaptively adjust its navigation behavior, a navigation intent vector of the logistics vehicle is constructed based on the task parameters. The purpose of step S202 is to transform the task information issued by the scheduling system into a structured representation that can be used for navigation behavior map recognition and calculation, thereby giving the navigation decision-making process clear task orientation and dynamic adaptability.

[0040] Task parameters are task description data generated by the scheduling system's central processing unit based on the logistics plan. Since the original task parameters are mostly symbolic or textual information, they cannot be directly used for navigation behavior modeling. Therefore, they need to be vectorized to map them into navigation intent vectors with mathematical computability and feature representation capabilities. The navigation intent vector is a high-dimensional feature representation built upon task semantic understanding. The construction process first defines the navigation intent space based on the task parameters, that is, designing a multi-dimensional feature vector structure through task labels, such as destination location encoding, time urgency factor, material hazard level, and path preference weight. Each feature dimension represents a behavioral tendency that should be prioritized or avoided during navigation.

[0041] In practice, the generation of navigation intent vectors is achieved using an embedding learning-based approach. The system extracts features and maps vectors to different types of task parameters using a pre-trained task semantic encoding model. For example, the destination workshop number is first converted into a one-hot vector using discrete label encoding, and then dimensionality is reduced through a linear transformation layer; task priority is converted into numerical time weight coefficients using a mapping function; material hazard levels are assigned higher path safety constraint weights to high-risk tasks by referencing a pre-established risk weight mapping table; and path preferences are generated based on historical task execution statistics to produce corresponding path offset vectors. After embedding these features into a unified vector space, feature concatenation and normalization are performed to ultimately generate a set of navigation intent vectors containing multi-dimensional task semantic features. These navigation intent vectors not only accurately express the expected attributes of the current task regarding the navigation path but also provide clear navigation goal guidance for the selection of behavior nodes and path weight evaluation in the subsequent navigation behavior graph.

[0042] S203. Construct a navigation behavior graph based on navigation intent vectors and historical environmental evolution data. The navigation behavior graph includes navigation behavior nodes, behavior paths, and path weights of behavior paths.

[0043] As a structured navigation knowledge representation model, the navigation behavior graph dynamically encodes the optional navigation behaviors, possible routes, and reliability of each route for a logistics vehicle under specific task-driven conditions through a graph structure. This provides a decision-making basis for the navigation control system that combines task orientation and environmental adaptability. The graph not only contains information on navigation behavior nodes and paths but also introduces path weights to represent the accessibility level of each path under historical environmental conditions. Constructing the navigation behavior graph may include steps S2031-S2034:

[0044] S2031. Determine environmental change events in the workshop based on historical environmental evolution data;

[0045] Environmental change events refer to sudden, periodic, or trending changes in the environment, either spatially or temporally, that have a potential impact on the normal passage capability of logistics vehicles. Examples include the frequent appearance of temporary obstacles, equipment activation and shutdown, high-frequency personnel movement, or short-term closure of operating areas. Identifying these events is crucial for improving the dynamic adaptability and risk avoidance capabilities of navigation behavior maps.

[0046] In the specific implementation process, a spatiotemporal distribution model is established using the historical environmental evolution data of the workshop where the logistics vehicle is located. Historical environmental evolution data refers to dynamic environmental status data collected over a relatively long period (such as the past week or month) by the environmental sensing system deployed in the logistics workshop. This includes obstacle distribution maps, traffic flow statistics, personnel activity trajectories, equipment operating status, and abnormal event records for different time periods. After timestamp alignment and spatial grid normalization, the historical environmental evolution data forms a multidimensional environmental feature sequence with temporal continuity and spatial mapping capabilities.

[0047] After establishing the data model, the system employs a time-series change point detection algorithm to identify environmental change events. The change point detection method used is a mean-variance joint detection mechanism based on a sliding window, combined with chi-square test and cumulative sum control chart algorithms to assess whether there are significant mean shifts or fluctuations in the environmental feature sequence. For example, within any spatial region, if the frequency of obstacles suddenly increases from an average of once per day to more than five times per day over several consecutive first preset time periods, and continues to reach a certain time threshold, it is judged as a sudden high-frequency obstacle event. To further improve detection accuracy, the system incorporates spatial neighborhood factors and historical periodic factors into the event identification criteria. Spatial neighborhood factors refer to the coordinated changes in the environmental state surrounding the event's location; if multiple adjacent grid points exhibit similar trends within the same time period, the event's credibility increases. Historical periodic factors are identified through Fourier transform or wavelet analysis to determine whether there are periodic peak events, such as daily rush hour or timed equipment testing, thus distinguishing between sporadic and periodic events.

[0048] After identifying environmental change events, the system uses a space-time tagging method to structurally encode these events. Each event includes attributes such as location coordinates, start time, duration, radius of influence, event type (e.g., high-frequency obstacles, traffic congestion, closed areas), and risk level. These structured event tags will serve as dynamic constraint factors in the navigation behavior map construction process, adjusting the path weight values ​​of corresponding behavioral paths to prioritize avoiding high-risk paths or implementing peak-shaving measures during peak travel times.

[0049] S2032. Determine navigation behavior nodes based on environmental change events, and determine the environmental feature vector of each navigation behavior node. The navigation behavior nodes include the main navigation behavior node and the transitional navigation behavior node.

[0050] In step S2032, a set of navigation behavior nodes with spatiotemporal sensitivity is constructed based on environmental change events, and an environmental feature vector reflecting its historical traffic characteristics is generated for each navigation behavior node. This may include the following steps: marking the location where the environmental change event occurs as a primary navigation behavior node; setting multiple transitional navigation behavior nodes between adjacent primary navigation behavior nodes based on a preset spatial division rule, the navigation behavior nodes including multiple transitional navigation behavior nodes between the primary navigation behavior node and adjacent primary navigation behavior nodes; dividing the historical environmental evolution data according to a first preset time period, and aggregating the historical environmental evolution data for each of the first preset time periods based on the spatial location of the navigation behavior node; extracting features from the aggregated historical environmental evolution data to obtain a first feature sub-vector of the navigation behavior node within each first preset time period; analyzing the fluctuation amplitude of the first feature sub-vector within a second preset time period to obtain a second feature sub-vector, the second preset time period consisting of multiple consecutive first preset time periods; and concatenating the first feature sub-vector with the first feature sub-vector to obtain the environmental feature vector.

[0051] Navigation behavior nodes refer to the locations where a logistics vehicle may stop, turn, or switch states during the completion of its navigation task. To improve the flexibility and granularity of navigation control, navigation behavior nodes include primary navigation behavior nodes and multiple transitional navigation behavior nodes between adjacent primary navigation behavior nodes. Primary navigation behavior nodes typically correspond to key locations in the workshop layout, such as entrances / exits, main road intersections, and the front end of equipment areas, while transitional navigation behavior nodes are used to connect intermediate locations between primary nodes, enhancing path continuity and map coverage.

[0052] An environmental feature vector is a comprehensive data representation used to describe the environmental state of the spatial region where a navigation node is located over different time periods. It is primarily used to assess whether the node is suitable as part of a navigation path. To more comprehensively characterize the static stability and dynamic fluctuations of the environment, the environmental feature vector is subdivided into two components: a first feature sub-vector and a second feature sub-vector.

[0053] The first feature sub-vector refers to the feature representation generated after statistical analysis of the environmental state of the area corresponding to the navigation behavior node within a first preset time period on multiple short time scales. The time span is relatively short, typically in units of 5 or 10 minutes, continuously covering several recent time periods. The first feature sub-vector is mainly used to reflect the environmental characteristics of the node within a recent time window, such as short-term traffic density, obstacle frequency, and the degree of personnel activity, to capture the instantaneous state of the local environment and short-term traffic patterns. Therefore, this sub-vector can be compared to short-term memory, helping the system quickly identify whether the current node is suitable for passage in the near future.

[0054] The second feature vector is a feature representation generated by modeling the environmental state evolution trend of a node region over a longer time range (i.e., a second preset time period) composed of multiple consecutive first preset time periods. The second feature vector focuses on the dynamic fluctuations of the environment over a longer period, including trend changes, periodic fluctuations, or sudden anomalies. For example, it assesses whether the frequency of obstacle occurrences gradually increases within a continuous hour, or whether traffic flow shows a peak-increasing trend. This vector is used to determine whether a node is in a potentially highly volatile or unstable environment, thereby proactively avoiding risky paths or optimizing travel times.

[0055] In practical implementation, to achieve high-resolution modeling of the dynamic environment and risk perception capabilities during path planning, the system structurally sets navigation behavior nodes based on historical environmental evolution data and enhances the responsiveness of each node to time-related changes through environmental feature modeling. First, to identify key environmental turning points in the navigation path, the system marks the locations of environmental change events as primary navigation behavior nodes. Environmental change events refer to abrupt environmental state changes identified by change point detection algorithms in historical environmental evolution data, such as a surge in obstacle frequency, a sharp decrease in traffic density, or equipment state switching. These events typically have sudden or periodic characteristics and directly impact the normal passage of logistics vehicles. To achieve precise control over these high-risk areas, the system maps the event locations to the navigation space grid using a spatial positioning mechanism and sets primary navigation behavior nodes at these locations, binding event attributes for subsequent dynamic weight adjustments and obstacle avoidance strategy generation in path selection.

[0056] After calibrating the main navigation behavior nodes, to enhance the path continuity and structural integrity of the navigation behavior map, the system needs to set multiple transitional navigation behavior nodes between adjacent main navigation behavior nodes. This operation aims to improve the spatial resolution and combination flexibility of the navigation path, avoiding issues such as path jumps or incomplete path representation due to excessive node spacing. The setting of transitional navigation behavior nodes is based on preset spatial division rules, which comprehensively consider the workshop's physical layout, historical trajectory density, and accessibility. In practice, the system can generate candidate node positions on the line connecting two main navigation behavior nodes using equidistant interpolation, and then filter out points with actual passage records as valid transitional navigation behavior nodes by combining historical navigation trajectory heatmaps. In this way, the system achieves refinement and completion of the path structure without adding redundant nodes, providing richer path combination resources for subsequent behavior sequence generation.

[0057] Subsequently, an environmental feature vector is constructed for each navigation behavior node, and the system performs environmental memory mapping based on historical environmental evolution data. Specifically, the system first divides the historical environmental evolution data into multiple first preset time periods according to the time dimension, for example, each time period is 10 minutes. For each navigation behavior node, the system extracts the environmental state data of the area where the node is located in each first preset time period from the historical data based on its spatial location. The environmental state data includes, but is not limited to, the following indicators: traffic density (the number of logistics vehicles passing through the node per unit time), obstacle occurrence frequency (the number of obstacle detection records), traffic flow speed (the average passing speed per unit time), and personnel activity intensity (based on personnel positioning or sensor statistics), etc.

[0058] For the historical environmental evolution data within each first time period, the system first performs spatial aggregation processing, that is, merges and statistically analyzes the data within a certain spatial range around the node into the node's representative value; then, it normalizes these environmental state indicators to eliminate the differences in the dimensions of different indicators; finally, it compresses the features of multiple indicators based on dimensionality reduction algorithms (such as principal component analysis PCA, t-SNE, or autoencoder) to obtain the first feature vector of fixed dimensions.

[0059] To characterize the environmental change trends of nodes over a longer time period, the system constructs a second preset time period by combining multiple consecutive first preset time periods. For example, six consecutive first preset time periods (60 minutes in total) are selected as a second preset time period. Within the second preset time period, the system performs trend analysis and fluctuation modeling on the environmental state indicators represented by multiple first feature sub-vectors corresponding to the navigation behavior node. Specifically, the system calculates key statistical features from the original environmental indicator sequence (such as the traffic density values ​​within the six time periods). The variation amplitude is the difference between the maximum and minimum values; the trend slope is a linear trend fitted using the least squares method; the standard deviation reflects the intensity of fluctuation within the time period; and the maximum jump value is the maximum difference between adjacent time periods, used to identify abrupt changes. These statistical features are normalized, and a fixed-dimensional second feature sub-vector is constructed using feature compression. The second feature sub-vector reflects the environmental dynamics of the node over a longer time scale and is used to determine whether the navigation behavior node is in a potentially high-fluctuation, unstable, or high-risk area. The system combines the first and second feature sub-vectors corresponding to each navigation behavior node to form a complete environmental feature vector for the navigation behavior node. The combination method can use vector concatenation to ensure the effective integration of short-term environmental memory and long-term environmental trend information.

[0060] S2033. Combine the navigation intent vector with the analysis of the environmental feature vector, calculate the environmental stability value of each navigation behavior node in each first preset time period, and determine the first preset time period in which the environmental stability value is greater than the preset environmental stability threshold and meets the constraint conditions of the navigation intent vector as the target passage time period.

[0061] In this embodiment, the purpose of step S2033 is to quantify the passage stability of each navigation behavior node in different time periods by fusing navigation intent vectors and environmental feature vectors, and to select the target passage period that meets the navigation goal based on task requirements, so as to achieve time optimization and risk avoidance in the path planning process. This not only realizes dynamic adaptation to environmental state, but also enables the navigation system to have the ability to select time windows based on task semantics, thereby improving the success rate of logistics vehicles in complex environments and the efficiency of task completion.

[0062] In the specific implementation process, the system first traverses the environmental feature vectors within each first preset time period, taking navigation behavior nodes as units, and calculates the corresponding environmental stability value. The environmental stability value is a quantitative indicator of the reliability of the node's passage status within a specific time period, and is obtained by jointly calculating the first feature sub-vector and the second feature sub-vector. The calculation of the environmental stability value adopts a fusion method based on weighted normalized scoring, which maps the stability index in the first feature sub-vector and the volatility index in the second feature sub-vector to the [0,1] interval and then performs a weighted combination. Among them, the higher the passage density and the lower the obstacle frequency, the better the static passage conditions are, and they are given higher positive weights; while the standard deviation and maximum variation of environmental indicators reflect smaller fluctuations, representing lower dynamic risks, and are also given positive weights. The system calculates the environmental stability value using the following formula: S(i,t)=α×R(i,t)+β×(1−V(i,t)). Where S(i,t) represents the environmental stability value of navigation node i within the first preset historical time period t corresponding to the current time period, R(i,t) represents the environmental memory feature score of navigation node i within the first preset historical time period t corresponding to the current time period, V(i,t) represents the environmental volatility score of navigation node i within the first preset historical time period t corresponding to the current time period, and α and β are empirical weight coefficients, satisfying α+β=1. Through the above calculation method, the system can integrate the static suitability and dynamic predictability of the spatial environment into a single stability evaluation value, forming a quantitative expression of the node's accessibility.

[0063] To ensure that the target travel time period aligns with the actual needs of the current task, the system, based on the calculated environmental stability value, further introduces a navigation intent vector as a filtering constraint. For example, if the task has a high priority, the navigation intent vector will include a larger time weight factor, indicating that time periods with high travel stability and low delay risk should be prioritized; if the task materials are hazardous, the route selection must avoid historically highly volatile travel areas. The system performs matching analysis with the environmental stability values ​​of navigation behavior nodes, filtering out time periods that do not meet the navigation intent vector constraints, and retaining only those time periods that simultaneously have high environmental stability values ​​and align with task-oriented tendencies as candidates.

[0064] After completing the above analysis, the system selects time periods within each navigation behavior node that have an environmental stability value greater than a preset environmental stability threshold and meet the navigation intent vector constraints, and marks these as target passage periods. The preset environmental stability threshold is a benchmark value set based on the system's minimum requirements for passage safety. The setting of the preset environmental stability threshold, combined with historical task completion rates and statistical data on abnormal path events, forms an empirical parameter used to eliminate time periods with excessively high environmental volatility or significant passage risks.

[0065] S2034. Determine the path weights and behavioral paths based on the target travel time period and environmental change events;

[0066] After constructing the navigation behavior node system and selecting the target travel time periods for each node, high-quality travel paths in the navigation behavior graph are further mined, and the path risk is quantitatively assessed through environmental dynamic factors to determine the path weight of each path and select the behavior paths suitable for the current navigation task. This may include the following steps: when there is a passable physical path between the target travel time periods of physically adjacent target navigation behavior nodes, a candidate behavior path is established between the target navigation behavior nodes, where the target navigation behavior nodes are those whose environmental change events meet preset conditions; the second feature sub-vectors of all target navigation behavior nodes in the candidate behavior paths within the overlapping recommended travel time periods are concatenated to obtain a target second feature sub-vector; the feature indicators in the target second feature sub-vector are weighted to obtain the travel difficulty of each candidate behavior path, the reciprocal of the travel difficulty is determined as the path weight, and the candidate behavior paths whose travel difficulty is less than a preset travel difficulty threshold are determined as the behavior paths.

[0067] In practice, the system first constructs candidate behavioral paths based on the target travel time period. A target navigation behavior node refers to a navigation behavior node within its corresponding spatial area that experiences an environmental change event that meets preset conditions during a specific time period. These preset conditions include, but are not limited to: event type (such as a sudden increase in obstacles, abnormal traffic density, or a sudden drop in traffic flow), whether the event occurred within the target travel time period or its adjacent period, whether the event's intensity exceeds a set threshold, and the frequency of the event's occurrence per unit time. The system marks the navigation behavior nodes associated with environmental change events that meet all preset conditions as target navigation behavior nodes.

[0068] In the navigation behavior graph, if two target navigation behavior nodes are physically connected by a passable path (e.g., there are no obstacles between them, the ground conditions are accessible, and historical trajectory records exist), then a passable physical path exists between the target navigation behavior nodes. Based on this, the system determines whether there is a temporal overlap in the target travel time periods of these two target navigation behavior nodes, i.e., whether there are recommended time periods for simultaneous travel. If both conditions are met, the system establishes a candidate behavior path between the two target navigation behavior nodes. The design logic of this operation is to ensure that the path is not only spatially connected but also has synchronous travel opportunities in the temporal dimension, thereby avoiding interruptions or waiting due to time misalignment during actual execution.

[0069] After constructing candidate behavioral pathways, the system further analyzes the traffic risk of each pathway based on environmental change events. To this end, the system first extracts all environmental change events covering the spatial range of the candidate behavioral pathways within the overlapping recommended travel periods, and generates a target second feature vector based on the environmental evolution data within these overlapping recommended travel periods. The target second feature vector is a vector representation formed by statistically modeling the environmental dynamic indicators (such as obstacle frequency change rate, traffic flow velocity variance, and frequency of traffic density abrupt changes) of the candidate behavioral pathway within the target travel period, reflecting the environmental fluctuation trend of this path during the recommended travel time. In this process, the system employs a sliding window mechanism and a weighted average strategy to merge and integrate the second feature vectors of each node on the path, thereby obtaining a target second feature vector covering the entire pathway.

[0070] After obtaining the target's second feature vector, the system further calculates the passage difficulty for each candidate behavioral path. Passage difficulty refers to the overall risk level of completing continuous passage instructions within the target passage period; a higher value indicates greater uncertainty and execution difficulty for the path during that period. The system assigns risk weights to the feature indicators in the target's second feature vector (e.g., assigning a higher weight to the obstacle frequency change rate to identify sudden obstacle risks) and performs a weighted summation to obtain the final passage difficulty score. To enable subsequent path selection algorithms to prioritize more stable paths, the system uses the reciprocal of the passage difficulty for each candidate behavioral path as its path weight. That is, path weight = 1 / passage difficulty, meaning that paths with lower passage risk have higher weights and higher priority in path optimization.

[0071] Finally, the system filters all candidate behavioral paths based on a preset accessibility difficulty threshold, identifying paths with accessibility difficulty below this threshold as behavioral paths. The accessibility difficulty threshold is set by the system based on historical navigation success rates and risk tolerance, and is used to exclude paths with insufficient accessibility stability during the target travel period. Through this filtering mechanism, a batch of behavioral paths with physical accessibility, time synchronization, and environmental stability is ultimately formed, serving as the basis for subsequent path generation and scheduling optimization.

[0072] For example, in a logistics workshop, there is a 3-meter-wide flat passage between two main navigation behavior nodes A and B. Historical trajectory data shows frequent passage between the two nodes, and their target passage periods both include an overlapping time period from 10:00 AM to 10:10 AM. The system confirms the existence of a passable physical path and temporal intersection between the two nodes, establishing a candidate behavior path AB. Further analysis of environmental change events along this path from 10:00 AM to 10:10 AM reveals a history of frequent equipment start-ups and shutdowns in the middle section of the path. Based on the node's second feature sub-vector, the system constructs the target second feature sub-vector for this path during this time period, calculating a passage difficulty of 0.45. The system sets a passage difficulty threshold of 0.5, considering the path risk acceptable, and calculates the path weight as 1 / 0.45 ≈ 2.22, incorporating it into the behavior path set for priority use by subsequent path planning algorithms. Through this process, the system effectively achieves temporal-spatial-environmental linkage optimization of paths, providing dynamic and robust path support for intelligent navigation in complex workshop environments.

[0073] After obtaining navigation behavior nodes, behavior paths, and path weights, a navigation behavior graph is constructed. The graph construction process begins with navigation behavior nodes as the graph node elements. Each navigation behavior node corresponds to a specific spatial location. In previous steps, node partitioning was completed by combining historical environmental evolution data and environmental change events, and an environmental feature vector was configured for each node to provide time-sensitive accessibility assessment criteria during path selection. Nodes include primary navigation behavior nodes and transitional navigation behavior nodes. Primary navigation behavior nodes represent spatial locations with significant environmental changes or mission criticality, while transitional navigation behavior nodes are used to enhance path continuity and the spatial coverage of the graph.

[0074] Subsequently, the system uses behavioral paths as edge elements in the graph structure for connection operations. Each behavioral path connects two navigation behavior nodes, representing the possible movement path of the logistics vehicle during task execution. The establishment of behavioral paths is based on the dual constraints of spatial accessibility and consistency of travel time periods. That is, not only must there be an actual passable physical path between the two nodes, but there must also be recommended travel time periods that overlap in historical data to ensure the synchronous availability of the path in space and time. For each established path, the system calculates the travel difficulty based on the target second feature vector of the candidate path within the recommended travel time period, and uses the reciprocal of the travel difficulty as the path weight. In the graph structure, the path weight is bound to the behavioral path as an edge attribute for subsequent path search algorithms to use.

[0075] The final generation of the navigation graph is achieved using a graph structure modeling engine. All navigation behavior nodes are treated as the node set of the graph, and all behavior paths and their weights are treated as the edge set. Adjacency matrices and path weight matrices are constructed to support graph traversal and optimal path search. The system represents the navigation behavior graph as a weighted directed graph, where each edge has directional and weight attributes, accurately reflecting the optimal travel direction and priority of the logistics vehicle at different time periods under task-driven conditions.

[0076] S204. Map the location data of the logistics vehicle to the navigation behavior graph to obtain the graph context information of the target navigation behavior node corresponding to the current location. The graph context information includes the historical passable paths, historical passability success rate and behavior risk level of the target navigation behavior node.

[0077] Step S204 involves fusing and mapping the current location data of the logistics vehicle with the navigation behavior graph to obtain the structured representation information of the current location in the navigation behavior graph, i.e., the graph context information. This provides a decision-making basis with historical experience support and risk perception capabilities for the generation of subsequent navigation behavior sequences and the dynamic adjustment of control parameters. The navigation behavior graph, as a structured graph model, has already been constructed in previous steps, including navigation behavior nodes, behavior paths, and path weights, capable of expressing the navigability structure of the workshop environment in the three dimensions of space, time, and task semantics. The location data consists of two-dimensional or three-dimensional spatial coordinate information collected in real time by the logistics vehicle's high-precision positioning module, reflecting its current physical spatial location. The purpose of mapping the location data to the navigation behavior graph is to determine the vehicle's current location node in the graph and obtain the graph structure attribute information surrounding that node, forming graph context information to guide the subsequent path planning and action generation process with historical reference and risk constraint capabilities.

[0078] In the specific implementation process, the system first receives the current location data of the logistics vehicle, typically in a continuous spatial representation under a global coordinate system, such as the x, y or x, y, z coordinate values ​​generated by the positioning module. Such continuous spatial coordinates cannot be directly located in the navigation behavior graph; a spatial discrete mapping mechanism is needed to assign them to navigation behavior nodes in the graph. The mapping mechanism employs a dual-judgment method based on distance threshold matching and topological adjacency constraints: First, the system calculates the spatial Euclidean distance between the current location and all navigation behavior nodes in the graph, filtering out several candidate nodes with the closest distance. Then, combining the vehicle's current motion direction, velocity vector, and the edge directions between nodes in the graph, the system determines the target navigation behavior node that best matches the vehicle's current navigation intent. This is the corresponding node of the logistics vehicle's current location in the navigation behavior graph. Based on this, the system extracts the graph context information of the target navigation behavior node, generated under the drive of historical data.

[0079] Graph context information refers to a structured set of knowledge built around target navigation behavior nodes, including three core elements: historical traversable paths, historical success rates, and behavioral risk levels. Historical traversable paths refer to the set of paths that started from the target navigation behavior node and successfully reached the next navigation behavior node during historical task execution, representing the traversable options of the target navigation behavior node under past environmental states and scheduling conditions. The system statistically analyzes the paths originating from the node in the historical navigation behavior sequence, selecting paths with success rates higher than an empirical threshold as valid paths, forming a set of historical traversable paths. Historical success rate refers to the actual completion rate of each path for the node in past tasks, calculated as the ratio of successful attempts to attempts, used to measure path reachability and execution stability. Behavioral risk level is a comprehensive score of the environmental fluctuations exhibited by the target node at different time periods, based on a weighted sum of the node's second feature vector and the frequency of historical events, reflecting the potential risk level of the node during task execution.

[0080] By acquiring the aforementioned contextual information, the system can fully understand the vehicle's current location's environmental adaptability, path selection tendencies, and potential risk exposure points in historical task execution before generating navigation behavior sequences. This provides a foundation for subsequent path planning with historical memory and dynamic risk perception capabilities. Especially when facing complex dynamic environments or sudden changes in task priorities, contextual information can provide more flexible and redundant decision-making basis for navigation behavior generation, improving the robustness and task completion efficiency of the navigation system.

[0081] S205. Generate a first navigation behavior sequence based on historical passable routes, historical passability success rate and environmental data. The first navigation behavior sequence contains multiple navigation control actions, and each navigation control action corresponds to a set of initial control parameters.

[0082] Step S205, based on the map context information corresponding to the current location of the logistics vehicle, combined with historical passable paths, historical success rates, and real-time environmental data of the current location, generates a first navigation behavior sequence reflecting the current task objective orientation, environmental adaptability, and historical success experience. The first navigation behavior sequence is a list of behaviors consisting of several navigation control actions arranged in chronological order. Each control action corresponds to an executable set of initial control parameters, used to guide the vehicle's operation instructions on a specific travel path, including driving direction, speed adjustment, steering angle, obstacle avoidance methods, etc. Generating a first navigation behavior sequence may include the following steps: determining the current traffic status of the historical passable routes based on the environmental data; identifying passable routes with a historical success rate greater than a preset success rate threshold and current traffic status meeting preset traffic conditions as target passable routes; determining target navigation control points for the target passable routes, the target navigation control points including turning points, obstacle avoidance points, and speed change points, and dividing the target passable routes into multiple passable sub-segments based on the target navigation control points; for each passable sub-segment, determining navigation control actions based on the starting position, ending position, and the target navigation control point corresponding to the starting position of the passable sub-segment, the navigation control actions including straight driving, turning and obstacle avoidance, and speed adjustment; determining initial control parameters based on the length and width of the passable sub-segment corresponding to each navigation control action and the positional distribution feature vector of the target navigation control points in the target passable route; and connecting the navigation control actions and the initial control parameters corresponding to the navigation control actions according to the temporal relationship of the passable sub-segments to obtain the first navigation behavior sequence.

[0083] In the specific implementation process, the system first needs to filter historical passable paths in the map context information to determine the paths with practical passability under the current circumstances. To this end, the system acquires real-time environmental data of the current location and its surrounding area through the environmental perception module, including obstacle distribution, traffic density, and dynamic object movement trajectories. This environmental data is then matched and analyzed with historical passable paths to determine whether each path is currently passable. Specifically, based on the spatial obstacle distribution and dynamic entity trajectory information collected by LiDAR, industrial cameras, or ultrasonic sensors in the current environmental data, the system quantifies the current passability of each historically passable path through obstacle coverage analysis and channel congestion index calculation. The obstacle coverage rate is calculated by dividing the path space into multiple grid areas and statistically analyzing the proportion of grid points occupied by obstacles, reflecting the static accessibility of the path. The number of grid areas is determined by the size of the logistics workshop and the pre-determined planning accuracy. The channel congestion index assesses the dynamic traffic pressure of the path through standardized processing of the ratio of dynamic target density per unit area to the historical average. After determining the current traffic status, the system further introduces the historical traffic success rate as a stability indicator, and selects those path segments whose traffic success rate is higher than the preset success rate threshold and whose current traffic status meets the set feasibility conditions as target traffic routes.

[0084] After determining the target passageway, the system further refines the path structure, identifying control points that play a crucial guiding role in navigation. These control points include turning points, obstacle avoidance points, and speed change points, corresponding to areas in the path where direction changes, obstacles intersect, or passage widths change, respectively. By analyzing the dynamic changes in the path's geometry and environmental data, the system constructs a path control point distribution model using methods such as angle thresholding to identify turning points, obstacle density changes to identify obstacle avoidance points, and passage cross-sectional area changes to identify speed change points. Subsequently, based on the locations of these control points, the system divides the target passageway into several sub-segments. Each sub-segment is bounded by the spatial area between two consecutive control points, possessing relatively singular navigation objectives and environmental characteristics, providing structured input units for subsequent actions and parameter generation.

[0085] For each passage sub-segment, the system generates appropriate navigation control actions based on its starting and ending points, as well as the type of control point corresponding to the starting point. These control actions include three basic types: straight-line driving, steering and obstacle avoidance, and speed adjustment. When the starting point of the passage sub-segment is a regular node and the path direction is stable, the system sets the navigation action to straight-line driving. If the starting point is a turning point, the system determines whether steering and obstacle avoidance are needed based on the angle of change in the path direction vector. If the path width or pedestrian density changes significantly, the system sets a speed adjustment action. In this way, the system deconstructs the complex path execution process into a sequence of behaviors composed of basic navigation actions, giving navigation control the advantages of composability and local adjustability, thus enhancing the system's response flexibility and control accuracy.

[0086] After setting the control actions, the system further generates initial control parameters for each navigation control action. These parameters include speed, steering angle, obstacle avoidance radius, acceleration, and expected passage time, with specific values ​​dynamically calculated based on the structural attributes of the passage sub-segments and the distribution characteristics of control points. The system first determines the action duration and speed boundaries based on the length and width of the path segment, then performs weighted adjustments based on the control point distribution feature vector (including control point density, type ratio, and positional dispersion). For example, in areas with dense obstacle avoidance points, the system increases the corresponding obstacle avoidance radius parameter while reducing the target speed; in wide, obstacle-free areas, the system increases the maximum speed and acceleration parameters to ensure passage efficiency. All parameters are generated using empirical rules and model fitting functions to ensure controllability and safety during actual execution. Finally, the system integrates all passage sub-segments according to their temporal relationship within the path, arranging the navigation control actions and initial control parameters corresponding to each sub-segment sequentially to form a complete first navigation behavior sequence.

[0087] S206. Calculate the adjustment change amount based on the path weights, environmental data, and behavioral risk levels in the navigation behavior map, and adjust the initial control parameters according to the adjustment change amount to obtain the second navigation behavior sequence;

[0088] After generating the first navigation behavior sequence, to further improve the stability of navigation behavior in dynamic environments and the robustness of task completion, in step S206, multi-dimensional input factors such as path weights in the navigation behavior graph, real-time environmental data, and the behavioral risk level of target navigation behavior nodes are introduced to perform stability analysis on the actual executability of each navigation control action, and an adjustment change is calculated accordingly. This adjustment change is used to correct the initial control parameters of each control action in the first navigation behavior sequence, thereby generating a second navigation behavior sequence that better meets the current task requirements and environmental conditions. This may include the following steps: analyzing the navigation stability feature vector of each navigation control action based on the path weights, environmental data, and behavioral risk levels in the navigation behavior graph; inputting the navigation stability feature vector into a preset adjustment function to calculate the adjustment change; and adjusting the initial control parameters corresponding to each navigation control action according to the navigation stability feature vector to obtain the second navigation behavior sequence.

[0089] In step S206, analyzing the navigation stability feature vector of each navigation control action based on path weights, environmental data, and behavioral risk levels in the navigation behavior map is a crucial prerequisite for achieving dynamic optimization of control parameters. By fusing analysis of the environmental adaptability, safety risk impact, and path weights representing traffic reliability in the map corresponding to each navigation control action, a set of stability indicators is extracted. These indicators quantify the executability and risk-bearing capacity of navigation control actions under current environmental and task conditions. The stability feature vector not only reflects the quality of navigation behavior within the spatial structure but also embodies its sensitivity to environmental disturbances, serving as a direct basis for subsequent calculations and adjustments. The process may include the following steps: determining the risk level score of the environmental data in multiple preset dimensions based on preset rules; combining the risk level scores into a traffic environment feature vector, wherein the environmental data is the environmental data of the traffic sub-segment corresponding to the navigation control action; performing a matching analysis between the traffic environment feature vector and the behavioral risk level to obtain a risk impact factor for each navigation control action, wherein the risk impact factor is used to characterize the degree of influence of the behavioral risk level on the navigation control action; and performing a weighted calculation on the path weight and the risk impact factor to obtain a navigation stability feature vector for each navigation control action.

[0090] During the analysis, the system first determines the traffic environment feature vector for each navigation control action corresponding to the traffic sub-segment by combining preset rules and environmental data. Scoring rules are pre-set based on task safety requirements and historical traffic experience, using a tiered threshold method or fuzzy scoring model to map the raw environmental data to a risk level score between 0 and 1. The traffic environment feature vector is a set of vectors that quantifies the complexity, frequency of dynamic changes, and degree of structural constraints the vehicle will face in that path segment. The system collects current environmental data for that path segment through a perception module, including obstacle density, frequency of dynamic object movement, path width change rate, and changes in ground friction coefficient. Each environmental indicator is standardized to form the traffic environment feature vector. Each dimension of features (k∈[1,n]) represents the risk level score of the corresponding attribute, such as obstacle density score, channel contraction rate score, etc. This vector is used to characterize the basic external conditions of path traffic stability and provide environmental background for subsequent risk impact analysis.

[0091] Based on the obtained environmental feature vector, the system performs a matching analysis with the behavioral risk level of target navigation behavior nodes in the navigation behavior map to obtain the risk impact factor for each navigation control action. The behavioral risk level is a node risk score calculated by the system during map construction based on historical environmental evolution data and event frequency. It is typically a floating-point number between 0 and 1; a higher score indicates a higher probability of the target navigation behavior node experiencing anomalies or interference in historical tasks. The matching analysis process uses a weighted vector inner product to fuse the environmental feature vector and the behavioral risk level, calculating the sensitivity of each environmental feature to behavioral risk. The specific calculation formula is as follows: ,in, This represents the risk impact factor for the i-th navigation control action. The weight of the k-th environmental feature in the risk response (obtained by the learning model). Let be the k-th dimension of the environmental feature vector, and r be the behavioral risk level value of the target navigation behavior node. Calculation results. This indicates the degree of risk exposure of the i-th navigation control action under the current environment and historical risk background. The larger the value, the more susceptible the action is to risk disturbances, and a higher safety margin needs to be given in subsequent parameter adjustments.

[0092] After acquiring the risk impact factors, the system further combines the path weights of the current path segment in the navigation behavior map to perform a weighted fusion of the risk impact factors and travel reliability, ultimately generating a navigation stability feature vector for each navigation control action. The path weights are values ​​calculated during map construction based on the path's historical travel stability and the reciprocal of the travel difficulty for the recommended time period, reflecting the overall reliability of the path. The stability feature vector is calculated using the following model: ,in, Let be the navigation stability feature vector for the i-th navigation control action. The path weights are the path segments. For its corresponding risk impact factors, and These are adjustable weighting coefficients used to control the proportion of reliability and risk factors in the feature vector, comprehensively evaluating the priority of the path and the impact of risks, and finally generating a navigation stability feature vector for subsequent control parameter adjustments.

[0093] After obtaining the navigation stability feature vector for each navigation control action, the system needs to further calculate the corresponding adjustment amount based on the feature vector, and dynamically optimize and adjust the initial control parameters of each navigation control action accordingly, thereby generating a second navigation behavior sequence with better execution stability and environmental adaptability.

[0094] In the specific implementation process, the system will assign a navigation stability feature vector to each navigation control action. As input, a preset adjustment function is determined accordingly, which maps the stability feature vector into changes in control parameters, thus constituting the adjustment change. Control parameters include speed, acceleration, steering angle, obstacle avoidance radius, and path holding time, each with its own specific adjustment logic. For example, for the target speed parameter v, the preset adjustment function is set as follows: ,in, Adjust the sensitivity coefficient for speed. This is a nonlinear compression function (such as sigmoid or tanh) for the stability eigenvector, used to avoid excessive adjustment. When This indicates that the stability of the action is low (e.g., a negative value or close to 0). A large negative output indicates a significant deceleration operation; while when... A value approaching 1 indicates high path stability, so the adjustment tends to be positive or close to zero, meaning the original speed is maintained or slightly increased. Other control parameters such as obstacle avoidance radius and acceleration are handled using similar strategy functions. The difference is that the adjustment direction of the obstacle avoidance radius is proportional to the risk, i.e.: ,in This is the maximum adjustment range for the obstacle avoidance radius. This indicates the degree of instability of the current action. The lower the stability, the larger the obstacle avoidance radius, thus leaving more room for dynamic adjustment of the path and reducing the risk of collision.

[0095] After the adjustment change is calculated, the system applies the adjustment change to the original initial control parameters, corrects and updates each parameter, and obtains the new control parameters. ,in These are the initial control parameters corresponding to the navigation control actions in the first navigation behavior sequence. This is the adjustment amount calculated based on the stability eigenvector. After the update, each control action has an optimized control strategy that reflects the path execution risk and passage reliability. Finally, the system combines all control actions and their updated parameters in sequence to construct the second navigation behavior sequence.

[0096] S207. Navigation control of the logistics vehicle is performed based on the second navigation behavior sequence.

[0097] After generating the second navigation behavior sequence, the system uses this sequence to control the logistics vehicle's navigation. Each navigation control action and its adjusted control parameters are sequentially input into the motion control module, driving the vehicle to perform specific operations along a predetermined path. During control, the system monitors the vehicle's status and environmental changes in real time to ensure stable execution of each control action within its corresponding sub-road segment, and triggers a fine-tuning mechanism when necessary to address unforeseen risks.

[0098] Please see Figure 3 This is a schematic diagram of the structure of a logistics vehicle navigation control system provided in an embodiment of this application. The logistics vehicle navigation control 300 specifically includes:

[0099] The acquisition module 301 is used to acquire the location data of the logistics vehicle at its current location, task parameters, environmental data corresponding to the location data, and historical environmental evolution data of the workshop where the logistics vehicle is located; the first construction module 302 is used to construct the navigation intent vector of the logistics vehicle based on the task parameters; the second construction module 303 is used to construct a navigation behavior graph based on the navigation intent vector and historical environmental evolution data, the navigation behavior graph including navigation behavior nodes, behavior paths, and path weights of behavior paths; the mapping module 304 is used to map the location data of the logistics vehicle to the navigation behavior graph, obtaining the graph context information of the target navigation behavior node corresponding to the current location, and the graph... The following information includes the historical passable paths, historical success rate, and behavior risk level of the target navigation behavior node; generation module 305 is used to generate a first navigation behavior sequence based on the historical passable paths, historical success rate, and environmental data. The first navigation behavior sequence contains multiple navigation control actions, each corresponding to a set of initial control parameters; adjustment module 306 is used to calculate the adjustment change based on the path weights, environmental data, and behavior risk level in the navigation behavior map, and adjust the initial control parameters according to the adjustment change to obtain a second navigation behavior sequence; control module 307 is used to perform navigation control on the logistics vehicle based on the second navigation behavior sequence.

[0100] Optionally, the second construction module 303 is specifically used for: determining the environmental change events of the workshop based on the historical environmental evolution data; determining the navigation behavior nodes based on the environmental change events, and determining the environmental feature vector of each navigation behavior node, wherein the navigation behavior nodes include main navigation behavior nodes and transitional navigation behavior nodes; analyzing the environmental feature vector in conjunction with the navigation intent vector, calculating the environmental stability value of each navigation behavior node in each first preset time period, and determining the first preset time period in which the environmental stability value is greater than a preset environmental stability threshold and meets the constraint conditions of the navigation intent vector as the target passage period; and determining the path weight and the behavior path based on the target passage period and the environmental change events.

[0101] Optionally, the second construction module 303 is further specifically used for: marking the location where the environmental change event occurs as a main navigation behavior node; setting multiple transitional navigation behavior nodes between adjacent main navigation behavior nodes based on a preset spatial division rule, wherein the navigation behavior nodes include multiple transitional navigation behavior nodes between the main navigation behavior node and adjacent main navigation behavior nodes; dividing the historical environmental evolution data according to a first preset time period, and aggregating the historical environmental evolution data of each first preset time period based on the spatial location of the navigation behavior node, performing feature extraction on the aggregated historical environmental evolution data to obtain a first feature sub-vector of the navigation behavior node in each first preset time period; analyzing the fluctuation amplitude of the first feature sub-vector in a second preset time period to obtain a second feature sub-vector, wherein the second preset time period consists of multiple consecutive first preset time periods; and concatenating the first feature sub-vector with the first feature sub-vector to obtain the environmental feature vector.

[0102] Optionally, the second construction module 303 is further specifically used for: establishing candidate behavior paths between target navigation behavior nodes when there is a passable physical path between adjacent target navigation behavior nodes in physical space during the target travel time period, wherein the target navigation behavior nodes are navigation behavior nodes whose environmental change events meet preset conditions; concatenating the second feature sub-vectors of all target navigation behavior nodes in the candidate behavior paths within the overlapping recommended travel time period to obtain a target second feature sub-vector; performing weighted calculation on the feature indicators in the target second feature sub-vector to obtain the travel difficulty of each candidate behavior path, determining the reciprocal of the travel difficulty as the path weight, and determining the candidate behavior paths whose travel difficulty is less than a preset travel difficulty threshold as the behavior paths.

[0103] Optionally, the generation module 305 is specifically used for: determining the current traffic status of the historical passable routes based on the environmental data; identifying passable routes with a historical success rate greater than a preset success rate threshold and whose current traffic status meets preset traffic conditions as target passable routes; determining target navigation control points for the target passable routes, the target navigation control points including turning points, obstacle avoidance points, and speed change points, and dividing the target passable routes into multiple passable sub-segments based on the target navigation control points; for each passable sub-segment, determining navigation control actions based on the starting position, ending position, and the target navigation control point corresponding to the starting position of the passable sub-segment, the navigation control actions including straight driving, turning and obstacle avoidance, and speed adjustment; determining initial control parameters based on the length and width of the passable sub-segment corresponding to each navigation control action and the positional distribution feature vector of the target navigation control points in the target passable route; and connecting the navigation control actions and the initial control parameters corresponding to the navigation control actions according to the temporal relationship of the passable sub-segments to obtain the first navigation behavior sequence.

[0104] Optionally, the adjustment module 306 is specifically used to: analyze the navigation stability feature vector of each navigation control action based on the path weights, environmental data and behavior risk levels in the navigation behavior map; input the navigation stability feature vector to calculate the adjustment change through a preset adjustment function, and adjust the initial control parameters corresponding to each navigation control action according to the navigation stability feature vector to obtain the second navigation behavior sequence.

[0105] Optionally, the adjustment module 306 is further specifically used for: determining the risk level score of the environmental data in multiple preset dimensions based on preset rules; combining the risk level scores into the traffic environment feature vector, wherein the environmental data is the environmental data of the traffic sub-segment corresponding to the navigation control action; performing matching analysis between the traffic environment feature vector and the behavior risk level to obtain a risk impact factor for each navigation control action, wherein the risk impact factor is used to characterize the degree of influence of the behavior risk level on the navigation control action; and performing weighted calculation on the path weight and the risk impact factor to obtain a navigation stability feature vector for each navigation control action.

[0106] It should be noted that the above embodiments of the apparatus are only illustrated by the division of the above functional modules. In practical applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above. In addition, the apparatus and method embodiments provided in the above embodiments belong to the same concept, and the specific implementation process can be found in the method embodiments, which will not be repeated here.

[0107] This embodiment also discloses an electronic device, as shown in the reference. Figure 4 The electronic device may include: at least one processor 401, at least one communication bus 402, a user interface 403, a network interface 404, and at least one memory 405. The communication bus 402 is used to enable communication between these components. The user interface 403 may include a display screen and a camera; optionally, the user interface 403 may also include a standard wired interface or a wireless interface. The network interface 404 may optionally include a standard wired interface or a wireless interface (such as a Wi-Fi interface). The processor 401 may include one or more processing cores. The processor 401 connects to various parts of the server using various interfaces and lines, and performs various functions and processes data by running or executing instructions, programs, code sets, or instruction sets stored in the memory 405, and by calling data stored in the memory 405. Optionally, the processor 401 may be implemented using at least one hardware form selected from digital signal processing, field-programmable gate arrays, and programmable logic arrays. The memory 405 may include random access memory (RAM) or read-only memory (ROM). Optionally, the memory 405 may include a non-transitory computer-readable medium. Memory 405 can be used to store instructions, programs, code, code sets, or instruction sets. For example... Figure 4 As shown, the memory 405, as a computer storage medium, may include an operating system, a network communication module, a user interface module, and an application program for a logistics vehicle navigation control method. Figure 4 In the electronic device shown, the user interface 403 is mainly used to provide an input interface for the user and to obtain the user input data; while the processor 401 can be used to call an application program stored in the memory 405 for a logistics vehicle navigation control method. When executed by one or more processors 401, the electronic device executes one or more methods as described in the above embodiments.

[0108] The foregoing description is merely an exemplary embodiment of this disclosure and should not be construed as limiting the scope of this disclosure. Any equivalent changes and modifications made in accordance with the teachings of this disclosure shall still fall within the scope of this disclosure. Other embodiments of this disclosure will be readily apparent to those skilled in the art upon consideration of the disclosure in this specification. This application is intended to cover any variations, uses, or adaptations of this disclosure that follow the general principles of this disclosure and include common knowledge or customary techniques in the art not described in this disclosure. The specification and embodiments are to be considered exemplary only, and the scope and spirit of this disclosure are defined by the claims.

Claims

1. A navigation control method for a logistics vehicle, characterized in that, Applied to a server, the method includes: Acquire the location data of the logistics vehicle at its current location, task parameters, environmental data corresponding to the location data, and historical environmental evolution data of the workshop where the logistics vehicle is located; Construct the navigation intent vector of the logistics vehicle based on the task parameters; A navigation behavior graph is constructed based on the navigation intent vector and the historical environmental evolution data. The navigation behavior graph includes navigation behavior nodes, behavior paths, and path weights of the behavior paths. The location data of the logistics vehicle is mapped to the navigation behavior graph to obtain the graph context information of the target navigation behavior node corresponding to the current location. The graph context information includes the historical passable paths, historical passability success rate and behavior risk level of the target navigation behavior node. A first navigation behavior sequence is generated based on the historical passable routes, the historical passability success rate and the environmental data. The first navigation behavior sequence contains multiple navigation control actions, and each navigation control action corresponds to a set of initial control parameters. The adjustment change is calculated based on the path weights, environmental data, and behavioral risk levels in the navigation behavior map, and the initial control parameters are adjusted according to the adjustment change to obtain the second navigation behavior sequence. The logistics vehicle is controlled for navigation based on the second navigation behavior sequence. The construction of the navigation behavior map based on the navigation intent vector and the historical environmental evolution data specifically includes: The environmental change events in the workshop are determined based on the historical environmental evolution data; The navigation behavior nodes are determined based on the environmental change events, and the environmental feature vector of each navigation behavior node is determined. The navigation behavior nodes include main navigation behavior nodes and transitional navigation behavior nodes. By combining the navigation intent vector with the environmental feature vector, the environmental stability value of each navigation behavior node is calculated in each first preset time period. The first preset time period in which the environmental stability value is greater than the preset environmental stability threshold and meets the constraint conditions of the navigation intent vector is determined as the target travel time period. The path weights and behavioral paths are determined based on the target travel time period and the environmental change events.

2. The navigation control method for a logistics vehicle according to claim 1, characterized in that, The environmental feature vector includes a first feature sub-vector and a second feature sub-vector. The process of determining the navigation behavior node based on the environmental change event and determining the environmental feature vector for each navigation behavior node specifically includes: The location where the environmental change event occurs is marked as the main navigation behavior node; Between adjacent main navigation behavior nodes, multiple transitional navigation behavior nodes are set based on preset spatial division rules. The navigation behavior nodes include multiple transitional navigation behavior nodes between the main navigation behavior node and adjacent main navigation behavior nodes. The historical environment evolution data is divided according to the first preset time period, and the historical environment evolution data of each first preset time period is aggregated based on the spatial location of the navigation behavior node. Feature extraction is performed on the aggregated historical environment evolution data to obtain the first feature sub-vector of the navigation behavior node in each first preset time period. Within a second preset time period, the fluctuation amplitude of the first feature sub-vector is analyzed to obtain the second feature sub-vector. The second preset time period consists of multiple consecutive first preset time periods. The environmental feature vector is obtained by concatenating the first feature sub-vector with the first feature sub-vector.

3. The navigation control method for a logistics vehicle according to claim 1, characterized in that, The determination of the path weight and the behavioral path based on the target travel time period and the environmental change event specifically includes: When there is an overlap in the recommended travel time between the target travel time of adjacent target navigation behavior nodes in physical space, and there is a passable physical path between the adjacent target navigation behavior nodes, a candidate behavior path is established between the target navigation behavior nodes, and the target navigation behavior node is the navigation behavior node whose environmental change event meets the preset conditions. The second feature sub-vectors of all target navigation behavior nodes in the candidate behavior path within the overlapping recommended passage period are concatenated to obtain the target second feature sub-vector. The feature indicators in the second feature sub-vector of the target are weighted and calculated to obtain the passage difficulty of each candidate behavior path. The reciprocal of the passage difficulty is determined as the path weight. The candidate behavior path whose passage difficulty is less than a preset passage difficulty threshold is determined as the behavior path.

4. The navigation control method for a logistics vehicle according to claim 1, characterized in that, The step of generating a first navigation behavior sequence based on the historical passable routes, the historical passability success rate, and the environmental data specifically includes: Based on the environmental data, the current access status of the historical passable routes is determined, and the passable routes with a historical access success rate greater than a preset success rate threshold and a current access status that meets preset access conditions are identified as target passable routes. The target navigation control points of the target passage are determined, including turning points, obstacle avoidance points and speed change points, and the target passage is divided into multiple passage sub-segments based on the target navigation control points; For each of the aforementioned traffic sub-segments, navigation control actions are determined based on the starting position, ending position, and target navigation control point corresponding to the starting position of the traffic sub-segment. The navigation control actions include straight-line driving, steering obstacle avoidance, and speed adjustment. The initial control parameters are determined based on the length and width of the passage sub-segment corresponding to each navigation control action and the positional distribution feature vector of the target navigation control point in the target passage. The navigation control actions and the corresponding initial control parameters are connected according to the temporal relationship of the passage sub-segments to obtain the first navigation behavior sequence.

5. The navigation control method for a logistics vehicle according to claim 4, characterized in that, The step of calculating the adjustment change based on the path weights, environmental data, and behavioral risk levels in the navigation behavior map, and adjusting the initial control parameters according to the adjustment change to obtain the second navigation behavior sequence, specifically includes: Based on the path weights, environmental data, and behavioral risk levels in the navigation behavior graph, analyze the navigation stability feature vector for each navigation control action; The navigation stability feature vector is input and the adjustment change is calculated through a preset adjustment function. The initial control parameters corresponding to each navigation control action are adjusted according to the navigation stability feature vector to obtain the second navigation behavior sequence.

6. The navigation control method for a logistics vehicle according to claim 5, characterized in that, The analysis of the navigation stability feature vector for each navigation control action based on the path weights in the navigation behavior map, the environmental data, and the behavioral risk level specifically includes: Based on preset rules, the risk level score of the environmental data is determined in multiple preset dimensions, and the risk level score is combined into a traffic environment feature vector. The environmental data is the environmental data of the traffic sub-road segment corresponding to the navigation control action. The traffic environment feature vector is matched and analyzed with the behavior risk level to obtain the risk impact factor of each navigation control action. The risk impact factor is used to characterize the degree of influence of the behavior risk level on the navigation control action. The navigation stability feature vector for each navigation control action is obtained by weighting the path weights and the risk impact factors.

7. A navigation control system for a logistics vehicle, characterized in that, include: The acquisition module is used to acquire the location data of the logistics vehicle at its current location, task parameters, environmental data corresponding to the location data, and historical environmental evolution data of the workshop where the logistics vehicle is located; The first construction module is used to construct the navigation intent vector of the logistics vehicle based on the task parameters; The second construction module is used to construct a navigation behavior graph based on the navigation intent vector and the historical environmental evolution data. The navigation behavior graph includes navigation behavior nodes, behavior paths, and path weights of the behavior paths. The mapping module is used to map the location data of the logistics vehicle to the navigation behavior graph to obtain the graph context information of the target navigation behavior node corresponding to the current location. The graph context information includes the historical passable paths, historical passability success rate and behavior risk level of the target navigation behavior node. The generation module is used to generate a first navigation behavior sequence based on the historical passable routes, the historical passability success rate and the environmental data. The first navigation behavior sequence contains multiple navigation control actions, and each navigation control action corresponds to a set of initial control parameters. The adjustment module is used to calculate the adjustment change amount based on the path weights, environmental data and behavior risk levels in the navigation behavior map, and adjust the initial control parameters according to the adjustment change amount to obtain the second navigation behavior sequence; The control module is used to perform navigation control on the logistics vehicle based on the second navigation behavior sequence; The second construction module is further configured to determine the environmental change events of the workshop based on the historical environmental evolution data; The navigation behavior nodes are determined based on the environmental change events, and the environmental feature vector of each navigation behavior node is determined. The navigation behavior nodes include main navigation behavior nodes and transitional navigation behavior nodes. By combining the navigation intent vector with the environmental feature vector, the environmental stability value of each navigation behavior node is calculated in each first preset time period. The first preset time period in which the environmental stability value is greater than the preset environmental stability threshold and meets the constraint conditions of the navigation intent vector is determined as the target travel time period. The path weights and behavioral paths are determined based on the target travel time period and the environmental change events.

8. An electronic device, characterized in that, include: One or more processors and memory; The memory is coupled to the one or more processors, the memory being used to store computer program code, the computer program code including computer instructions, the one or more processors invoking the computer instructions to cause the electronic device to perform the method as described in any one of claims 1-6.

9. A computer-readable storage medium comprising instructions, characterized in that, When the instructions are executed on an electronic device, the electronic device causes the electronic device to perform the method as described in any one of claims 1-6.

Images