Intelligent forklift management system and method based on AI collision avoidance

Abstract

The invention provides an intelligent forklift management system and method based on AI collision avoidance. The method comprises the steps that three-dimensional data, attribute data, historical working data and historical working environment data of a forklift are obtained; constructing a digital twin system of the forklift; the size of a passing limiting space of the forklift is determined according to the digital twin system, and the size of a driving detection area is determined; determining a to-be-detected area according to the driving detection area and the driving route, and acquiring data of the area; recognizing obstacles according to the regional data, extracting feature angular points, selecting feature control points, and judging whether the feature control points meet preset requirements or not; if feature control points meeting requirements exist, the feature control points serve as basic control points, remaining control points are solved in combination with the current pose of the forklift, a preset route planning strategy and constraint conditions, alternative routes are generated and transmitted to a route tracking program, the real-time detection state is kept, and the driving route of the forklift is adjusted. According to the scheme, the working efficiency of the forklift can be improved, and the safety risk can be remarkably reduced.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence technology, and particularly relates to an intelligent forklift management system and method based on AI anti-collision. Background Art

[0002] An intelligent forklift is an automated device with functions such as driverless operation, remote monitoring, and warehouse management, and is widely used in scenarios such as repetitive handling, heavy handling intensity, and harsh working environments. The existing intelligent forklift management system has low operating efficiency and low safety. Summary of the Invention

[0003] Based on the above problems, the present invention proposes an intelligent forklift management system and method based on AI anti-collision. By comprehensively utilizing AI technology, digital twin, and real-time data processing, an efficient and safe intelligent forklift management system is constructed, which can not only improve the working efficiency of the forklift, but also significantly reduce safety risks, and has broad application prospects.

[0004] In view of this, one aspect of the present invention proposes an intelligent forklift management method based on AI anti-collision, including: Obtaining the three-dimensional data of the forklift, the forklift attribute data, the forklift historical working data, and the forklift historical working environment data, and constructing a three-dimensional model of the forklift according to the three-dimensional data of the forklift; Generating a digital twin system of the forklift according to the three-dimensional model of the forklift, the forklift attribute data, the forklift historical working data, and the forklift historical working environment data; Determining the size of the passing restricted space of the forklift according to the digital twin system of the forklift; Determining the size of the driving detection area of the forklift according to the size of the passing restricted space; Determining the area to be detected in the driving direction according to the size of the driving detection area and the driving route of the forklift; Obtaining the area data of the area to be detected; Determining the obstacle data according to the area data; Extracting all feature corner points of the obstacle from the obstacle data, selecting the feature control points, and determining whether the feature control points meet the preset requirements; If there are no feature control points that meet the preset requirements, it means that it is no longer possible to pass ahead, control the forklift to execute an emergency stop and issue an alarm, and wait for the obstacle to be removed before continuing to run; When there are feature control points that meet the preset requirements, using the feature control points that meet the preset requirements as the basic control points, combining the current pose information of the forklift, the preset route planning strategy, and the constraint conditions, obtaining the remaining required control points, and generating an alternative path by combining the basic control points and the remaining required control points; Transmit the alternative path to the path tracking program and maintain the real-time detection status; Adjust the driving route of the forklift according to the alternative path.

[0005] Optionally, the step of determining the size of the passing restricted space of the forklift according to the forklift digital twin system includes: Determine the basic size of the passing restricted space of the forklift by using the three-dimensional model of the forklift in the forklift digital twin system; Simulate the working process of the forklift in the forklift digital twin system by using the historical working data of the forklift and the historical working environment data of the forklift, and determine the offset amount during the operation of the forklift during the simulation process; Modify the basic size of the passing restricted space according to the offset amount to obtain the size of the passing restricted space.

[0006] Optionally, the step of determining the size of the driving detection area of the forklift according to the size of the passing restricted space includes: Define the basic parameters of the driving detection area according to the operating characteristics of the forklift and the size of the passing restricted space; Conduct an initial calculation according to the size of the restricted space and determine the size of the basic detection area. Specifically: determine the minimum channel width as the width of the basic detection area; determine the minimum channel height as the height of the basic detection area; determine the length of the basic detection area based on the minimum turning radius; Use the following formula to calculate the size of the driving detection area: Size of driving detection area = Size of basic detection area + Safety margin; where the value of the safety margin is obtained through big data analysis based on the historical working data of the forklift; When calculating the size of the driving detection area, adaptively adjust the size of the driving detection area according to the ground conditions and environmental complexity; Verify the calculated size of the driving detection area through simulation tests to ensure that the forklift can drive safely under this detection area size; During the operation of the forklift, continuously monitor the environmental changes and dynamically adjust the size of the driving detection area according to the new data.

[0007] Optionally, the step of determining the area to be detected in the driving direction according to the size of the driving detection area and the driving route of the forklift includes: Determine the first area to be detected according to the size of the driving detection area and the driving route of the forklift; Obtain the driving status data and load status data of the forklift; Extract the current speed from the driving status data and dynamically adjust the detection area length of the first area to be detected according to the current speed to obtain the second detection area; Extract the steering angle from the driving state data, and adjust the detection area shape of the second area to be detected according to the steering angle to obtain the third area to be detected; Calculate the influence data of the load on the braking performance according to the load state data, and adjust the longitudinal distance and lateral range of the detection area of the third detection area according to the influence data to obtain the area to be detected.

[0008] Optionally, the step of determining the obstacle data according to the area data includes: Extract the area image data of the area to be detected from the area data; Input the area image data into a preset obstacle detection model; The obstacle detection model performs obstacle recognition and extraction on the area image data to obtain obstacle image data; Perform obstacle segmentation according to the obstacle image data to obtain obstacle segmentation data; Perform obstacle contour and center calculation according to the obstacle segmentation data, and perform coordinate transformation on the calculation results to obtain the coordinate sets of the obstacle contour points and the center point.

[0009] Optionally, the step of extracting all feature corner points of the obstacle from the obstacle data, selecting feature control points, and determining whether the feature control points meet the preset requirements includes: Extract the geometric feature information of the obstacle from the obstacle data; Use a feature detection algorithm to process the geometric feature information and extract all feature corner points; Select feature control points according to the distribution and importance of the feature corner points; Verify the selected feature control points to determine whether they meet the preset requirements; Output the feature control points that meet the preset requirements and their coordinates for subsequent obstacle recognition, path planning, or obstacle avoidance decision-making.

[0010] Optionally, when there are feature control points that meet the preset requirements, using the feature control points that meet the preset requirements as the basic control points, combining the current pose information of the forklift, the preset route planning strategy, and the constraint conditions, obtaining the remaining required control points, and generating an alternative path by combining the basic control points and the remaining required control points includes: Use the feature control points that meet the preset requirements as the basic control points; Obtain the current pose information from the forklift's sensor system, including the position and attitude of the forklift; Determine the preset route planning strategy; Define the constraint conditions in the path planning; Calculate the remaining required control points based on the basic control points, the current pose information of the forklift, the route planning strategy, and the constraint conditions; Combine the basic control points and the remaining required control points to generate alternative paths.

[0011] Optionally, the step of extracting the regional image data of the area to be detected from the regional data includes: Set a maximum detection distance threshold, establish a distance calculation method based on depth information, and construct a distance filtering algorithm to establish a distance threshold judgment model; Obtain the original image data of the regional data; Combine the distance threshold judgment model, register the depth information of the original image data, and eliminate the image areas exceeding the distance threshold to obtain the regional image data.

[0012] Optionally, the step of simulating the working process of the forklift in the forklift digital twin system by using the forklift historical working data and the forklift historical working environment data, and determining the offset when the forklift is running during the simulation includes: Process and analyze the forklift historical working data, including: cleaning abnormal data; standardizing the data format; extracting key features; Process and analyze the forklift historical working environment data, including: reconstructing the historical scene; extracting environmental features; constructing an environmental model; Conduct data correlation analysis on the forklift historical working data and the forklift historical working environment data, including: establishing a spatio-temporal correspondence relationship; identifying key influencing factors; constructing a data correlation model; Construct a simulation scenario according to the forklift digital twin system, including: Load the basic model: load the 3D model of the forklift; set physical parameters; configure dynamic characteristics; Reproduce the environmental scenario: construct the ground model; add obstacles; set environmental parameters; Simulate the interaction relationship: define collision detection; set the friction coefficient; configure dynamic response; Conduct dynamic simulation, including: Establish a dynamic model: construct a mass distribution model; set inertia parameters; define joint constraints; Conduct kinematic analysis: calculate steering characteristics; analyze acceleration characteristics; simulate the braking process; Analyze the influence of the load: simulate the change of the load; calculate the center of gravity offset; analyze the stability; Calculate the offset, including: Determine the reference trajectory: extract the ideal path; set key nodes; generate a reference trajectory; Simulate the actual trajectory: execute dynamic simulation; record the motion trajectory; extract position data; Offset analysis: Calculate the position deviation; Analyze the attitude deviation; Statistically analyze the offset pattern.

[0013] Another aspect of the present invention provides an intelligent forklift management system based on AI anti-collision, which is used to execute an intelligent forklift management method based on AI anti-collision, including: a server and detection sensors; The server is configured to: Obtain the three-dimensional data of the forklift, the forklift attribute data, the historical working data of the forklift, and the historical working environment data of the forklift, and construct a three-dimensional model of the forklift according to the three-dimensional data of the forklift; Generate a digital twin system of the forklift according to the three-dimensional model of the forklift, the forklift attribute data, the historical working data of the forklift, and the historical working environment data of the forklift; Determine the size of the passing restricted space of the forklift according to the digital twin system of the forklift; Determine the size of the driving detection area of the forklift according to the size of the passing restricted space; Determine the area to be detected in the driving direction according to the size of the driving detection area and the driving route of the forklift; Obtain the area data of the area to be detected; Determine the obstacle data according to the area data; Extract all the feature corner points of the obstacle from the obstacle data, select the feature control points, and determine whether the feature control points meet the preset requirements; If there are no feature control points that meet the preset requirements, it means that it is no longer possible to pass ahead. Control the forklift to execute an emergency stop and issue an alarm, and continue to run after the obstacle is removed; When there are feature control points that meet the preset requirements, use the feature control points that meet the preset requirements as the basic control points, combine the current pose information of the forklift, the preset route planning strategy, and the constraint conditions, obtain the remaining required control points, and generate an alternative path by combining the basic control points and the remaining required control points; Transmit the alternative path to the path tracking program and maintain a real-time detection state; Adjust the driving route of the forklift according to the alternative path.

[0014] Adopting the technical solution of the present invention, an intelligent forklift management method based on AI anti-collision includes: obtaining the three-dimensional data of the forklift, the forklift attribute data, the forklift historical working data, and the forklift historical working environment data, and constructing a three-dimensional model of the forklift according to the three-dimensional data of the forklift; generating a digital twin system of the forklift according to the three-dimensional model of the forklift, the forklift attribute data, the forklift historical working data, and the forklift historical working environment data; determining the size of the passing restricted space of the forklift according to the digital twin system of the forklift; determining the size of the driving detection area of the forklift according to the size of the passing restricted space; determining the area to be detected in the driving direction according to the size of the driving detection area and the driving route of the forklift; obtaining the area data of the area to be detected; determining the obstacle data according to the area data; extracting all the feature corner points of the obstacle from the obstacle data, selecting the feature control points, and determining whether the feature control points meet the preset requirements; if there are no feature control points that meet the preset requirements, it means that it is no longer possible to pass ahead, controlling the forklift to execute an emergency stop and issue an alarm, and continuing to run after the obstacle is removed; when there are feature control points that meet the preset requirements, using the feature control points that meet the preset requirements as the basic control points, combining the current pose information of the forklift, the preset route planning strategy, and the constraint conditions, obtaining the remaining required control points, and generating an alternative path by combining the basic control points and the remaining required control points; transmitting the alternative path to the path tracking program and maintaining a real-time detection state; adjusting the driving route of the forklift according to the alternative path. By comprehensively using AI technology, digital twin, and real-time data processing, an efficient and safe intelligent forklift management system is constructed, which can not only improve the working efficiency of the forklift, but also significantly reduce the safety risk, and has a wide application prospect. Description of the Drawings

[0015] Figure 1 is a flowchart of an intelligent forklift management method based on AI anti-collision provided by an embodiment of the present invention; Figure 2 is a schematic block diagram of an intelligent forklift management system based on AI anti-collision provided by an embodiment of the present invention. Detailed Embodiments

[0016] In order to more clearly understand the above-mentioned objects, features, and advantages of the present invention, the present invention will be further described in detail below with reference to the drawings and specific embodiments. It should be noted that, without conflict, the embodiments of the present application and the features in the embodiments may be combined with each other.

[0017] In the following description, many specific details are set forth in order to fully understand the present invention. However, the present invention may be implemented in other ways different from those described herein. Therefore, the protection scope of the present invention is not limited by the specific embodiments disclosed below.

[0018] In the description and claims of this application and the above-mentioned drawings, terms such as "first", "second", etc. are used to distinguish different objects, rather than to describe a specific order. In addition, the terms "comprising" and "having" and any variations thereof are intended to cover non-exclusive inclusion. For example, a process, method, system, product or device that includes a series of steps or units is not limited to the listed steps or units, but optionally further includes steps or units not listed, or optionally further includes other steps or units inherent to these processes, methods, products or devices.

[0019] Reference to "embodiment" herein means that a particular feature, structure, or characteristic described in connection with the embodiment can be included in at least one embodiment of this application. The phrase appears in various places in the specification and does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment mutually exclusive with other embodiments. Those skilled in the art will explicitly and implicitly understand that the embodiments described herein can be combined with other embodiments.

[0020] Reference is made below Figures 1 to 2 to describe a smart forklift management system and method based on AI anti-collision provided according to some embodiments of the present invention.

[0021] As Figure 1 shown, an embodiment of the present invention provides a smart forklift management method based on AI anti-collision, including: Obtain the three-dimensional data, attribute data, historical working data, and historical working environment data of the forklift, and construct a three-dimensional model of the forklift according to the three-dimensional data of the forklift; In this step, three-dimensional geometric data of the forklift is obtained using technologies such as laser scanning, structured light, or photogrammetry as the three-dimensional data of the forklift (these technologies can capture the shape and size information of the forklift and generate high-precision three-dimensional point cloud data); basic information of the forklift is collected, including model, load capacity, power type (electric or internal combustion), manufacturer, etc. (these data can usually be obtained through the technical manual of the forklift or the API provided by the manufacturer) as the forklift attribute data; through the control system or management software of the forklift, historical usage records of the forklift are obtained, including working hours, operation frequency, fault records, etc. (these data help analyze the usage efficiency and maintenance requirements of the forklift) as the forklift historical working data; relevant data of the forklift working environment is monitored, such as temperature, humidity, ground type, obstacle distribution, etc. (these data can be obtained through environmental sensors or historical record systems) as the forklift historical working environment data. The obtained data is cleaned to remove noise and redundant information, and the data from different sources is integrated into a unified database; the obtained three-dimensional data is converted into a visual three-dimensional model using three-dimensional modeling software (such as Blender, AutoCAD, or SolidWorks) (point cloud processing technology (such as PCL library) can be used to convert point cloud data into a mesh model); the forklift attribute data and the forklift historical working environment data are associated with the three-dimensional model to form a complete digital twin model (this can be achieved through a database management system (such as MySQL or MongoDB)); through actual measurement and comparison, the accuracy and reliability of the generated three-dimensional model are verified (virtual reality (VR) technology can be used for simulation testing); according to the verification results, the model is optimized to ensure its effectiveness and accuracy in practical applications. Through this step, relevant data of the forklift can be effectively obtained and an accurate three-dimensional model can be constructed, providing support for subsequent intelligent management and decision-making.

[0022] Generate a forklift digital twin system of the forklift according to the forklift three-dimensional model, the forklift attribute data, the forklift historical working data, and the forklift historical working environment data; In this step, the forklift attribute data, the forklift historical work data, and the forklift historical work environment data from different sources are converted into a unified format for subsequent processing and analysis; the obtained three-dimensional model of the forklift is imported into the digital twin platform or simulation software; the attribute data of the forklift (such as load capacity, power type, etc.) is associated with the three-dimensional model to ensure that the digital twin can reflect the actual performance of the forklift; the historical work data (such as usage frequency, fault records, etc.) and the work environment data (such as temperature, humidity, etc.) are integrated into the digital twin system for real-time monitoring and analysis; sensors (such as GPS, accelerometers, temperature sensors, etc.) are installed on the forklift to obtain the operating status and environmental changes of the forklift in real time. These data will continuously update the digital twin model to ensure that it reflects the current actual situation; data stream processing technologies (such as Apache Kafka or Apache Flink) are used to process and analyze the sensor data in real time to ensure that the digital twin system can quickly respond to environmental changes; the digital twin system is used for simulation to predict the performance of the forklift in different working environments; according to the simulation results and real-time data, the operating parameters and path planning of the forklift are adjusted to improve work efficiency and safety. Through this step, a digital twin system of the forklift can be effectively generated, thus realizing the intelligent management and optimized operation of the forklift.

[0023] Determine the size of the passing restricted space of the forklift according to the forklift digital twin system; Determine the size of the driving detection area of the forklift according to the size of the passing restricted space; Determine the area to be detected in the driving direction according to the size of the driving detection area and the driving route of the forklift; Obtain the area data (including image data, spatial position data, environmental data, etc.) of the area to be detected; In this step, necessary sensor devices are installed on the forklift and / or within the working area of the forklift, including cameras (for acquiring image data), lidar (for acquiring spatial position data), and environmental sensors (for acquiring environmental data such as temperature, humidity, etc.); the camera captures the image of the area to be detected in real time, and a binocular vision camera can be used to obtain higher-depth information and stereoscopic images; the lidar or other positioning systems (such as GPS, IMU, etc.) are used to obtain the spatial position and distribution information of the objects within the area to be detected, and these data can form a three-dimensional point cloud to help understand the spatial structure of the environment; the environmental sensors collect information such as temperature, humidity, and light intensity, and these data help analyze the impact of the environment on the detection task; the data obtained from different sensors are fused to form a comprehensive data set, which can be achieved through data processing algorithms (such as Kalman filtering, sensor fusion technology, etc.) to improve the accuracy and reliability of the data; during the data acquisition process, the changes in the area to be detected are monitored in real time, and the area data are dynamically updated according to the new data, which can be achieved through the embedded AI analysis module to ensure that the system can quickly respond to environmental changes; the obtained area data are stored in the database for subsequent analysis and processing. Data analysis tools can be used to deeply analyze the data to extract useful information and patterns. Through this step, the images, spatial positions, and environmental data of the area to be detected can be comprehensively obtained, providing a rich information basis for subsequent analysis; through data fusion and real-time update, the understanding and analysis accuracy of the area to be detected can be improved, thereby enhancing the efficiency and accuracy of the detection task.

[0024] Determine the obstacle data according to the said area data; Extract all the feature corner points of the obstacles from the said obstacle data, select the feature control points, and determine whether the feature control points meet the preset requirements; If there are no feature control points that meet the preset requirements, it means that it is no longer possible to pass ahead. Control the forklift to execute an emergency stop and issue an alarm, and continue to run after the obstacle is removed; In this step, during the operation of the forklift, the front environment is monitored in real time to extract feature control points. If the system detects that there are no feature control points that meet the preset requirements, it indicates that there are obstacles ahead. Once it is confirmed that passage ahead is impossible, the system should immediately control the forklift to perform an emergency stop, which can be achieved in the following ways: sending an emergency stop command to the forklift's control system to immediately cut off the power and ensure that the forklift stops within the shortest time; monitoring the current speed and position of the forklift to ensure the safety of the emergency stop process; while performing the emergency stop, the system should trigger the alarm mechanism to emit audible and visual alarms to alert surrounding personnel and other equipment, which can be achieved through the alarm device on the forklift; after the forklift stops, the system should continue to monitor the front environment to confirm the status of the obstacle. If the obstacle is removed or no longer affects the forklift's travel path, the system will be ready to resume operation; once it is confirmed that the obstacle is removed, the system should send a command to make the forklift resume operation, which includes: re-evaluating the current environment and starting the forklift after ensuring safety; continuing to execute the original travel route or adjusting the travel path according to the new environmental information. Through this step, through real-time monitoring and the emergency stop mechanism, it is possible to effectively avoid collisions between the forklift and obstacles, ensuring the safety of the operator and the surrounding environment; the system can quickly respond to environmental changes, handle emergencies in a timely manner, and reduce the likelihood of accidents; after the obstacle is removed, the forklift can quickly resume operation, reducing downtime and improving the overall operation efficiency.

[0025] When there are feature control points that meet the preset requirements, using the feature control points that meet the preset requirements as the basic control points, combining the current pose information of the forklift, the preset route planning strategy, and the constraint conditions, calculate the remaining required control points, and generate alternative paths by combining the basic control points and the remaining required control points. Transmit the alternative path to the path tracking program and maintain a real-time detection state. In this step, the generated alternative paths are transmitted to the path tracking program through a communication interface (such as CAN bus, Wi-Fi, or Bluetooth). This process needs to ensure the integrity and real-time nature of the data so that the path tracking program can quickly receive and process this information. The path tracking program needs to continuously monitor the current position and status of the forklift, which can be achieved through on-vehicle sensors (such as lidar, cameras, GPS, etc.) to obtain the motion state of the forklift and information about the surrounding environment in real time. During the real-time detection process, if an obstacle or other factors affecting driving are detected, the path tracking program should be able to dynamically adjust the current path. This may involve recalculating the path or selecting other options from the alternative paths. The system should establish a feedback mechanism to feedback the real-time state of the vehicle and the path tracking results to the path planning module for further optimization and adjustment. This feedback can help the system continuously improve the accuracy of path planning and tracking. Through this step, the real-time detection and dynamic adjustment mechanism can ensure that the vehicle always travels along the optimal path, reducing the risk of deviation and collision. The system can quickly respond to real-time environmental changes to ensure safe operation in complex and dynamic environments. Through effective path planning and real-time adjustment, the forklift or vehicle can complete tasks more efficiently, reducing downtime.

[0026] Adjust the driving route of the forklift according to the alternative path.

[0027] In this embodiment, by comprehensively utilizing AI technology, digital twin, and real-time data processing, an efficient and safe intelligent forklift management system is constructed, which can not only improve the working efficiency of the forklift but also significantly reduce safety risks, having broad application prospects.

[0028] In some possible implementation manners of the present invention, the step of determining the passing restricted space size of the forklift according to the forklift digital twin system includes: Utilize the three-dimensional model of the forklift in the forklift digital twin system to determine the basic passing restricted space size of the forklift; In this step, in the forklift digital twin system, a 3D model of the forklift is utilized to conduct boundary envelope calculations based on the 3D model of the forklift, analyzing the external dimensions, turning radius, working range, etc. of the forklift to ensure the accurate assessment of the forklift's movement capabilities within a specific space; according to the forklift's working environment data (such as aisle width, obstacle position, etc.), define the passing restricted space for the forklift (this space should consider the size and operating characteristics of the forklift to ensure safe passage); combine the 3D model of the forklift and the defined restricted space to calculate the maximum passing size of the forklift within this space (i.e., the size of the basic passing restricted space), which includes evaluating the space requirements of the forklift in different operating states (such as lifting goods, turning, etc.). Among them, conducting boundary envelope calculations based on the 3D model of the forklift includes: obtaining the static limit size parameters of the forklift (such as measuring the length, width, and height of the forklift body; measuring the maximum extended length of the fork arms; measuring the maximum lifting height of the fork arms); constructing the forklift movement envelope (such as calculating the horizontal movement trajectory under the minimum turning radius; calculating the vertical movement trajectory during the operation of the fork arms; generating a complete 3D movement envelope body). Among them, calculating the maximum passing size of the forklift within this space includes: establishing a space margin calculation model (including: setting a basic safety margin; dynamically adjusting the margin according to speed; adjusting the margin according to the load state); generating the final passing restricted space (including: combining the static envelope volume; adding a dynamic compensation space; superimposing a safety margin space); outputting standardized parameters (including: determining the minimum aisle width; determining the minimum aisle height; determining the minimum turning radius).

[0029] Simulate the working process of the forklift in the forklift digital twin system by using the forklift historical working data and the forklift historical working environment data, and determine the offset during the operation of the forklift during the simulation process; In this step, according to the forklift historical working data, analyze the load mass distribution, operating speed distribution, and steering angle distribution; calculate the tilt angle under different loads, braking distance under different speeds, and lateral offset under different steering angles based on the forklift historical working data, and establish a dynamic compensation model; analyze the influence of environmental factors by using the forklift historical working environment data, including: evaluating the road surface friction coefficient, identifying slope changes, detecting ground unevenness, to obtain road surface analysis data; identifying the positions of fixed obstacles, evaluating the activity rules of dynamic obstacles, determining the safe operation boundary, to obtain environmental constraint analysis data; simulate the working process of the forklift in the forklift digital twin system according to all the above analysis data and the dynamic compensation model, and determine the offset during the operation of the forklift during the simulation process.

[0030] Modify the size of the basic passing restricted space according to the offset to obtain the size of the passing restricted space.

[0031] In this step, the calculated offset is applied to the size of the basic passage restriction space. Specifically, it can be calculated by the following formula: the size of the new passage restriction space = the size of the basic passage restriction space + the offset, which can ensure that the new passage restriction space can reflect the actual operating ability of the forklift in the current environment; verify the modified passage restriction space to ensure its feasibility in actual operation, which can be completed through simulation or actual testing to ensure that the forklift can pass safely within the new restricted space. In this step, by dynamically adjusting the passage restriction space, the forklift can better adapt to different working environments and operating conditions; the real-time updated passage restriction space can effectively avoid collisions and accidents and improve the safety of the forklift.

[0032] In this embodiment, the boundary envelope calculation provides the basic space requirements, the dynamic characteristic analysis supplements the additional space requirements during the movement process, the environmental factor analysis provides the external constraint conditions, and the final passage restriction space is the result of comprehensive consideration of all the above factors. By accurately calculating the passage restriction space of the forklift, collisions of the forklift in narrow or complex environments can be effectively avoided, thereby improving the operation safety; after clarifying the passage restriction space of the forklift, the driving route and operation strategy of the forklift can be optimized, reducing unnecessary detours and waiting times and improving work efficiency.

[0033] In some possible embodiments of the present invention, the step of determining the size of the driving detection area of the forklift according to the size of the passage restriction space includes: Define the basic parameters of the driving detection area according to the operating characteristics of the forklift and the size of the passage restriction space (these parameters include the shape, size and position of the detection area); Perform an initial calculation according to the size of the restricted space and determine the size of the basic detection area. Specifically: determine the minimum passage width as the width of the basic detection area; determine the minimum passage height as the height of the basic detection area; determine the length of the basic detection area based on the minimum turning radius; Calculate the size of the driving detection area using the following formula: the size of the driving detection area = the size of the basic detection area + the safety margin; where the value of the safety margin is obtained through big data analysis based on the historical working data of the forklift (the safety margin is to ensure that the forklift has enough space for operation during driving to avoid collisions and accidents); When calculating the size of the driving detection area, adaptively adjust the size of the driving detection area according to the ground condition and environmental complexity (which can be obtained through sensor data or a real-time monitoring system); In this step, adjustments are made according to the ground conditions, specifically: considering the change in the friction coefficient (evaluating the current ground adhesion situation; adjusting the estimated braking distance; correspondingly adjusting the size of the detection area), and considering the influence of the slope (calculating the influence of the slope on braking; adjusting the longitudinal distance of the detection area). Adjustments are made according to the environmental complexity, specifically: evaluating the obstacle density; adjusting the accuracy requirements of the detection area; optimizing the shape of the detection area.

[0034] Verify the size of the calculated driving detection area through simulation tests to ensure that the forklift can drive safely within this detection area size; During the operation of the forklift, continuously monitor the environmental changes and dynamically adjust the size of the driving detection area according to the new data.

[0035] In this step, first, establish an optimization model for the detection area, including: setting the basic safety factor; establishing a dynamic adjustment strategy; defining the optimization objective function; second, perform adaptive adjustment of the detection area, including: calculating the optimal detection area parameters in real time; smoothly transitioning the change of the detection area; ensuring detection continuity; finally, output the standardized detection area parameters, including: generating a description of the detection area boundary; defining the priority of the detection area; providing a detection area update strategy.

[0036] This embodiment can effectively avoid collisions during the driving of the forklift by accurately calculating the size of the driving detection area, thereby improving the operation safety; after clarifying the driving detection area, the driving path and operation strategy of the forklift can be optimized, reducing unnecessary detours and waiting times, and improving work efficiency.

[0037] In some possible implementation manners of the present invention, the step of determining the area to be detected in the driving direction according to the size of the driving detection area and the driving route of the forklift includes: Determine the first area to be detected according to the size of the driving detection area and the driving route of the forklift; Obtain the driving state data and load state data of the forklift; Extract the current speed from the driving state data and dynamically adjust the detection area length of the first area to be detected according to the current speed to obtain a second detection area; In this step, calculate the braking distance at the current speed, calculate the reaction time safety margin according to the braking distance, generate a speed-related detection area adjustment length according to the reaction time safety margin, and adjust the detection area length of the first area to be detected according to the speed-related detection area adjustment length to obtain a second detection area.

[0038] Extract the steering angle from the driving state data and adjust the detection area shape of the second area to be detected according to the steering angle to obtain a third area to be detected; In this step, the steering prediction trajectory is calculated based on the steering angle, and the arc expansion of the second detection area is adjusted according to the steering prediction trajectory to ensure that all possible movement areas are covered, resulting in the third area to be detected.

[0039] According to the load status data, the influence data of the load on the braking performance is calculated, and the longitudinal distance and lateral range of the detection area of the third detection area are adjusted according to the influence data, resulting in the area to be detected.

[0040] The solution of this embodiment can dynamically adjust the length and shape of the area to be detected, enabling the forklift to more effectively identify and avoid obstacles during driving, reducing the collision risk; it can adjust the detection area according to the real-time driving state and load conditions to ensure the safety and stability of the forklift under different operating conditions; through the precise setting of the area to be detected, the forklift can perform material handling more efficiently in a complex environment, reducing the downtime and maintenance costs caused by collisions. The solution of this embodiment can significantly improve the operating ability of the forklift in dynamic and complex environments, providing strong support for the intelligent forklift management system.

[0041] In some possible implementation manners of the present invention, the step of determining the obstacle data according to the area data includes: Extracting the area image data of the area to be detected from the area data (removing the images beyond the distance range); In this step, image information is extracted from the data of the area to be detected, ensuring that the images beyond the set distance range are removed. The purpose of this step is to focus on the area related to forklift operation and reduce unnecessary data processing. Extracting the area image data of the area to be detected from the area data may specifically include: setting a maximum detection distance threshold, establishing a distance calculation method based on depth information, constructing a distance filtering algorithm, and thus establishing a distance threshold judgment model; obtaining the original image data of the area data; combining the distance threshold judgment model, performing depth information registration on the original image data and removing the image areas beyond the distance threshold to obtain the area image data.

[0042] Inputting the area image data into a preset obstacle detection model; In this step, the extracted area image data is input into a preset obstacle detection model. This model is based on a deep learning algorithm and can effectively identify and classify obstacles in the image. Inputting the area image data into a preset obstacle detection model includes: performing data standardization processing (unifying the image size; normalizing the pixel values; adjusting the image format); establishing a data batch processing mechanism (setting the processing batch size; organizing the data input queue; implementing a parallel processing strategy); and inputting the processed area image data into the preset obstacle detection model.

[0043] The obstacle detection model performs obstacle recognition and extraction on the regional image data to obtain obstacle image data; In this step, the obstacle detection model processes the input regional image data, identifies the obstacles therein, and extracts the image data of the obstacles. This process is the key to realizing intelligent anti-collision and can real-time identify potential dangers around the forklift. The identification and extraction using the obstacle detection model include: extracting color features, texture features, and shape features from the regional image data; constructing a feature pyramid based on the color features, texture features, and shape features; performing sliding window detection and merging the detection results; calculating the overlap degree of detection frames; screening the optimal detection results; and outputting the obstacle image data.

[0044] Perform obstacle segmentation based on the obstacle image data to obtain obstacle segmentation data; In this step, perform obstacle segmentation based on the extracted obstacle image data to obtain the segmentation data of the obstacles. This step separates the obstacles from the background through image processing technology for subsequent analysis. Performing obstacle segmentation includes: establishing segmentation preprocessing (performing edge enhancement; performing region growing; applying morphological processing); implementing precise segmentation (performing region segmentation; optimizing the boundary; generating a segmentation mask).

[0045] Calculate the obstacle contour and center based on the obstacle segmentation data, and perform coordinate transformation on the calculation results to obtain the coordinate sets of the obstacle contour points and the center point.

[0046] In this step, based on the segmentation data of the obstacles, calculate the contour and center point of the obstacles, and perform coordinate transformation on the calculation results. This process converts the geometric information of the obstacles into a coordinate set that can be used for forklift navigation and obstacle avoidance. Calculating the obstacle contour and center includes: Contour extraction and processing: Detecting the contour boundary; Smoothing the contour curve; Extracting key contour points; Center point calculation: Calculating the centroid coordinates; Verifying the validity of the center point; Recording the position of the center point; Coordinate system transformation: Establishing a coordinate transformation model; Performing coordinate mapping; Outputting the final coordinate set.

[0047] In this embodiment, the system can accurately identify and locate the obstacles around the forklift, thereby improving the safety and operation efficiency of the forklift. It not only relies on advanced image processing technology but also combines the powerful capabilities of AI algorithms to ensure effective collision prevention in complex environments.

[0048] In some possible implementation manners of the present invention, the step of extracting all feature corner points of the obstacle from the obstacle data, selecting feature control points, and determining whether the feature control points meet the preset requirements includes: Extract the geometric feature information of the obstacle from the obstacle data (including contour, edge, and other relevant data); Use feature detection algorithms (such as Harris corner detection, FAST algorithm, or SIFT algorithm, etc.) to process the geometric feature information and extract all feature corner points (these corner points are usually where the shape of the obstacle changes significantly and can effectively describe the geometric features of the obstacle); In this step, extracting the feature corner points of the obstacle includes: Perform corner detection preprocessing: Denoise the image; Enhance edge features; Adjust contrast; Apply the corner detection algorithm: Calculate the image gradient; Construct the corner response function; Perform non-maximum suppression; Conduct preliminary corner point screening: Set the response threshold; Filter weak corner points; Remove duplicate corner points.

[0049] In the implementation example of the present invention, it may further include the step of optimizing the feature corner points, specifically: Calculate the corner point feature descriptor: Extract local region features; Construct the feature vector; Standardize the feature description; Perform feature matching: Establish the feature similarity metric; Find the best matching points; Verify the reliability of the match; Conduct position refinement: Sub-pixel level corner point localization; Update the corner point coordinates; Verify the localization accuracy.

[0050] Select the feature control points according to the distribution and importance of the feature corner points; It can be understood that some criteria (such as the response value of the corner point, the distance from other corner points, etc.) can be set to screen out the most representative control points, and these control points will be used for subsequent obstacle recognition and positioning. In this step, the selection of feature control points includes: Establish the control point selection criteria: Define the importance scoring rules; Set the selection threshold; Establish the priority sorting mechanism; Perform clustering analysis: Calculate the distance between corner points; Form corner point clusters; Determine the representative points of the clusters; Select based on scene understanding: Analyze the structural features of the obstacle; Identify the key turning points; Determine the candidate set of control points.

[0051] Verify the selected feature control points to determine whether they meet the preset requirements; In this step, the preset requirements may include: Whether the number of control points is within a reasonable range; Whether the distribution of control points is uniform and covers the key areas of the obstacle; Whether the geometric features of the control points meet the expectations (such as angles, distances, etc.).

[0052] Output the feature control points that meet the preset requirements and their coordinates for subsequent obstacle recognition, path planning, or obstacle avoidance decision-making.

[0053] In the solution of this embodiment, by extracting and selecting feature control points, the system can more accurately identify and locate obstacles, reducing the possibility of misidentification; the effective selection and verification of feature control points can improve the system's adaptability to complex environments, ensuring the safe operation of the forklift in a dynamic environment; accurate feature control points provide reliable data support for subsequent path planning and obstacle avoidance decision-making, improving the operation efficiency and safety of the forklift.

[0054] In some possible implementation manners of the present invention, the step of, when there are feature control points that meet the preset requirements, using the feature control points that meet the preset requirements as the basic control points, combining the current pose information of the forklift, the preset route planning strategy, and the constraint conditions, obtaining the remaining required control points, and generating an alternative path by combining the basic control points and the remaining required control points includes: Use the feature control points that meet the preset requirements as the basic control points; Obtain the current pose information from the sensor system of the forklift, including the position (coordinates) and attitude (direction) of the forklift; Determine the preset route planning strategy; It can be understood that this may include the shortest path algorithm, obstacle avoidance strategy, smooth path generation, etc.; select a suitable strategy to adapt to the current environment and task requirements.

[0055] Clarify the constraint conditions in path planning (such as the maximum turning radius, minimum safety distance, speed limit, etc.; these constraint conditions will affect the selection of control points and the generation of paths); According to the basic control points, the current pose information of the forklift, the route planning strategy, and the constraint conditions, calculate the remaining required control points (this can be achieved through path interpolation algorithms or optimization algorithms to ensure the coherence and feasibility of the path); In this step, calculate the remaining required control points: Basic control point analysis: Extract the coordinates of the basic control points; calculate the distance between the basic control points; evaluate the distribution characteristics; Construct a point set supplement strategy: Combine the current pose information of the forklift, the route planning strategy, and the constraint conditions to predict the first driving trajectory of the forklift; Identify the sparse area of control points on the first driving trajectory; Calculate the expected point spacing; Determine the number and position of supplementary points; Generate the remaining required control points: Based on the number and position of the supplementary points, apply the interpolation algorithm to optimize the point distribution to obtain the remaining required control points, and verify the rationality of the point positions.

[0056] Combine the basic control points and the remaining required control points to generate an alternative path.

[0057] In this step, a path smoothing algorithm (such as B-spline or Bezier curve) can be used to optimize the path to make it smoother and more natural; specifically, it includes: Path fitting calculation: Select the type of spline curve; Calculate the curve parameters; Generate the path curve; Path smoothing process: Eliminate the mutation points; Optimize the curvature distribution; Ensure continuity; Preliminary feasibility verification: Check the motion constraints; Verify the collision risk; Evaluate the path quality; Establish optimization objectives: Define the path length term; Define the smoothness term; Define the safety term; Execute iterative optimization: Calculate the objective function value; Update the path parameters; Check the convergence condition; Generate multiple alternative paths: Adjust the optimization parameters; Calculate multiple paths in parallel; Screen the feasible solutions; Establish an evaluation index system: Calculate the path length; Evaluate the smoothness; Analyze the safety margin; Execute comprehensive scoring: Set the index weights; Calculate the comprehensive score; Sort the alternative solutions; Output the final path set: Select the optimal path; Save the alternative solutions; Generate the path description.

[0058] Path verification: Verify the generated alternative paths to ensure that they meet all the constraint conditions and are feasible in actual operation.

[0059] The solution of this embodiment, combined with the characteristic control points and the current pose information of the forklift, can generate a more accurate and effective path, reducing errors; By setting the constraint conditions and adjusting the control points in real time, it ensures the safe driving of the forklift in a complex environment and avoids collisions; The generated alternative paths can improve the operation efficiency of the forklift, reduce the driving time and energy consumption, and improve the overall operation efficiency.

[0060] In some possible implementation manners of the present invention, the step of extracting the region image data of the region to be detected from the region data includes: Set the maximum detection distance threshold, establish a distance calculation method based on depth information, and construct a distance filtering algorithm to establish a distance threshold judgment model; It is understandable that, according to the application scenario and safety requirements, a maximum detection distance threshold (e.g., 2 meters, 5 meters, etc.) can be set, which will be used to determine whether the object is within the detectable range; dynamically adjust the threshold according to actual tests and environmental changes to ensure the sensitivity and accuracy of the system; use depth sensors (such as lidar, depth cameras, etc.) to obtain the depth information of the environment, which can provide distance data from each pixel to the object; calculate the actual distance between the object and the sensor through the pixel values ​​in the depth image, which can be achieved using simple geometric formulas or deep image processing algorithms; pre-process the acquired depth data, including denoising, smoothing, etc., to improve the quality of the data; implement a filtering algorithm to check whether each calculated distance is less than the set maximum detection distance threshold; and establish a judgment model based on the output of the filtering algorithm. The model can be a simple rule engine or a machine learning-based model, depending on the complexity of the application; if a machine learning model is used, it is necessary to collect labeled data for training. The model can learn distance judgment under different environments and conditions; integrate the model into a real-time system to continuously monitor the environment and determine whether the object is within the detectable range based on the depth information. By setting a reasonable distance threshold and using depth information, the distance of an object can be accurately determined; objects beyond the detection range can be filtered out in a timely manner to reduce potential collision risks; and through effective distance filtering, unnecessary calculations and processing can be reduced, thereby improving system efficiency.

[0061] Acquire original image data of the regional data; In combination with the distance threshold judgment model, the original image data is registered for depth information and the image area exceeding the distance threshold is eliminated to obtain regional image data.

[0062] In this embodiment, image information is extracted from the data of the area to be detected, ensuring that images beyond a set distance range are removed, focusing on areas related to forklift operation, and reducing unnecessary data processing.

[0063] In some possible implementations of the present invention, the step of simulating the working process of a forklift in the forklift digital twin system using the forklift historical working data and the forklift historical working environment data, and determining the offset of the forklift during operation during the simulation process, includes: Process and analyze the historical working data of forklifts, including: cleaning abnormal data; standardizing data format; extracting key features; Process and analyze the historical working environment data of forklifts, including: reconstructing historical scenes; extracting environmental features; and building environmental models; Conduct data association analysis on the forklift's historical working data and forklift's historical working environment data, including: establishing time-space correspondence; identifying key influencing factors; and building a data association model; Construct a simulation scenario based on the forklift digital twin system, including: Load the basic model: Load the 3D model of the forklift; Set physical parameters; Configure dynamic characteristics; Recreate the environmental scenario: Construct the ground model; Add obstacles; Set environmental parameters; Simulate interaction relationships: Define collision detection; Set the friction coefficient; Configure dynamic responses; Conduct dynamic simulation, including: Establish a dynamic model: Construct a mass distribution model; Set inertia parameters; Define joint constraints; Conduct kinematic analysis: Calculate steering characteristics; Analyze acceleration characteristics; Simulate the braking process; Analyze the influence of load: Simulate load changes; Calculate the center of gravity offset; Analyze stability; Calculate the offset, including: Determine the reference trajectory: Extract the ideal path; Set key nodes; Generate a reference trajectory; Simulate the actual trajectory: Execute dynamic simulation; Record the movement trajectory; Extract position data; Offset analysis: Calculate the position deviation; Analyze the attitude deviation; Statistically analyze the offset law.

[0064] In this embodiment, data preprocessing provides basic data support for the simulation; Scenario construction provides environmental conditions for dynamic simulation; Dynamic simulation generates actual motion data; Offset calculation analyzes the simulation results; Model optimization ensures simulation accuracy. The solution of this embodiment is based on accurate modeling of historical data, comprehensive consideration of multi-factor influences, accurate prediction of dynamic offsets; An efficient data processing mechanism, parallel simulation calculations, and fast result analysis; Scenario adaptation ability, adaptation to load changes, adaptation to environmental conditions; A complete verification mechanism, continuous optimization and update, and stable performance output. This solution realizes the accurate prediction of the forklift operation offset by constructing a complete digital twin simulation system, providing a reliable compensation basis for actual operation. At the same time, through a continuous model optimization mechanism, the accuracy of the prediction results and the stable operation of the system are ensured.

[0065] Please refer to Figure 2 , Another embodiment of the present invention provides an intelligent forklift management system based on AI anti-collision, which is used to execute an intelligent forklift management method based on AI anti-collision, including: a server and detection sensors; The server is configured to: Obtain the 3D data of the forklift, forklift attribute data, historical working data of the forklift, and historical working environment data of the forklift, and construct a 3D model of the forklift according to the 3D data of the forklift; Generate a forklift digital twin system for the forklift based on the three-dimensional model of the forklift, the forklift attribute data, the historical working data of the forklift, and the historical working environment data of the forklift; Determine the size of the passing restricted space of the forklift according to the forklift digital twin system; Determine the size of the driving detection area of the forklift according to the size of the passing restricted space; Determine the area to be detected in the driving direction according to the size of the driving detection area and the driving route of the forklift; Obtain the area data of the area to be detected; Determine the obstacle data according to the area data; Extract all the characteristic corner points of the obstacle from the obstacle data, select the characteristic control points, and determine whether the characteristic control points meet the preset requirements; If there are no characteristic control points that meet the preset requirements, it means that it is no longer possible to pass ahead. Control the forklift to execute an emergency stop and issue an alarm, and wait for the obstacle to be removed before continuing to run; When there are characteristic control points that meet the preset requirements, use the characteristic control points that meet the preset requirements as the basic control points, combine the current pose information of the forklift, the preset route planning strategy and the constraint conditions, obtain the remaining required control points, and generate an alternative path by combining the basic control points and the remaining required control points; Transmit the alternative path to the path tracking program and maintain a real-time detection state; Adjust the driving route of the forklift according to the alternative path.

[0066] It should be known that Figure 2 The block diagram of the intelligent forklift management system based on AI anti-collision shown is only for illustration, and the number of each module shown does not limit the protection scope of the present invention. The intelligent forklift management system based on AI anti-collision provided in this embodiment can be used to execute the corresponding implementation solutions of the intelligent forklift management method based on AI anti-collision. For the specific implementation process, please refer to the description of each method embodiment, which will not be elaborated here.

[0067] It should be noted that for the foregoing method embodiments, for the sake of simple description, they are all expressed as a series of action combinations. However, those skilled in the art should know that this application is not limited by the described action sequence, because according to this application, certain steps can be performed in other sequences or simultaneously. Secondly, those skilled in the art should also know that the embodiments described in the specification are all preferred embodiments, and the actions and modules involved are not necessarily essential to this application.

[0068] In the above embodiments, the descriptions of each embodiment have their own emphases. For the parts not detailed in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0069] In several embodiments provided by the present application, it should be understood that the disclosed device can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the above division of units is only a logical function division. In actual implementation, there may be other division methods. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the displayed or discussed coupling or direct coupling or communication connection between each other can be through some interfaces. The indirect coupling or communication connection of the device or unit can be in electrical or other forms.

[0070] The units described above as separate components may or may not be physically separated. The components displayed as units may or may not be physical units, that is, they can be located in one place, or they can be distributed to multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0071] In addition, in each embodiment of the present application, each functional unit can be integrated in a processing unit, or each unit can exist physically alone, or two or more units can be integrated in one unit. The above integrated unit can be implemented in the form of hardware or in the form of a software functional unit.

[0072] If the above integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable memory. Based on this understanding, the technical solution of the present application, in essence, or the part that contributes to the prior art, or all or part of this technical solution, can be embodied in the form of a software product. This computer software product is stored in a memory and includes several instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the above methods in each embodiment of the present application. And the aforementioned memory includes: USB flash drive, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), mobile hard disk, magnetic disk, or optical disk, etc., which can store program codes.

[0073] Those of ordinary skill in the art can understand that all or part of the steps in the above methods of the embodiments can be completed by instructing relevant hardware through a program. This program can be stored in a computer-readable memory. The memory can include: flash drive, read-only memory (abbreviation: ROM), random access memory (abbreviation: RAM), magnetic disk, or optical disk, etc.

[0074] The above has introduced the embodiments of the present application in detail. Specific examples are used in this article to elaborate on the principles and implementation manners of the present application. The description of the above embodiments is only used to help understand the method and its core idea of the present application; at the same time, for those of ordinary skill in the art, according to the idea of the present application, there will be changes in the specific implementation manners and application scopes. In summary, the content of this specification should not be construed as a limitation to the present application.

[0075] Although the present invention is disclosed as above, the present invention is not limited thereto. Any person skilled in the art can easily think of changes or substitutions without departing from the spirit and scope of the present invention, and can make various modifications and alterations, including combinations of the above different functions and implementation steps, including software and hardware implementation manners, all within the protection scope of the present invention.

Claims

1. A smart forklift management method based on AI collision avoidance, characterized in that: include: Acquire forklift three-dimensional data, forklift attribute data, forklift historical working data and forklift historical working environment data of the forklift, and construct a forklift three-dimensional model of the forklift according to the forklift three-dimensional data; generating a forklift digital twin system of the forklift according to the forklift three-dimensional model, the forklift attribute data, the forklift historical working data and the forklift historical working environment data; Determining the size of the restricted space for the forklift to pass through according to the forklift digital twin system; Determining the size of the forklift's driving detection zone according to the size of the restricted space; Determine the area to be detected in the driving direction according to the size of the driving detection area and the driving route of the forklift; Acquire regional data of the area to be detected; determining obstacle data based on the area data; Extracting all characteristic corner points of the obstacle from the obstacle data, selecting characteristic control points, and determining whether the characteristic control points meet preset requirements; If there is no characteristic control point that meets the preset requirements, it means that the road ahead is blocked, and the forklift is controlled to perform an emergency stop and sound an alarm, and continue to operate after the obstacle is removed; When there are feature control points that meet the preset requirements, the feature control points that meet the preset requirements are used as basic control points, and the remaining required control points are obtained by combining the current posture information of the forklift, the preset route planning strategy and constraints, and an alternative path is generated by combining the basic control points and the remaining required control points; Transmitting the candidate path to a path tracking program and maintaining a real-time detection state; The driving route of the forklift is adjusted according to the alternative path.

2. The intelligent forklift management method based on AI collision avoidance according to claim 1 is characterized in that: The step of determining the size of the restricted space for the forklift to pass through according to the forklift digital twin system comprises: In the forklift digital twin system, the forklift 3D model is used to determine the size of the basic restricted space for the forklift to pass through; Using the forklift historical working data and the forklift historical working environment data to simulate the working process of the forklift in the forklift digital twin system, and determining the offset of the forklift during operation during the simulation process; The basic pass-limit space size is modified according to the offset to obtain the pass-limit space size.

3. The intelligent forklift management method based on AI collision avoidance according to claim 2 is characterized in that: The step of determining the size of the forklift's driving detection zone according to the size of the restricted space comprises: Define the basic parameters of the driving detection zone according to the operating characteristics of the forklift and the size of the restricted space; Perform initial calculations based on the size of the restricted space and determine the size of the basic detection area, specifically: determine the minimum channel width as the basic detection area width; determine the minimum channel height as the basic detection area height; determine the basic detection area length based on the minimum turning radius; The size of the driving detection zone is calculated using the following formula: Driving detection zone size = basic detection zone size + safety margin; the value of the safety margin is obtained by big data analysis based on the historical working data of the forklift; When calculating the size of the driving detection area, the size of the driving detection area is adaptively adjusted according to the ground conditions and environmental complexity; Verify the calculated driving detection zone size through simulation tests to ensure that the forklift can drive safely within the detection zone size; During the operation of the forklift, environmental changes are continuously monitored and the size of the driving detection area is dynamically adjusted based on new data.

4. The intelligent forklift management method based on AI collision avoidance according to claim 3 is characterized in that: The step of determining the area to be detected in the driving direction according to the size of the driving detection area and the driving route of the forklift includes: Determine a first area to be detected according to the size of the driving detection area and the driving route of the forklift; Obtaining driving status data and load status data of the forklift; Extracting the current speed from the driving state data, dynamically adjusting the length of the detection area of ​​the first to-be-detected area according to the current speed, to obtain a second detection area; Extracting a steering angle from the driving state data, and adjusting a detection area shape of the second to-be-detected area according to the steering angle to obtain a third to-be-detected area; According to the load state data, the influence data of the load on the braking performance is calculated, and the longitudinal distance and the lateral range of the detection area of ​​the third detection area are adjusted according to the influence data to obtain the area to be detected.

5. The intelligent forklift management method based on AI collision avoidance according to claim 4 is characterized in that: The step of determining obstacle data according to the area data comprises: Extracting regional image data of the area to be detected from the regional data; Inputting the regional image data into a preset obstacle detection model; The obstacle detection model performs obstacle recognition and extraction on the area image data to obtain obstacle image data; Performing obstacle segmentation according to the obstacle image data to obtain obstacle segmentation data; The obstacle contour and center are calculated according to the obstacle segmentation data, and the calculation results are subjected to coordinate transformation to obtain a coordinate set of obstacle contour points and a center point.

6. The intelligent forklift management method based on AI collision avoidance according to claim 5 is characterized in that: The step of extracting all characteristic corner points of the obstacle from the obstacle data, selecting characteristic control points, and determining whether the characteristic control points meet preset requirements includes: Extracting geometric feature information of the obstacle from the obstacle data; Use feature detection algorithm to process geometric feature information and extract all feature corner points; Select feature control points based on the distribution and importance of feature corner points; Verify the selected feature control points to determine whether they meet the preset requirements; The characteristic control points and their coordinates that meet the preset requirements are output for subsequent obstacle identification, path planning or obstacle avoidance decisions.

7. The intelligent forklift management method based on AI collision avoidance according to claim 6 is characterized in that: The step of taking the feature control points meeting the preset requirements as the basic control points when there are feature control points meeting the preset requirements, combining the current posture information of the forklift, the preset route planning strategy and the constraint conditions, obtaining the remaining required control points, and generating the alternative path by combining the basic control points and the remaining required control points includes: The characteristic control points that meet the preset requirements are used as basic control points; Obtain current posture information from the forklift's sensor system, including the forklift's position and posture; Determine the preset route planning strategy; Clarify the constraints in path planning; Calculate the remaining required control points based on the basic control points, the current posture information of the forklift, the route planning strategy and the constraints; Combine the base control points with the remaining required control points to generate an alternative path.

8. The intelligent forklift management method based on AI collision avoidance according to claim 7 is characterized in that: The step of extracting the regional image data of the area to be detected from the regional data comprises: Set the maximum detection distance threshold, establish a distance calculation method based on depth information, and construct a distance filtering algorithm to establish a distance threshold judgment model; Acquire original image data of the regional data; In combination with the distance threshold judgment model, the original image data is registered for depth information and the image area exceeding the distance threshold is eliminated to obtain regional image data.

9. The intelligent forklift management method based on AI collision avoidance according to claim 8 is characterized in that: The step of simulating the working process of the forklift in the forklift digital twin system by using the forklift historical working data and the forklift historical working environment data, and determining the offset of the forklift during operation during the simulation process, comprises: Process and analyze the historical working data of forklifts, including: cleaning abnormal data; standardizing data format; extracting key features; Process and analyze the historical working environment data of forklifts, including: reconstructing historical scenes; extracting environmental features; and building environmental models; Conduct data association analysis on the forklift's historical working data and forklift's historical working environment data, including: establishing time-space correspondence; identifying key influencing factors; and building a data association model; A simulation scenario is constructed based on the forklift digital twin system, including: Load basic model: load the forklift 3D model; set physical parameters; configure dynamic characteristics; Environmental scene reproduction: build ground model; add obstacles; set environmental parameters; Interaction simulation: define collision detection; set friction coefficient; configure dynamic response; Perform dynamic simulations, including: Establish a dynamic model: construct a mass distribution model; set inertia parameters; define joint constraints; Perform kinematic analysis: calculate steering characteristics; analyze acceleration characteristics; simulate braking process; Load impact analysis: simulate load changes; calculate center of gravity shift; analyze stability; Calculate the offset, including: Determine the reference trajectory: extract the ideal path; set key nodes; generate the reference trajectory; Actual trajectory simulation: perform dynamic simulation; record motion trajectory; extract position data; Offset analysis: calculate position deviation; analyze posture deviation; statistically analyze offset rules.

10. An intelligent forklift management system based on AI collision avoidance, used to execute the intelligent forklift management method based on AI collision avoidance as claimed in any one of claims 1 to 9, characterized in that: include: Servers and detection sensors; The server is configured to: Acquire forklift three-dimensional data, forklift attribute data, forklift historical working data and forklift historical working environment data of the forklift, and construct a forklift three-dimensional model of the forklift according to the forklift three-dimensional data; generating a forklift digital twin system of the forklift according to the forklift three-dimensional model, the forklift attribute data, the forklift historical working data and the forklift historical working environment data; Determining the size of the restricted space for the forklift to pass through according to the forklift digital twin system; Determining the size of the forklift's driving detection zone according to the size of the restricted space; Determine the area to be detected in the driving direction according to the size of the driving detection area and the driving route of the forklift; Acquire regional data of the area to be detected; determining obstacle data based on the area data; Extracting all characteristic corner points of the obstacle from the obstacle data, selecting characteristic control points, and determining whether the characteristic control points meet preset requirements; If there is no characteristic control point that meets the preset requirements, it means that the road ahead is blocked, and the forklift is controlled to perform an emergency stop and sound an alarm, and continue to operate after the obstacle is removed; When there are feature control points that meet the preset requirements, the feature control points that meet the preset requirements are used as basic control points, and the remaining required control points are obtained by combining the current posture information of the forklift, the preset route planning strategy and constraints, and an alternative path is generated by combining the basic control points and the remaining required control points; Transmitting the candidate path to a path tracking program and maintaining a real-time detection state; The driving route of the forklift is adjusted according to the alternative path.

Images