An automated parking control method for ski resorts and a transport vehicle based on Internet of Things (IoT) sensing.

Abstract

This invention discloses an automatic parking control method and transport vehicle for ski resorts based on IoT sensing, belonging to the field of IoT control technology. By acquiring various IoT sensor information, this invention can comprehensively understand the surrounding environment of the ski resort transport vehicle. The fusion of ultrasonic, geomagnetic, visual, and lidar sensor information can accurately reflect the vehicle's status. An intelligent driving control processing algorithm is used to determine the initial vehicle environmental state vector, providing an accurate basis for subsequent decisions. Determining the parking path planning label helps to assess the risk of path intersection with the control strategy in advance. Based on this label, the decision to issue an automatic parking control strategy can be made, improving parking safety and efficiency, avoiding unnecessary collision risks, optimizing parking lot resource utilization, and adapting to the complex and ever-changing parking needs of ski resorts.

Description

Technical Field

[0001] This invention relates to the field of Internet of Things (IoT) control technology, specifically to an automatic parking control method and transport vehicle for ski resorts based on IoT sensing. Background Technology

[0002] Automated parking of transport vehicles in ski resort areas faces numerous challenges. Traditional parking systems often fail to comprehensively acquire information about the vehicle's surrounding environment, and relying on a single type of sensor is insufficient to accurately depict the complex ski resort environment. For example, snow may cover some ground markings or affect the normal operation of sensors, and a single sensor cannot accurately determine the vehicle's position and the situation of surrounding obstacles. Due to the lack of accurate initial environmental condition assessment of the vehicle, the risk of intersection with the control strategy path cannot be effectively estimated during parking path planning, easily leading to low parking efficiency, increased collision risk, and ineffective utilization of parking resources. Therefore, a technical solution is needed that can accurately analyze the vehicle's condition and rationally plan the parking path to solve the problems in automated parking of ski resort transport vehicles. Summary of the Invention

[0003] This invention provides an automatic parking control method and transport vehicle for ski resorts based on Internet of Things (IoT) sensing, which can solve or partially solve the technical problems involved in the background art.

[0004] This invention provides an automatic parking control method for ski resorts based on IoT sensing, applied to an IoT sensing control system. The method includes: acquiring first IoT sensing information of a transport vehicle to be controlled, wherein the transport vehicle to be controlled is a transport vehicle in the ski resort area; the first IoT sensing information includes ultrasonic sensing information, geomagnetic sensing information, visual sensing information, and lidar sensing information.

[0005] The intelligent driving control processing algorithm is used to process the first IoT sensor information of the vehicle to be controlled to determine the initial vehicle environment state vector of the vehicle to be controlled; the intelligent driving control processing algorithm is used to process the initial vehicle environment state vector of the vehicle to be controlled to determine the parking path planning label of the vehicle to be controlled, the parking path planning label of the vehicle to be controlled represents the probability that the vehicle to be controlled has a path intersection point with the automatic parking control strategy during the automatic parking process; based on the parking path planning label of the vehicle to be controlled, it is determined whether to issue the automatic parking control strategy to the vehicle to be controlled.

[0006] This invention provides a transport vehicle that is communicatively connected to an IoT sensing and control system. The transport vehicle is used to receive automatic parking control strategies issued by the IoT sensing and control system. The IoT sensing and control system includes at least one processor and a memory. The memory stores computer execution instructions. The at least one processor executes the computer execution instructions stored in the memory, causing the at least one processor to perform the above-described method.

[0007] This invention provides an IoT sensing and control system, including at least one processor and a memory; the memory stores computer execution instructions; the at least one processor executes the computer execution instructions stored in the memory, causing the at least one processor to perform the above-described method.

[0008] This invention provides a readable storage medium on which a program or instruction is stored, and when the program or instruction is executed by a processor, it implements the steps of the above method.

[0009] In this embodiment of the invention, by acquiring various IoT sensor information, the surrounding environment of the ski resort transport vehicle can be comprehensively understood. The fusion of ultrasonic sensor information, geomagnetic sensor information, visual sensor information, and lidar sensor information can accurately reflect the vehicle's status. The intelligent driving control processing algorithm determines the initial vehicle environmental state vector, providing an accurate basis for subsequent decisions. Determining the parking path planning label helps to assess the risk of intersection with the control strategy path in advance. Based on this label, the decision to issue an automatic parking control strategy can be made, improving parking safety and efficiency, avoiding unnecessary collision risks, optimizing parking lot resource utilization, and adapting to the complex and ever-changing parking needs of ski resorts. Attached Figure Description

[0010] Figure 1 This is a flowchart of an automatic parking control method for ski resorts based on Internet of Things (IoT) sensing, provided as an embodiment of the present invention.

[0011] Figure 2 This is a schematic diagram of the structure of an IoT sensing and control system provided in an embodiment of the present invention. Detailed Implementation

[0012] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the embodiments of the present invention.

[0013] In the embodiments of this invention, the terms "first," "second," etc., are used to distinguish similar objects and not to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that embodiments of this invention can be implemented in orders other than those illustrated or described herein, and the objects distinguished by "first," "second," etc., are generally of the same class, without limiting the number of objects; for example, a first object can be one or more. Furthermore, in the embodiments of this invention, "and / or" indicates at least one of the connected objects, and the character " / " generally indicates that the preceding and following objects have an "or" relationship.

[0014] Figure 1 An automatic parking control method for ski resorts based on Internet of Things (IoT) sensing is shown, which is applied to an IoT sensing control system. The method includes the following steps 110-140.

[0015] Step 110: The IoT sensing control system acquires the first IoT sensing information of the transport vehicle to be controlled, wherein the transport vehicle to be controlled is a transport vehicle in the ski resort area; the first IoT sensing information includes ultrasonic sensing information, geomagnetic sensing information, visual sensing information and lidar sensing information.

[0016] Step 120: The IoT sensing control system uses the intelligent driving control processing algorithm to process the first IoT sensing information of the transport vehicle to be controlled, so as to determine the initial vehicle environment state vector of the transport vehicle to be controlled.

[0017] Step 130: The IoT sensing control system uses the intelligent driving control processing algorithm to process the initial vehicle environment state vector of the vehicle to be controlled in order to determine the parking path planning label of the vehicle to be controlled. The parking path planning label of the vehicle to be controlled represents the probability that the vehicle to be controlled has a path intersection point with the automatic parking control strategy during the automatic parking process.

[0018] Step 140: The IoT sensor control system determines whether to issue the automatic parking control strategy to the vehicle under control based on the parking path planning tag of the vehicle under control.

[0019] In step 140, the parking path planning label of the vehicle to be controlled includes a first parking path planning label and a second parking path planning label. The first parking path planning label represents the probability that the vehicle to be controlled will match the path intersection point when an automatic parking control strategy is issued to the vehicle to be controlled. The second parking path planning label represents the probability that the vehicle to be controlled will match the path intersection point when an automatic parking control strategy is not issued to the vehicle to be controlled. Based on this, the step of determining whether to issue the automatic parking control strategy to the vehicle under control based on the parking path planning label of the vehicle under control further includes: if the first discrimination mechanism and the second discrimination mechanism are met, then the automatic parking control strategy is issued to the vehicle under control. The first discrimination mechanism is met when the weight of the first parking path planning label is greater than the weight of the second parking path planning label, and the second discrimination mechanism is met when the weight difference between the first parking path planning label and the second parking path planning label is greater than a set weight; if at least one of the first discrimination mechanism and the second discrimination mechanism is not met, then it is determined that the automatic parking control strategy will not be issued to the vehicle under control.

[0020] In this embodiment of the invention, the IoT sensor control system plays a key role in the automatic parking control of the ski resort transport vehicle. First, in step 110, the IoT sensor control system begins to acquire the first IoT sensor information of the transport vehicle to be controlled. In this embodiment of the invention, the transport vehicle to be controlled is a transport vehicle in the ski resort area, and the first IoT sensor information includes ultrasonic sensor information, geomagnetic sensor information, visual sensor information, and lidar sensor information.

[0021] For ultrasonic sensing information, in a ski resort environment, ultrasonic sensors can emit ultrasonic signals at a high frequency, such as 20-50 times per second. These signals are reflected back after encountering objects around the transport vehicle (such as other vehicles, snowdrifts, parking lot boundary facilities, etc.). By accurately measuring the time difference between the emitted signal and the received reflected signal, the distance between the transport vehicle and surrounding objects can be determined. For example, in a standard ski resort parking lot scenario, the effective detection range of ultrasonic sensors can reach 0.2-5 meters, with an accuracy down to the centimeter level, such as ±3 centimeters. This provides important data for the system to accurately assess the nearby environment around the transport vehicle.

[0022] Geomagnetic sensing information is acquired through geomagnetic sensors. These sensors are installed at specific locations on the transport vehicle or in corresponding areas of the parking lot. Due to the existence of the Earth's magnetic field, the surrounding magnetic field changes to varying degrees depending on the location of the transport vehicle within the parking lot, influenced by the vehicle's ferromagnetic components (such as the engine and chassis). For example, geomagnetic sensors can detect variations in magnetic field strength between 0.3 and 0.6 Gauss, and accurately determine the transport vehicle's position and orientation based on these geomagnetic variations. This method is unaffected by environmental factors such as light or snow accumulation, providing the system with stable position reference information.

[0023] Visual sensing information primarily relies on cameras installed around the transport vehicle or surveillance cameras in the parking lot. The cameras acquire images at a certain frame rate (e.g., 30 frames per second). In a ski resort environment, the cameras can capture the visual scene around the transport vehicle, including the layout of the parking lot (such as the division of parking spaces and the direction of aisles), the position and movement of other vehicles, and potential obstacles (such as temporarily placed ski equipment, snow piles, etc.). Using computer vision algorithms, the system can identify the shape, color, texture, and other features of various objects from these images, thereby determining their type and relative position. For example, the vision system can accurately identify other vehicles within a range of 5-20 meters from the transport vehicle and determine their direction of travel.

[0024] The sensing information from the lidar device is provided by the lidar sensor. Lidar constructs a three-dimensional point cloud map of the surrounding environment by emitting a laser beam and receiving reflected light. In complex environments like ski resorts, lidar can quickly scan a large area around a transport vehicle. For example, it can scan an area with a radius of 10-20 meters within 0.1-0.5 seconds, with a resolution of 1°-3°. By analyzing the time and intensity of the reflected light, lidar can accurately depict the outlines and heights of surrounding objects, clearly revealing both snow-covered objects and complex parking facility structures, providing the system with comprehensive and accurate environmental data.

[0025] Next, in step 120, the IoT sensing control system uses the intelligent driving control processing algorithm to process the acquired first IoT sensing information to determine the initial vehicle environment state vector of the transport vehicle to be controlled. The intelligent driving control processing algorithm is an AI algorithm model that comprehensively considers the characteristics and interrelationships of various sensing information. For example, the algorithm constructs a multi-dimensional vector to describe the initial environmental state of the transport vehicle based on distance data from ultrasonic sensing information, position and attitude data from geomagnetic sensing information, object recognition and position data from visual sensing information, and environmental contour data from lidar sensing information. This vector may include the distance vector between the transport vehicle and surrounding objects, the coordinate vector of the transport vehicle in the parking lot, and the type and distribution vector of surrounding objects. For example, in a specific scenario, the initial vehicle environment state vector can be represented as an array containing 10-20 elements, each element representing a different environmental state parameter. In this way, the system can comprehensively and accurately grasp the current environmental state of the transport vehicle, providing detailed data support for subsequent parking path planning.

[0026] Then, in step 130, the IoT sensing control system continues to process the initial vehicle environment state vector using the intelligent driving control processing algorithm to determine the parking path planning label for the transport vehicle to be controlled. This parking path planning label represents the probability that the transport vehicle to be controlled will have a path intersection point with the automatic parking control strategy during the automatic parking process. During the processing, the intelligent driving control processing algorithm analyzes various parameters in the initial vehicle environment state vector to predict the possible driving trajectory of the transport vehicle under different parking operations. For example, based on the current position of the transport vehicle, the parking situation of surrounding vehicles, and the layout of the parking lot, the algorithm will simulate multiple possible parking paths. In this process, the algorithm will consider the influence of various factors on the path, such as the width of the parking lot lane (e.g., the standard lane width is 3-5 meters), the spacing of other vehicles (generally requiring a safe distance of 0.5-1.5 meters), and the turning radius of the transport vehicle itself (the turning radius of different types of transport vehicles may be between 3-8 meters), etc. Through comprehensive analysis of these factors, the algorithm will calculate the intersection point of each possible path with the preset path in the automatic parking control strategy and convert it into a parking path planning label. The probability value in the parking path planning label can be represented by a number between 0 and 1. For example, 0.3 means there is a 30% probability that there is a path intersection.

[0027] Finally, in step 140, the IoT sensing control system determines whether to issue an automatic parking control strategy to the vehicle under control based on the parking path planning tag of the vehicle under control. In this embodiment, the parking path planning tag of the vehicle under control includes a first parking path planning tag and a second parking path planning tag. The first parking path planning tag represents the probability that the vehicle under control will match a path intersection point when an automatic parking control strategy is issued; the second parking path planning tag represents the probability that the vehicle under control will match a path intersection point when an automatic parking control strategy is not issued. In the specific judgment process, if both the first and second judgment mechanisms are met, an automatic parking control strategy is issued to the vehicle under control. Meeting the first judgment mechanism means that the weight of the first parking path planning tag is greater than the weight of the second parking path planning tag. In this embodiment, the weights can be values ​​set according to actual conditions, for example, the weight of the first parking path planning tag is 0.6, and the weight of the second parking path planning tag is 0.3. The second discrimination mechanism is met when the weight difference between the first and second parking path planning labels is greater than a set weight. For example, if the set weight is 0.2, the condition is met when the weight difference is 0.3 (0.6-0.3=0.3). If at least one of the first and second discrimination mechanisms is not met, it is determined that no automatic parking control strategy will be issued to the vehicle under control. This decision-making mechanism can effectively avoid path conflicts during parking, improving the safety and efficiency of automatic parking. Through this series of operations, the IoT sensor control system can accurately control and manage the automatic parking process of the vehicle under control in a ski resort environment.

[0028] Throughout the process, the intelligent driving control processing algorithm plays a central role. It requires continuous updates and optimization to adapt to the complex and ever-changing environment of ski resorts. For example, during peak ski season, with high vehicle traffic in parking lots, the algorithm needs to process sensor information more quickly and adjust parking path planning labels and decision-making strategies in a timely manner. Simultaneously, with the continuous development of sensor technology, such as the emergence of higher-resolution visual sensors and LiDAR with longer detection ranges, the intelligent driving control processing algorithm also needs corresponding improvements to better utilize this new data and enhance the accuracy and reliability of automatic parking control.

[0029] The IoT sensor control system also needs to consider the security of data transmission and storage. Since it involves the automatic parking control of numerous transport vehicles in the ski resort, a large amount of sensor information, vehicle status information, and control strategy information needs to be transmitted and stored in the system. To ensure data security, the system can employ encryption technologies to encrypt the data. For example, the AES (Advanced Encryption Standard) algorithm can be used to encrypt the data collected by the sensors. During data transmission, SSL (Secure Sockets Layer) or TLS (Transport Layer Security) protocols are used for secure transmission to prevent data theft or tampering. For data storage, a secure database management system is used, with strict access permissions set so that only authorized personnel or equipment can access and modify the relevant data.

[0030] Furthermore, system compatibility and scalability are also important considerations. With the development of ski resorts and the increase in the number of transport vehicles, the IoT sensor control system needs to be compatible with different types of sensors and transport vehicle equipment. For example, the system should be able to easily integrate newly introduced ultrasonic sensors with special functions or new types of electric transport vehicles into the control system. At the same time, the system also needs to have good scalability to allow for the addition of new functional modules in the future, such as integration with other intelligent management systems of the ski resort (e.g., ticketing systems, ski equipment rental systems, etc.) to achieve more comprehensive intelligent management.

[0031] The layout and installation of sensors also require careful design. For ultrasonic sensors, they need to be rationally distributed around the vehicle body according to its shape and size to ensure comprehensive close-range detection. For example, multiple ultrasonic sensors can be evenly installed on the front and rear bumpers and both sides of the vehicle to form a complete close-range detection network. The installation location of the geomagnetic sensors needs to consider areas with minimal magnetic field interference to ensure measurement accuracy. Visual sensors and lidar should be installed in positions that provide the best field of view, generally on the top of the vehicle or at a high position at the front, avoiding obstruction of vision by other objects.

[0032] Thus, through the above series of steps and technical means, effective control and management of the automatic parking process of the uncontrolled transport vehicle in the ski resort environment is achieved, improving the safety, efficiency and intelligence level of parking.

[0033] In one possible scenario, the intelligent driving control processing algorithm could be the Transformer algorithm. The following will detail the process by which the IoT sensor control system uses the Transformer algorithm to process the first IoT sensor information to determine the initial vehicle environmental state vector.

[0034] The Transformer algorithm is a deep learning model based on an attention mechanism. Its core advantage lies in its ability to effectively process long sequence data and capture long-distance dependencies within the data. In this embodiment of the invention, the first IoT sensing information of the transport vehicle can be regarded as a complex multimodal sequence data, and the Transformer algorithm can effectively uncover its inherent relationships.

[0035] I. Characteristics and Preprocessing of First-Earth Internet of Things Sensor Information

[0036] (a) Ultrasonic sensing information

[0037] Ultrasonic sensing information is presented as a series of distance values. For example, distances to surrounding objects (other vehicles, snowdrifts, parking lot boundaries, etc.) are measured at regular time intervals (e.g., every 0.1 seconds) in different directions around a ski transport vehicle. These distance values ​​may range from 0.1 meters to 10 meters, depending on the sensor's installation location and the surrounding environment.

[0038] Because ultrasonic sensors may have a certain measurement error (e.g., ±0.05 meters), the raw distance data needs to be filtered. A mean filtering method can be used, for example, taking the average of five consecutive measurements as the current valid distance value to reduce the influence of noise. Then, the processed distance values ​​are arranged according to the sensor's installation position to form a distance value sequence, which serves as the input to the Transformer algorithm.

[0039] (ii) Geomagnetic sensing information

[0040] The geomagnetic sensing information primarily reflects the changes in the transport vehicle's position relative to the Earth's magnetic field. It may include information such as magnetic field strength and direction. For example, the magnetic field strength value may range from 0.2 Gauss to 0.6 Gauss, and varies with the position and attitude of the transport vehicle.

[0041] First, the collected geomagnetic data is normalized, mapping the magnetic field strength values ​​to a range of 0 to 1, so that it can be processed on the same numerical scale as other sensor information. Then, based on the transport vehicle's direction of travel, the geomagnetic direction information is converted into a relative direction value relative to the vehicle, for example, converting the geomagnetic direction into a clockwise angle value (0° to 360°) relative to the vehicle's heading. The processed geomagnetic intensity and direction information are combined into a vector sequence, which is then used as input to the Transformer algorithm.

[0042] (III) Visual sensing information

[0043] Visual sensing information is image data captured by cameras. In a ski resort environment, images contain a wealth of information, such as the layout of the parking lot (parking lines, lane markings, etc.), the appearance and location of other vehicles, and the presence of skiers and their equipment. The image resolution may be 1920×1080 pixels, and each pixel contains information from the RGB color channels.

[0044] First, the image is cropped and scaled to a size suitable for processing by the Transformer algorithm, for example, to 224×224 pixels. Then, a Convolutional Neural Network (CNN) is used to extract features from the image. CNNs can extract features such as edges, textures, and object shapes. The extracted feature vectors are arranged in a specific order to form the input sequence of visual sensing information.

[0045] (iv) LiDAR sensing information

[0046] LiDAR sensing information is three-dimensional point cloud data. In a ski resort environment, it can accurately depict the three-dimensional shape, height, and distance of objects around the transport vehicle. Each point in the point cloud data contains three-dimensional coordinates (x, y, z) and information such as reflection intensity. For example, a single scan may yield tens of thousands of point cloud data points, with their coordinates possibly ranging within a spherical area with a radius of 10 meters centered on the transport vehicle.

[0047] First, the point cloud data is downsampled to reduce the data volume while retaining key shape information. A voxel grid filtering method can be used to divide the point cloud space into voxels of a certain size (e.g., 0.1m × 0.1m × 0.1m), retaining only one representative point within each voxel. Then, the processed point cloud data is converted into a format suitable for the Transformer algorithm, for example, by combining the point cloud coordinates and reflection intensity information into a vector sequence.

[0048] II. Input Construction for the Transformer Algorithm

[0049] 1) Multimodal data fusion

[0050] The preprocessed ultrasonic sensor information sequence, geomagnetic sensor information vector sequence, visual sensor information feature vector sequence, and lidar sensor information vector sequence are concatenated in a specific order to form a unified input sequence. This input sequence contains multifaceted information about the environment surrounding the transport vehicle, comprehensively describing the vehicle's initial perception state.

[0051] 2) Location coding

[0052] Because the attention mechanism in the Transformer algorithm is insensitive to positional information in the input sequence, positional encoding needs to be added to the input sequence. Positional encoding can be a combination of sine and cosine functions, adding a position-related encoded value to each element in the input sequence. In this way, the Transformer algorithm can distinguish sensor information elements at different positions when processing the input sequence, thus better capturing the relationships between them.

[0053] III. Processing Procedure of the Transformer Algorithm

[0054] (a) Multi-head attention mechanism

[0055] 1) Calculate the query, key, and value.

[0056] In the encoder part of the Transformer, each element in the input sequence is first transformed linearly to compute query, key, and value vectors. For each element in the input sequence, the query vector is used to match the key vectors of other elements to determine the degree of attention paid to the value vectors of other elements. For example, for a distance value element in ultrasonic sensing information, its query vector will interact with the key vectors of geomagnetic, visual, and lidar sensing information elements.

[0057] 2) Attention Calculation

[0058] By calculating the dot product of the query vector and the key vector, and then scaling and normalizing using the Softmax function, the attention weight of each element relative to other elements is obtained. Then, the attention weights are multiplied by the corresponding value vectors and summed to obtain a new attention-weighted vector. Due to the use of a multi-head attention mechanism, multiple such attention calculation processes are performed in parallel (e.g., attention calculation with 8 heads), with each head focusing on a different representational subspace of the input sequence, thereby capturing richer relational information.

[0059] (ii) Feedforward Neural Network

[0060] 1) Fully connected layer processing

[0061] The vector processed by the multi-head attention mechanism is then passed through a feedforward neural network (FFN). The FFN consists of two fully connected layers, interspersed with a non-linear activation function (such as ReLU). The number of neurons in each fully connected layer can be adjusted according to specific needs; for example, the first fully connected layer may have 512 neurons, and the second may have 256 neurons. The FFN further performs non-linear transformations on the attention-weighted vector, enhancing the model's expressive power.

[0062] (III) Layer Normalization and Residual Connectivity

[0063] 1) Layer normalization

[0064] After each multi-head attention layer and feedforward neural network layer, layer normalization is performed. Layer normalization normalizes the input of each neuron, making its mean 0 and variance 1. This accelerates the model training process and improves the model's stability.

[0065] 2) Residual connection

[0066] Simultaneously, residual connections are used to sum the input and output of each sub-layer (multi-head attention layer or feedforward neural network layer). Residual connections allow gradients to backpropagate more effectively, avoiding the vanishing or exploding gradient problems that occur in deep networks, enabling the model to better learn the mapping relationship between the input sequence and the output.

[0067] IV. Determination of the Initial Vehicle Environmental State Vector

[0068] 1) Output layer design

[0069] In the decoder part of the Transformer algorithm, an output layer is designed to generate the initial vehicle environment state vector. The number of neurons in the output layer is determined by the number of vehicle environment state parameters that need to be represented. For example, if 10 parameters are needed to represent the vehicle's position coordinates (x, y), orientation angle, type of surrounding objects, and distance, then the output layer can be set to 10 neurons.

[0070] 2) Vector generation

[0071] After processing by the Transformer algorithm, the neurons in the output layer generate corresponding numerical values. These values ​​are combined to form the initial vehicle environment state vector. For example, the output position coordinates might be relative to a fixed reference point in the parking lot, the orientation angle might be relative to a certain direction (such as due north), the types of surrounding objects can be represented by codes (such as 0 for snowdrifts, 1 for other vehicles, etc.), and the distance value is the distance to surrounding objects calculated based on various sensor information. This initial vehicle environment state vector comprehensively describes the initial state of the transport vehicle in the ski resort environment, providing important basic data for subsequent operations such as parking path planning.

[0072] In the process of processing the first IoT sensor information using the Transformer algorithm in the aforementioned IoT sensing control system, the final generated initial vehicle environment state vector is a multi-dimensional vector that comprehensively reflects the initial state of the transport vehicle in the ski resort environment. The following will illustrate this initial vehicle environment state vector in detail through specific numerical examples.

[0073] First, considering the vehicle's position information in three-dimensional space, the initial vehicle environment state vector may contain values ​​representing the vehicle's coordinates. For example, if we take a fixed corner of the parking lot as the origin (0, 0, 0), the transport vehicle's position coordinates in this coordinate system might be (x = 10.5 meters, y = 15.2 meters, z = 0 meters). In this embodiment of the invention, the x and y coordinates represent the vehicle's position on the horizontal plane relative to the origin, while the z coordinate is 0 meters because the vehicle is on flat ground. The accuracy of this coordinate value can reach the centimeter level, for example, ±0.01 meters, because the result of fusing multiple sensor information can provide relatively accurate positioning.

[0074] Besides location coordinates, the vehicle's orientation angle is also an important parameter. In this example, with due north as 0° and clockwise as the positive direction, the transport vehicle's orientation angle is 45°. This angle value is also derived by combining geomagnetic sensor information and visual sensor information such as lane lines or the relative positions of other vehicles. The accuracy of the angle can be set to ±1°, which is sufficient to reflect the approximate orientation of the vehicle for subsequent parking path planning.

[0075] Next, information about objects surrounding the vehicle is also a crucial component of the initial vehicle environment state vector. Taking the objects closest to the vehicle as an example, sensors might detect a snowdrift 3.2 meters in front of the vehicle, another transport vehicle 4.5 meters to its left rear, and a parking lot boundary facility (such as a guardrail) 2.8 meters to its right. The types of these objects can be represented by codes, such as setting the snowdrift as 0, the transport vehicle as 1, and the parking lot boundary facility as 2. Thus, in the initial vehicle environment state vector, the snowdrift in front might be represented as (type=0, distance=3.2 meters, direction=0°), the transport vehicle to its left rear as (type=1, distance=4.5 meters, direction=225°), and the parking lot boundary facility to its right as (type=2, distance=2.8 meters, direction=90°). The distance accuracy can also reach the centimeter level, and the direction accuracy is ±1°.

[0076] The vehicle's own speed information is also included in this vector. For example, if the transport vehicle is moving slowly at a speed of 0.5 m / s, this speed value is calculated using the vehicle's own speed sensor or based on continuous position information, with an accuracy of ±0.05 m / s. In some cases, the direction of the speed may also be included. If the vehicle is traveling in a straight line in a certain direction, the speed direction is at the same angle as the vehicle's orientation (45°). If the vehicle is turning, the speed direction will change depending on the turning situation.

[0077] Vehicle dimensions are also crucial for parking planning. For example, this transport vehicle is 6 meters long, 2.5 meters wide, and 3 meters high. These dimensions can be pre-set and stored in the system, or verified through visual sensing or other methods. The accuracy of the dimensions may vary depending on the vehicle's actual manufacturing standards, with length and width accuracy at ±0.05 meters and height accuracy at ±0.1 meters.

[0078] Next, consider the vehicle's wheel status, such as the wheel steering angle. For example, if the current wheel steering angle is 5°, this angle represents the wheel's offset angle relative to the vehicle's straight-line travel, with an accuracy of ±0.5°. The wheel steering angle is crucial for determining the vehicle's trajectory and for parking maneuvers.

[0079] In addition, special conditions of the vehicle's surrounding environment are also reflected in the vector. For example, visual sensing information might detect ground conditions around the vehicle, revealing snow accumulation within 2-5 meters in front of the vehicle with a depth of 0.1 meters. Snow depth can be measured using lidar or other specialized sensors with an accuracy of ±0.01 meters. This ground condition information affects vehicle driving and parking operations, and needs to be considered when calculating braking distance and steering friction.

[0080] Additionally, the initial vehicle environment state vector may also contain information related to the parking lot layout. For example, the type of parking area the vehicle is in; if it's a perpendicular parking area, it's encoded as 0, and if it's a parallel parking area, it's encoded as 1. Furthermore, the distance and direction of the vehicle to the nearest parking space are also included; for example, the nearest perpendicular parking space is 6 meters to the left front of the vehicle, at a 30° angle.

[0081] Regarding the relative speed between the vehicle and surrounding vehicles or objects, this information is also reflected if there are moving objects nearby. For example, if a transport vehicle to the left rear of the vehicle is moving away from it at a speed of 0.3 m / s, this relative speed information is calculated through continuous position monitoring with an accuracy of ±0.05 m / s.

[0082] Vehicle acceleration information is also part of the vector. For example, if a vehicle's current acceleration is 0.1 m / s², this value represents the rate of change of the vehicle's velocity, with an accuracy of ±0.01 m / s². The direction of acceleration can also be included; for example, if the vehicle is accelerating forward, the direction of acceleration is at the same angle as the vehicle's heading, which is 45°.

[0083] Furthermore, the vehicle's suspension system status may also be reflected in the vector. For example, sensors may detect the degree of compression in the vehicle's suspension system, such as 0.03 meters of compression in the left front suspension and 0.02 meters in compression in the right rear suspension. This value reflects the vehicle's load distribution and the impact of road surface smoothness on the vehicle, with an accuracy of ±0.005 meters.

[0084] The vehicle's lighting status can also be part of the initial vehicle environment state vector. For example, the headlights are on, represented by 1 for on and 0 for off; the brake lights are off. Lighting status information is significant in interactions with other vehicles and in safe driving in low-light ski resort environments.

[0085] Regarding the vehicle's battery or fuel level, for electric transport vehicles, the remaining battery level is 50%, with an accuracy of ±5%; for fuel-powered transport vehicles, the remaining fuel level is 30 liters, with an accuracy of ±1 liter. While this information may seem unrelated to parking path planning, it can be useful in certain situations, such as when considering whether there is sufficient energy to move to a designated parking location or whether charging equipment needs to be activated during parking.

[0086] The vehicle's communication status may also be included. For example, a communication signal strength of -60dBm between the vehicle and the IoT sensor control system indicates a normal communication connection but a moderate signal strength. Communication status information ensures stable data exchange between the vehicle and the control system, enabling timely receipt of parking commands or uploading of vehicle status information.

[0087] Information about the vehicle's interior environment, such as the interior temperature of 20°C with an accuracy of ±1°C, and whether there are passengers inside the vehicle (e.g., 1 for one passenger and 0 for no passenger). While this information may not have a significant impact on the parking operation itself, it may be considered in some comprehensive management scenarios (such as selecting a more suitable parking location based on the number of people inside the vehicle to facilitate passenger boarding and alighting).

[0088] The status of the vehicle's sensors also needs to be considered in the initial vehicle environment state vector. For example, the operating status of an ultrasonic sensor is represented by 1 for normal operation and 0 for malfunction. If an ultrasonic sensor malfunctions, it will affect the vehicle's accurate perception of its surroundings, especially in the detection of nearby obstacles. Therefore, the system needs to know the status of each sensor in order to make appropriate adjustments or prompt maintenance when processing sensor information.

[0089] External identification information of the vehicle may also be included. For example, unique identification information such as the vehicle's license plate number or identification code is crucial for accurately identifying vehicles and recording parking information in parking management systems.

[0090] The initial vehicle environment state vector comprehensively and meticulously describes the initial state of the transport vehicle in the ski resort environment by containing so many dimensions of information with specific numerical values. Each element in these numerical feature vectors is not isolated but interconnected and mutually influential. For example, the vehicle's position coordinates and the distance and direction information of surrounding objects together determine the feasible path for the vehicle during parking; information such as the vehicle's speed, acceleration, and wheel steering angle reflects the vehicle's dynamic driving state, which needs to be combined with surrounding environmental information to ensure safe and efficient parking operations during parking path planning; information such as the vehicle's internal environment, battery or fuel level, and communication status, while seemingly auxiliary in some cases, plays an indispensable role in overall vehicle management and optimizing the parking experience. Furthermore, the values ​​in this vector are not static. As the vehicle moves, the surrounding environment changes, and sensors continuously monitor, the IoT sensor control system constantly updates this vector, providing the latest and most accurate basic data for subsequent operations such as parking path planning and the formulation of automatic parking control strategies.

[0091] In actual system operation, due to the complexity of sensor information and the dynamic changes in the environment, each value of the initial vehicle environmental state vector may be affected by multiple factors. For example, in snowy conditions, visual sensing information may be affected by snowflakes, leading to errors in judging the distance and type of surrounding objects. This necessitates supplementation and correction using other sensor information (such as lidar sensing information). Similarly, geomagnetic sensing information may be affected by metal structures or other magnetic field sources around the ski resort, thus affecting the accuracy of the vehicle's position and orientation angle. Therefore, the IoT sensor control system needs to continuously optimize algorithms and fusion strategies to ensure that the initial vehicle environmental state vector can reflect the vehicle's actual state as accurately as possible.

[0092] Meanwhile, the emphasis of this initial vehicle environment state vector may differ depending on the ski resort environment and the type of transport vehicle. In large ski resorts with high traffic volume and complex parking layouts, the relative speed of the vehicle to surrounding vehicles and parking layout information may be more important. However, in smaller ski resorts or with specific transport vehicles (such as small vehicles specifically used for transporting ski equipment), the vehicle's dimensions and cargo loading (if any) may become more critical information. Therefore, IoT sensor control systems need to flexibly adjust the composition and weights of each value in the initial vehicle environment state vector according to the specific application scenario to achieve optimized automatic parking control.

[0093] In terms of data transmission and storage, this initial vehicle environment state vector, containing numerous numerical features, requires effective management. Due to the large data volume and the need for real-time updates, employing efficient data compression algorithms can reduce bandwidth requirements and storage space consumption. For example, lossless compression algorithms can be used to compress the values ​​in the vector, improving data transmission and storage efficiency without sacrificing data accuracy. Simultaneously, to ensure data security, encryption technologies such as AES (Advanced Encryption Standard) can be used during transmission to encrypt the vector data, preventing theft or tampering during transmission. Regarding storage, the data should be stored in a reliable database system with strict access permissions, allowing only authorized devices and personnel to access and modify this data.

[0094] In summary, the initial vehicle environment state vector comprehensively reflects the initial state of the transport vehicle in the ski resort environment through a series of specific numerical feature vectors. These values ​​cover information on the vehicle's position, attitude, surrounding environment, and its own state, and play a fundamental and crucial role in the automatic parking process of the entire IoT sensing control system.

[0095] Based on steps 110-140 above, the debugging method for the intelligent driving control processing algorithm includes: acquiring the second IoT sensor information of multiple historical transport vehicles and the prior information of each historical transport vehicle, wherein the prior information of the historical transport vehicle characterizes whether there are related path intersections between the historical transport vehicle and the automatic parking control strategy during the automatic parking process; using the Transformer algorithm to perform state feature mining on the second IoT sensor information of the multiple historical transport vehicles to obtain the initial vehicle environment state vector of each historical transport vehicle; using the Transformer algorithm based on the initial vehicle environment state vector of each historical transport vehicle to determine the initial vehicle environment state vector of each historical transport vehicle. The parking path planning label of the historical transport vehicle represents the probability that the historical transport vehicle matches the path intersection point; based on the initial vehicle environment state vector of each historical transport vehicle, the prior confidence of each historical transport vehicle is determined, and the prior confidence of the historical transport vehicle is used to adjust the initial vehicle environment state vector of the historical transport vehicle to the target mode; based on the prior basis, parking path planning label and prior confidence of each historical transport vehicle, the Transformer algorithm is debugged to obtain the intelligent driving control processing algorithm, which is used to determine the probability that the transport vehicle to be controlled matches the path intersection point.

[0096] In the entire technical solution, the debugging of the intelligent driving control processing algorithm is crucial for improving the accuracy and effectiveness of automatic parking control.

[0097] First, the second IoT sensing information of multiple historical transport vehicles and the prior information of each historical transport vehicle are acquired. The second IoT sensing information in this embodiment of the invention covers various types of sensing data, similar in nature and function to the first IoT sensing information mentioned earlier. For example, ultrasonic sensing information can provide distance information of objects within a certain range around the vehicle. In historical data, at a certain moment, the ultrasonic sensor at the front of a historical transport vehicle might detect a distance of 1.2 meters from an obstacle, while the side sensor might detect a distance of 2.5 meters from another object. Geomagnetic sensing information reflects the vehicle's state in the geomagnetic field. If expressed as a numerical value of geomagnetic intensity, the measured geomagnetic intensity at a certain location might be 0.4 Gauss, thus helping to determine the approximate location of the vehicle in the parking lot. Visual sensing information contains rich image data content. From the images, the layout elements of the parking lot can be identified, such as parking lines, lanes, and the relative positions of other vehicles. For example, visual sensing information can indicate that there are three other vehicles around a historical transport vehicle, and the lateral distance to the nearest vehicle is 1.8 meters. LiDAR sensing information constructs a three-dimensional point cloud map of the environment around the vehicle. For example, in the LiDAR scan data of a certain historical transport vehicle, the outline of an object with a height of 0.5 meters is accurately depicted at a distance of 5 meters from the vehicle in a certain direction.

[0098] Meanwhile, the prior information of each historical transport vehicle is crucial, as it clearly indicates whether there were any path intersections between the historical transport vehicle and the automatic parking control strategy during the automatic parking process. This information is based on actual parking operation records; the prior information of some historical transport vehicles indicates that path intersections with the control strategy did indeed occur during parking, while that of others indicates that such intersections did not occur.

[0099] Next, the Transformer algorithm is used to mine state features from the second IoT sensor information of multiple historical transport vehicles, thereby obtaining the initial vehicle environment state vector for each historical transport vehicle. Taking a specific historical transport vehicle as an example, its initial vehicle environment state vector contains a wealth of meaningful numerical information. The vehicle's position coordinates might be (9.5 m, 13.2 m, 0 m), where the z-coordinate of 0 m indicates that the vehicle is on a level surface. The vehicle's orientation angle is, for example, 40°, relative to a reference direction (such as due north). Regarding the information about objects around the vehicle, there might be a snowdrift (type code 0) 3.0 m to the right front of the vehicle, and another transport vehicle (type code 1) 4.0 m to the left rear of the vehicle. The vehicle's own speed is 0.4 m / s, reflecting the vehicle's current driving state. The wheel steering angle is, for example, 8°, which is important for analyzing the vehicle's trajectory and parking operations. The vehicle is 5.8 meters long, 2.3 meters wide, and 2.8 meters high. These dimensions are essential for parking path planning. For example, when determining whether a vehicle can smoothly enter a parking space, the relationship between the vehicle's width and the parking space's width needs to be considered.

[0100] Then, based on this initial vehicle environment state vector, the Transformer algorithm determines the parking path planning label for each historical transport vehicle. This label represents the probability that the historical transport vehicle will match a path intersection point. For example, the parking path planning label of a certain historical transport vehicle might indicate that it has a 0.35 probability of having a path intersection point. This probability is calculated by comprehensively considering multiple factors in the vehicle's initial environment state vector. For example, the vehicle's location is close to the intersection area of ​​the parking lot's passageways, the surrounding vehicles are parked relatively densely, and factors such as the vehicle's own speed and turning angle all contribute to the algorithm's determination of a certain risk of path intersection.

[0101] Next, based on the initial vehicle environment state vectors of each historical transport vehicle, the prior confidence level for each historical transport vehicle is determined. The prior confidence level is used to adjust the initial vehicle environment state vectors of historical transport vehicles to the target mode. For example, if the prior confidence level of a certain historical transport vehicle is 0.7, this value reflects the reliability of the vehicle's initial environment state vector or its importance weight in the overall data. When the prior confidence level is high, more attention will be given to the various values ​​in the initial environment state vector of that vehicle when debugging the Transformer algorithm. For example, the influence weight of key values ​​such as vehicle position coordinates and distances to surrounding objects will be adjusted according to the prior confidence level during algorithm debugging. If the prior confidence level is 0.7, then when calculating the probability related to path intersections, key information such as vehicle position coordinates may be given a higher weight ratio, possibly increasing from the original 30% weight to 40%, thereby more accurately adjusting the initial vehicle environment state vector to the target mode to better adapt to the algorithm debugging needs.

[0102] Finally, based on the prior information, parking path planning labels, and prior confidence scores of each historical transport vehicle, the Transformer algorithm is debugged to obtain the intelligent driving control processing algorithm. During this process, if the prior information of a certain historical transport vehicle indicates the existence of a path intersection point, and the parking path planning label shows a high path intersection probability (e.g., 0.6), while the prior confidence score is also high (e.g., 0.8), then the relevant parameters in the Transformer algorithm will be adjusted during debugging. For example, the weighting of factors such as vehicle speed and the distribution of surrounding objects when calculating the path intersection probability might be adjusted. For instance, if the original weight of vehicle speed in calculating the path intersection probability was 0.1, due to the aforementioned situation, its weight might be adjusted to 0.15, so that the algorithm can more accurately determine the probability of matching the path intersection point when processing the transport vehicle under control. This debugging process is an iterative optimization process. By comprehensively analyzing and processing data from multiple historical transport vehicles, the parameters of the Transformer algorithm are continuously adjusted, ultimately resulting in an intelligent driving control processing algorithm suitable for determining the probability of matching the path intersection point of the transport vehicle under control.

[0103] Utilizing rich information from historical transport vehicles to debug the Transformer algorithm, resulting in an intelligent driving control processing algorithm, offers several significant advantages. In terms of accuracy, taking the values ​​in the vehicle's environmental state vector as an example, precise information such as vehicle position and distances to surrounding objects is effectively used during algorithm debugging, enabling the intelligent driving control processing algorithm to more accurately determine the probability of the controlled transport vehicle matching path intersections. Regarding parking safety, accurate determination of parking path planning label probabilities helps identify potential path intersection risks in advance, thereby avoiding dangerous situations such as collisions during automatic parking. The introduction of prior confidence provides a reasonable basis for adjusting the initial vehicle environmental state vector, allowing the algorithm to be optimized in a targeted manner based on the reliability of different historical transport vehicle data. Overall, this debugging method optimizes the algorithm's control over the automatic parking of transport vehicles, improves the reliability and efficiency of the entire parking system, reduces parking failures or safety accidents caused by algorithm inaccuracies, and better adapts to the automatic parking needs of complex environments such as ski resorts.

[0104] In a preferred embodiment, the plurality of historical transport vehicles includes a plurality of first transport vehicles to which an automatic parking control strategy has been issued and a plurality of second transport vehicles to which the automatic parking control strategy has not been issued; the step of determining the parking path planning label of each historical transport vehicle based on the initial vehicle environment state vector of each historical transport vehicle using the Transformer algorithm includes: determining the parking path planning label of each first transport vehicle based on a first debugging annotation and the initial vehicle environment state vector of the plurality of first transport vehicles using the Transformer algorithm, wherein the first debugging annotation represents the result of the automatic parking control strategy being issued; and determining the parking path planning label of each second transport vehicle based on a second debugging annotation and the initial vehicle environment state vector of the plurality of second transport vehicles using the Transformer algorithm, wherein the second debugging annotation represents the result of the automatic parking control strategy not being issued.

[0105] In this technical solution, multiple historical transport vehicles are divided into multiple first transport vehicles that have been issued automatic parking control strategies and multiple second transport vehicles that have not been issued automatic parking control strategies. This classification helps to analyze parking-related situations in more detail under different circumstances.

[0106] For the first transport vehicle, the Transformer algorithm is used to determine its parking path planning label based on the first debugging annotation and the initial vehicle environment state vectors of multiple first transport vehicles when determining its parking path planning label. The first debugging annotation represents the result of the issued automatic parking control strategy. For example, in the initial vehicle environment state vector of a certain first transport vehicle, the vehicle's coordinates are (10.2 m, 14.5 m, 0 m), the orientation angle is 45°, there is an object 2.8 m away from the vehicle with a parking facility code of 2, and the vehicle speed is 0.3 m / s. Based on such an initial vehicle environment state vector, and combined with the first debugging annotation, which contains information such as the interaction information between the transport vehicle and other vehicles within a specific time after the automatic parking control strategy is issued, such as the minimum safe distance from another vehicle during parking being 0.5 m, the Transformer algorithm determines the parking path planning label of the first transport vehicle by comprehensively analyzing these factors. For example, this label might indicate that the first transport vehicle has a 0.4 probability of having a path intersection. This probability is derived by taking into account factors such as the vehicle's initial position, speed, surrounding objects, and the interaction results after the automatic parking control strategy is issued.

[0107] For the second transport vehicle, the Transformer algorithm is used to determine its parking path planning label based on the second debugging annotation and the initial vehicle environment state vectors of multiple second transport vehicles. The second debugging annotation represents the result of not having an automatic parking control strategy issued. Taking a certain second transport vehicle as an example, its initial vehicle environment state vector shows vehicle coordinates (8.8 m, 11.3 m, 0 m), an orientation angle of 35°, an object 3.2 meters away from the vehicle with a snowdrift type (coded 0), and a vehicle speed of 0.2 m / s. The second debugging annotation may contain information about why an automatic parking control strategy was not issued, such as an overly complex surrounding environment or the vehicle's own state being unsuitable for automatic parking. The Transformer algorithm determines the parking path planning label of the second transport vehicle based on this information and the initial vehicle environment state vector. For example, this label might indicate that the second transport vehicle has a 0.3 probability of having a path intersection point. In this embodiment, the probability is derived by comprehensively considering the vehicle's initial state and the relevant factors of not having an automatic parking control strategy issued.

[0108] By processing the first and second transport vehicles separately, a more comprehensive and detailed analysis of their parking path planning under different strategies can be achieved. For the first transport vehicle, since an automatic parking control strategy has been issued, the relevant information in the first debugging annotation reflects various situations related to the automatic parking control strategy during actual parking operations. These situations, combined with the initial vehicle environment state vector, can more accurately determine the parking path planning label. For the second transport vehicle, although an automatic parking control strategy has not been issued, the information in the second debugging annotation is equally important. It provides relevant factors regarding the vehicle's failure to perform automatic parking operations. Together with the initial vehicle environment state vector, it determines the parking path planning label, thus helping to understand the parking probability of the vehicle in the parking lot environment from another perspective.

[0109] By determining parking path planning labels for the first and second transport vehicles based on different debugging annotations and initial vehicle environment state vectors, the accuracy of parking path planning can be improved. Numerical examples show that combining the values ​​in the state vectors (vehicle position, surrounding object types, and distances) with information from the debugging annotations more accurately reflects the vehicle's parking path under different conditions. This helps optimize automatic parking control strategies, better cope with complex parking environments, improve the overall performance and reliability of the parking system, and reduce risks and uncertainties during the parking process.

[0110] In an optional embodiment, determining the prior confidence of each historical transport vehicle based on its initial vehicle environment state vector includes: performing feature mapping on the initial vehicle environment state vector of each historical transport vehicle to obtain the driving control decision mapping vector of each historical transport vehicle; and determining the prior confidence of each historical transport vehicle based on its driving control decision mapping vector.

[0111] Further, the driving control decision mapping vector of the historical transport vehicle includes the local control decision mapping vector of the historical transport vehicle in multiple attention indicators; the step of determining the prior confidence of each historical transport vehicle based on the driving control decision mapping vector of each historical transport vehicle includes: for any historical transport vehicle, determining the first mapping vector comparison result based on the local control decision mapping vector of the any historical transport vehicle in the first attention indicator and the local control decision mapping vector of each historical transport vehicle in the first attention indicator, wherein the first attention indicator is any one of the multiple attention indicators; based on the local control decision mapping vector of the any historical transport vehicle in the first attention indicator, determining the first mapping vector comparison result based on the local control decision mapping vector of the any historical transport vehicle in the first attention indicator and the local control decision mapping vector of each historical transport vehicle in the first attention indicator, determining the prior confidence of each historical transport vehicle in the first attention indicator; The local control decision mapping vectors of the two attention indicators and the local control decision mapping vectors of each historical transport vehicle for the second attention indicator are used to determine the comparison result of the second mapping vector. The second attention indicator is any attention indicator other than the first attention indicator among the plurality of attention indicators. Based on the comparison result of the first mapping vector and the comparison result of the second mapping vector, the local control decision mapping correlation degree between the first attention indicator and the second attention indicator for any historical transport vehicle is determined. Based on the local control decision mapping correlation degree between each historical transport vehicle for every two attention indicators, the prior confidence degree of each historical transport vehicle is determined.

[0112] Furthermore, determining the first mapping vector comparison result based on the local control decision mapping vector of any historical transport vehicle in the first attention index and the local control decision mapping vectors of all historical transport vehicles in the first attention index includes: determining the mean vector of the local control decision mapping of the first attention index based on the basic confidence level of each historical transport vehicle and the local control decision mapping vector of each historical transport vehicle in the first attention index; determining the reinforcement vector of the local control decision mapping of any historical transport vehicle in the first attention index based on the basic confidence level of any historical transport vehicle and the local control decision mapping vector of any historical transport vehicle in the first attention index; and using the vector operation value between the reinforcement vector of the local control decision mapping of any historical transport vehicle in the first attention index and the mean vector of the local control decision mapping of the first attention index as the first mapping vector comparison result.

[0113] In this embodiment, the entire technical solution focuses on the key step of determining the prior confidence level based on the initial vehicle environment state vector of each historical transport vehicle. Its specific operation process involves multiple levels of complex calculations and logical judgments.

[0114] First, feature mapping is performed on the initial vehicle environment state vectors of each historical transport vehicle to obtain the driving control decision mapping vector for each historical transport vehicle. The initial vehicle environment state vectors of historical transport vehicles contain rich vehicle-related information, which is an important basis for determining vehicle driving control decisions. For example, in the initial vehicle environment state vector of a certain historical transport vehicle, the vehicle's coordinate position is (11.5 meters, 16.2 meters, 0 meters). In this embodiment of the invention, the coordinates represent the vehicle's position in a specific coordinate system of the parking lot, where the z-coordinate of 0 meters indicates that the vehicle is on the same horizontal plane. The vehicle's orientation angle is 50°, which reflects the vehicle's direction relative to a fixed direction (such as due north). The information of objects around the vehicle is also an important part of the vector, such as an object 3.8 meters to the right rear of the vehicle, whose type is another vehicle (e.g., coded as 1), the vehicle's current driving speed is 0.4 meters / second, and the wheel steering angle is 10°, etc. Through feature mapping operations, the information in these initial vehicle environment state vectors is transformed into driving control decision mapping vectors. These mapping vectors contain mapping information related to how the vehicle makes driving control decisions. They are no longer just the original vehicle state information, but information that has been transformed through a specific mapping relationship and is more directly related to driving control decisions.

[0115] The driving control decision mapping vector of historical transport vehicles includes local control decision mapping vectors for multiple attention indicators. This means that within this vector, for different attention indicators, there exist corresponding local control decision mapping parts. For any historical transport vehicle, determining the prior confidence level is a multi-step and interrelated process.

[0116] When determining the first mapping vector comparison result, it is necessary to determine the mean vector of the local control decision mapping of the first attention index based on the basic confidence level of each historical transport vehicle and the local control decision mapping vector of the first attention index. For example, there is a group of historical transport vehicles, where the basic confidence levels of three historical transport vehicles are 0.6, 0.7, and 0.8, respectively. For the local control decision mapping vectors of these three historical transport vehicles under the first attention index, for example, they are vector A=[a1, a2, a3], vector B=[b1, b2, b3], and vector C=[c1, c2, c3], respectively (in this embodiment of the invention, the vector elements represent values ​​related to the local control decision mapping, and the specific values ​​depend on the vehicle state and the mapping logic of the algorithm). For example, according to the idea of ​​weighted averaging (but the specific weighting coefficients are determined by the algorithm), the mean vector of the local control decision mapping of the first attention index is calculated. At the same time, based on the basic confidence level of any historical transport vehicle and its local control decision mapping vector of the first attention index, the reinforcement vector of the local control decision mapping of the first attention index for that transport vehicle is determined. For example, for a historical transport vehicle with a base confidence level of 0.7, its local control decision mapping vector under the first attention metric is vector D = [d1, d2, d3]. Through specific algorithmic operations (determined by the algorithm's internal logic), the base confidence level is used to perform calculations with the elements of vector D to obtain the local control decision mapping reinforcement vector. Then, the vector operation value between the reinforcement vector of the historical transport vehicle's local control decision mapping under the first attention metric and the mean vector of the local control decision mapping under the first attention metric is used as the comparison result of the first mapping vector. The calculation method of this vector operation value is based on the algorithm's predefined operation rules, and it reflects the degree of difference between the historical transport vehicle and the overall mean under the first attention metric.

[0117] Next, based on the local control decision mapping vector of any historical transport vehicle under the second attention index and the local control decision mapping vectors of all historical transport vehicles under the second attention index, the comparison result of the second mapping vector is determined. In this embodiment of the invention, the second attention index is any attention index other than the first attention index among multiple attention indices. For example, under the second attention index, there is another set of relevant data for historical transport vehicles. For a specific historical transport vehicle, its local control decision mapping vector under the second attention index is vector E=[e1, e2, e3], and the local control decision mapping vectors of other historical transport vehicles under the second attention index are vector F=[f1, f2, f3], vector G=[g1, g2, g3], etc. Following a method similar to that used to determine the first mapping vector comparison result, the mean vector of the local control decision mapping for the second attention index is first determined based on the baseline confidence level of each historical transport vehicle and its local control decision mapping vector for the second attention index. Then, the reinforcement vector of the local control decision mapping for the specific historical transport vehicle in the second attention index is determined based on its baseline confidence level and its local control decision mapping vector for the second attention index. Finally, the vector operation value between these two vectors is calculated to obtain the second mapping vector comparison result. This result also reflects the difference between the historical transport vehicle and the overall mean under the second attention index.

[0118] Then, based on the comparison results of the first and second mapping vectors, the correlation degree of local control decision mapping between the first and second attention indicators for any historical transport vehicle is determined. This correlation degree is determined based on the two mapping vector comparison results obtained earlier, calculated through a specific algorithmic logic. For example, if the first mapping vector comparison result is vector H = [h1, h2, h3] and the second mapping vector comparison result is vector I = [i1, i2, i3], the algorithm obtains the correlation degree of local control decision mapping between the first and second attention indicators for the historical transport vehicle by performing specific combination operations on the elements of these two vectors (defined internally by the algorithm). This correlation degree reflects the closeness of the relationship between the local control decision mapping of the historical transport vehicle under these two attention indicators; it is an indicator that comprehensively considers the comparison results of the vehicle and the overall situation under two different attention indicators.

[0119] Finally, based on the correlation degree of the local control decision mapping between each pair of attention indicators for each historical transport vehicle, the prior confidence of each historical transport vehicle is determined. For example, the correlation degrees of the local control decision mapping between multiple different pairs of attention indicators for a certain historical transport vehicle are correlation degree 1, correlation degree 2, correlation degree 3, etc. (each correlation degree is a value obtained through the previous calculation method). The algorithm obtains the prior confidence of the historical transport vehicle by comprehensively calculating these correlation degrees (which may involve weighted summation, normalization, etc., depending on the internal logic of the algorithm). For example, the calculated prior confidence of a certain historical transport vehicle is 0.75. This prior confidence is of great significance in the entire technical solution. It is used to adjust the initial vehicle environment state vector of the historical transport vehicle to the target mode, thus playing a key role in subsequent algorithm debugging and other processes. For example, when adjusting the initial vehicle environment state vector, the prior confidence may affect the weight adjustment of different elements in the vector, or perform operations such as modifying certain elements, so that the initial vehicle environment state vector better meets the actual control requirements.

[0120] The prior confidence level is determined through such a complex and systematic calculation process. From the perspective of feature vector values, the calculation is based on the initial vehicle environmental state vector values, such as the vehicle's position and speed. This method greatly improves the accuracy of the prior confidence level determination. The improved accuracy of the prior confidence level allows for more precise adjustment of the initial vehicle environmental state vector towards the target mode. This helps improve the accuracy and reliability of the entire algorithm in automatic parking control, enabling the automatic parking system to more meticulously consider different vehicle states and environmental conditions. In real-world automatic parking scenarios, a more accurate prior confidence level can optimize parking-related decision-making, reduce parking risks caused by inaccurate decisions, improve parking efficiency, and better adapt to complex and changing parking environment conditions, whether in densely packed vehicles or with special obstacles, enabling more effective automatic parking operations.

[0121] In one exemplary technical solution, the step of debugging the Transformer algorithm based on the prior information, parking path planning labels, and prior confidence scores of each historical transport vehicle to obtain the intelligent driving control processing algorithm includes: for any historical transport vehicle, determining the first path planning node error corresponding to the historical transport vehicle based on the prior information and parking path planning labels of the historical transport vehicle; determining the first path planning node weighted error of the historical transport vehicle based on the first path planning node error and the prior confidence score of the historical transport vehicle; determining the first training error based on the first path planning node weighted error of each historical transport vehicle; and debugging the Transformer algorithm based on the first training error to obtain the intelligent driving control processing algorithm.

[0122] In this exemplary technical solution, the entire process revolves around debugging the Transformer algorithm based on the prior information of each historical transport vehicle, parking path planning labels, and prior confidence to obtain the intelligent driving control processing algorithm.

[0123] First, for any historical transport vehicle, the first path planning node error is determined based on its prior information and parking path planning label. The prior information of a historical transport vehicle indicates whether there are any path intersections related to the automatic parking control strategy during the automatic parking process, while the parking path planning label represents the probability that the vehicle matches a path intersection. For example, if the prior information of a certain historical transport vehicle indicates the existence of a path intersection, its parking path planning label shows a probability of 0.6 for such an intersection. If a path intersection does indeed exist, the first path planning node error may be a value related to the difference between the two. This value is determined based on the internal logic of the algorithm and is related to the comparison between the specific representations of the prior information and the parking path planning label and the actual results.

[0124] Next, based on the error of the first path planning node corresponding to any given historical transport vehicle and its prior confidence level, the weighted error of the first path planning node for that historical transport vehicle is determined. For example, if the prior confidence level of a certain historical transport vehicle is 0.7, and its first path planning node error is a specific value (derived from the previous calculation), the weighted error of the first path planning node is obtained by performing calculations on the first path planning node error and the prior confidence level according to the algorithm's rules (this calculation method is based on the algorithm's logic). The prior confidence level here plays a role in weighting the first path planning node error, so that the errors of historical transport vehicles with different confidence levels have different weights in subsequent calculations.

[0125] Then, the first training error is determined based on the weighted error of the first path planning nodes for each historical transport vehicle. For example, if there are multiple historical transport vehicles, each with its own weighted error for the first path planning nodes, these weighted errors are combined in a specific way to form the first training error. For instance, it might be obtained by summing the weighted errors of the first path planning nodes of all historical transport vehicles or by using a weighted average method (determined by the algorithm's internal logic). This first training error reflects the overall error of all historical transport vehicles in terms of path planning nodes, comprehensively considering the influence of prior information, parking path planning labels, and prior confidence levels for each historical transport vehicle.

[0126] Finally, the Transformer algorithm was debugged based on the first training error to obtain the intelligent driving control processing algorithm. The first training error serves as feedback information, used to adjust the parameters in the Transformer algorithm. Since the Transformer algorithm played a crucial role in determining the initial vehicle environment state vector and parking path planning labels of historical transport vehicles, debugging it using the first training error can optimize the algorithm's performance, enabling it to more accurately determine the probability of matching path intersections when processing the transport vehicles under control. For example, if the first training error is large, it indicates a significant deviation in the previous processing of historical transport vehicle data by the Transformer algorithm. In such cases, adjustments to the algorithm's parameters are necessary, such as adjusting the weights of factors like vehicle position and surrounding objects when calculating parking path planning labels, or adjusting coefficients in certain calculation steps when determining prior confidence levels, thereby obtaining an optimized intelligent driving control processing algorithm.

[0127] This method of determining training errors and debugging the Transformer algorithm based on prior information from historical transport vehicles, parking path planning labels, and prior confidence levels, from a specific numerical perspective, involves factors such as the probability values ​​of parking path planning labels for historical transport vehicles and prior confidence levels, all of which participate in the algorithm's optimization process. This helps improve the accuracy of the intelligent driving control processing algorithm, enabling it to more accurately judge the automatic parking status of the controlled transport vehicle, reduce errors in parking path planning, and thus improve the reliability and efficiency of the entire automatic parking system, better adapting to different parking scenarios and vehicle states.

[0128] In another exemplary technical solution, the plurality of historical transport vehicles includes a plurality of first transport vehicles to which an automatic parking control strategy has been issued and a plurality of second transport vehicles to which the automatic parking control strategy has not been issued; the step of debugging the Transformer algorithm based on the prior information, parking path planning labels, and prior confidence of each historical transport vehicle to obtain the intelligent driving control processing algorithm includes: determining a first training error based on the prior information, parking path planning labels, and prior confidence of each historical transport vehicle; determining a second training error based on the initial state vector comparison results between the initial vehicle environment state vectors of the plurality of first transport vehicles and the initial vehicle environment state vectors of the plurality of second transport vehicles; and debugging the Transformer algorithm based on the first training error and the second training error to obtain the intelligent driving control processing algorithm.

[0129] In this exemplary technical solution, the entire solution focuses on debugging the Transformer algorithm based on relevant information from multiple historical transport vehicles (including multiple first transport vehicles that have been issued automatic parking control strategies and multiple second transport vehicles that have not been issued automatic parking control strategies) to obtain the intelligent driving control processing algorithm.

[0130] First, the first training error is determined based on the prior information, parking path planning labels, and prior confidence scores of each historical transport vehicle. For each historical transport vehicle, the prior information clarifies whether there are any path intersections related to the automatic parking control strategy during the automatic parking process. The parking path planning label represents the probability that the vehicle matches a path intersection, and the prior confidence score measures the reliability of the vehicle's initial vehicle environment state vector or its importance weight in the overall data. For example, the prior information of one first transport vehicle indicates the existence of a path intersection, its parking path planning label shows a probability of 0.6 for the existence of a path intersection, and its prior confidence score is 0.7. The prior information of another first transport vehicle indicates that there are no path intersections, its parking path planning label shows a probability of 0.2 for the existence of a path intersection, and its prior confidence score is 0.6. A similar situation exists for the second transport vehicle.

[0131] When determining the first training error, an intermediate error is first determined based on the prior information and parking path planning label of each historical transport vehicle (similar to the first path planning node error mentioned earlier, but considering all historical transport vehicles). This intermediate error is a value determined by the algorithm's internal logic based on the relationship between the two. Then, this intermediate error is calculated with the prior confidence level according to the algorithm's specified method to obtain the weighted error corresponding to each historical transport vehicle. These weighted errors are combined in a specific way (such as summation or weighted averaging, determined by the algorithm's internal logic) to form the first training error.

[0132] Next, the second training error is determined based on the comparison results of the initial vehicle environment state vectors of multiple first transport vehicles and multiple second transport vehicles. The initial vehicle environment state vector of a first transport vehicle includes information such as the vehicle's position coordinates, orientation angle, surrounding object information, vehicle speed, and vehicle dimensions. For example, in the initial vehicle environment state vector of one first transport vehicle, the vehicle's position coordinates are (10.5 m, 13.2 m, 0 m), the orientation angle is 45°, there is an object (type 1) 2.8 m in front, the vehicle speed is 0.3 m / s, the vehicle length is 5.8 m, and the width is 2.3 m. The initial vehicle environment state vector of another first transport vehicle has different values ​​for these parameters. The initial vehicle environment state vectors of the second transport vehicles also have similar components and their own values.

[0133] Determining the initial state vector comparison results is a complex process. For example, for vehicle position coordinates, it might involve calculating the average or variance of the coordinate differences between multiple first and second transport vehicles on a certain coordinate axis (the specific calculation method is determined by the algorithm). For vehicle speed, it might involve calculating the distribution of the speed difference between the two vehicles, etc. By comprehensively calculating these comparison results from different aspects according to the algorithm's internal logic, a second training error is obtained. This second training error reflects the impact of the differences between the first and second transport vehicles in their initial vehicle environmental states on algorithm debugging.

[0134] Finally, the Transformer algorithm was debugged based on the first and second training errors to obtain the intelligent driving control processing algorithm. The first and second training errors, as feedback information from two different sources, jointly influence the debugging of the Transformer algorithm. If the first training error is large, it indicates a significant deviation in the algorithm regarding prior assumptions, parking path planning labels, and prior confidence levels, potentially requiring adjustments to the weights or internal logic relationships when calculating these parameters. If the second training error is large, for example, due to differences in the initial vehicle environment state vectors of the first and second transport vehicles, it may be necessary to adjust the relevant parameters when processing the initial vehicle environment states of different types of transport vehicles (first and second transport vehicles), such as the degree of emphasis placed on different parameters for different vehicle types during feature mapping. Through this debugging process, the Transformer algorithm can be optimized into an intelligent driving control processing algorithm, thereby more accurately determining the probability of matching path intersections when processing the transport vehicles under control.

[0135] By comprehensively considering both the first and second training errors, the Transformer algorithm is debugged. Numerical examples show that factors such as the vehicle's parking path planning label probability, prior confidence, and the values ​​in the initial vehicle environment state vector are all involved in the algorithm debugging. This helps improve the accuracy of the intelligent driving control processing algorithm, enabling it to more accurately handle the automatic parking situations of different types of transport vehicles (the first and second transport vehicles), reduce errors in parking path planning, improve the reliability and efficiency of the entire automatic parking system, and enhance the system's adaptability to different parking scenarios and vehicle states.

[0136] In another exemplary technical solution, the step of debugging the Transformer algorithm based on the prior information, parking path planning labels, and prior confidence scores of each historical transport vehicle to obtain the intelligent driving control processing algorithm includes: determining a first training error based on the prior information, parking path planning labels, and prior confidence scores of each historical transport vehicle; determining hyperparameter data based on the algorithm weights of the Transformer algorithm, wherein the hyperparameter data characterizes the long-distance dependency performance of the Transformer algorithm; determining a third training error based on the hyperparameter data and the number of historical transport vehicles; and debugging the Transformer algorithm based on the first training error and the third training error to obtain the intelligent driving control processing algorithm.

[0137] In this exemplary technical solution, the entire process revolves around debugging the Transformer algorithm based on the prior information of each historical transport vehicle, parking path planning labels, and prior confidence, thereby obtaining the intelligent driving control processing algorithm. This process is of great significance in the automatic parking scenario of ski resorts.

[0138] Ski resorts present unique challenges, including heavy snowfall, complex terrain, and frequent movement of people and equipment. These factors place high demands on automated parking systems for transport vehicles. In such scenarios, historical data on transport vehicles becomes crucial for optimizing automated parking algorithms.

[0139] First, the first training error is determined based on the prior information, parking path planning label, and prior confidence level of each historical transport vehicle. For each historical transport vehicle performing automatic parking tasks at the ski resort, the prior information reflects whether there is a path intersection with the automatic parking control strategy during the automatic parking process. In the ski resort parking lot, due to the potentially compact layout of parking spaces and the possibility of snowdrifts, temporary storage of ski equipment, and irregular parking of other vehicles, the existence of path intersections directly affects the safety and efficiency of parking. The parking planning label represents the probability that the vehicle matches a path intersection. For example, if the prior information of a historical transport vehicle indicates the existence of a path intersection, its parking path planning label has a probability of 0.55, and the prior confidence level is 0.65.

[0140] Determining the first training error requires following specific algorithmic logic. First, an intermediate value is determined based on the relationship between prior information and parking path planning labels. This intermediate value is related to the specific values ​​and relationships of the prior information and parking path planning labels and is obtained according to the algorithm's internal calculation rules. In the ski resort scenario, the calculation of this intermediate value needs to consider the impact of the ski resort's unique environment. For example, snow cover may cause some deviation in the sensor's measurement of the distance to surrounding objects, and this deviation will be reflected in the numerical relationship between the prior information and parking path planning labels. Then, this intermediate value is calculated with the prior confidence level according to the prescribed algorithm to obtain an error correlation value for each historical transport vehicle. Finally, by comprehensively processing these error correlation values ​​of all historical transport vehicles (e.g., summation, weighted averaging, etc., determined by the algorithm's internal logic), the first training error is determined.

[0141] Next, hyperparameter data is determined based on the algorithm weights of the Transformer algorithm. This hyperparameter data characterizes the long-range dependency performance of the Transformer algorithm. In the ski resort automated parking scenario, the algorithm weights in the Transformer algorithm are a crucial part of its internal structure. These weights determine how the algorithm processes input data and how it transmits information between different neurons or modules. For example, in a ski resort parking lot, a transport vehicle may need to travel from a distant entrance to its parking position, requiring the processing of environmental information over a long distance. In the multi-head attention mechanism of the Transformer algorithm, different heads may have different weights, which affect the degree of attention paid to different representation subspaces of the input. By specifically analyzing and processing these algorithm weights, hyperparameter data can be obtained. This hyperparameter data reflects the algorithm's performance in handling long-range dependencies, which are crucial when processing multi-sensor data in scenarios like ski resort automated parking, because the vehicle's state and surrounding environmental information may have complex relationships at different temporal and spatial scales. For example, distant snowdrifts or other vehicles on the road may affect the current parking path of the transport vehicle, and the algorithm needs to accurately process this long-range correlation information.

[0142] Then, the third training error is determined based on the hyperparameter data and the number of historical transport vehicles. The hyperparameter data, as an indicator of the algorithm's performance in handling long-range dependencies, is combined with the number of historical transport vehicles to determine the third training error. In the scenario of automatic parking at a ski resort, if the hyperparameter data indicates a certain bias in the algorithm's handling of long-range dependencies, and the number of historical transport vehicles is large, this may mean that this bias will be amplified during the processing of large amounts of data. For example, a ski resort may have a large number of transport vehicles performing parking tasks at different times, accumulating a large amount of historical data. When there is a bias in the algorithm's handling of long-range dependencies, this bias will produce a larger value when determining the third training error as the number of historical transport vehicles increases. The specific determination method is based on the logic set internally by the algorithm, which may involve performing some mathematical operations or logical judgments on the hyperparameter data and the number of historical transport vehicles to obtain a third training error that reflects the impact of this relationship on algorithm debugging.

[0143] Finally, the Transformer algorithm was debugged based on the first and third training errors to obtain the intelligent driving control processing algorithm. Both the first and third training errors are crucial for debugging the Transformer algorithm. The first training error reflects the algorithm's deviation in processing information related to automatic vehicle parking from the perspectives of prior knowledge of historical transport vehicles, parking path planning labels, and prior confidence. The third training error reflects another type of deviation from the perspectives of the Transformer algorithm's long-distance dependency performance and the number of historical transport vehicles.

[0144] During the debugging of automatic parking at the ski resort, a large first training error indicates a significant problem in processing historical parking information of transport vehicles. Given the unique environment of a ski resort, adjustments may be needed to the calculations related to prior information, parking path planning labels, and prior confidence levels. For example, since snow accumulation can affect vehicle trajectory and the detection of surrounding objects, the weights of factors such as distance to snow-covered obstacles and vehicle speed (considering the special circumstances of driving on snow) need to be adjusted when calculating parking path planning labels, or the calculation method for prior confidence levels may need to be modified. A large third training error may require adjustments to the weights in the Transformer algorithm, particularly those related to handling long-distance dependencies. For example, adjusting the weight allocation of different heads in a multi-head attention mechanism, or adjusting the weights of different positional information when processing sequential data, can optimize the algorithm's ability to handle long-distance dependencies. By comprehensively considering these two types of training errors and thoroughly debugging the Transformer algorithm, a more accurate intelligent driving control algorithm for handling automatic parking of transport vehicles at ski resorts can be obtained.

[0145] The Transformer algorithm is debugged by combining the first and third training errors. Numerical examples show that the probability of parking path planning labels for historical transport vehicles, prior confidence levels, and hyperparameter data related to the weights of the Transformer algorithm are all involved. In the context of automated parking at ski resorts, this helps improve the accuracy of the intelligent driving control processing algorithm, enabling it to better handle the complex relationships in automated parking at ski resorts, such as the impact of snow accumulation, irregular parking, and long-distance driving, reducing parking errors, improving the reliability and efficiency of the parking system, and enhancing its adaptability to different parking conditions at ski resorts.

[0146] In other exemplary technical solutions, the step of debugging the Transformer algorithm based on the prior information, parking path planning labels, and prior confidence of each historical transport vehicle to obtain the intelligent driving control processing algorithm includes: determining a first training error based on the prior information, parking path planning labels, and prior confidence of each historical transport vehicle; determining a fourth training error based on the prior confidence of each historical transport vehicle and the number of historical transport vehicles; and debugging the Transformer algorithm based on the first training error and the fourth training error to obtain the intelligent driving control processing algorithm.

[0147] In detail, the above technical solution revolves around debugging the Transformer algorithm based on prior data from each historical transport vehicle, parking path planning labels, and prior confidence levels to obtain the intelligent driving control processing algorithm. The unique environment of ski resorts presents numerous challenges to automated parking. The layout of the parking lot can become complex due to snow accumulation, temporary parking of ski equipment, etc., and snow-covered ground can affect vehicle movement and sensor perception capabilities. In such scenarios, debugging the Transformer algorithm based on prior data from each historical transport vehicle, parking path planning labels, and prior confidence levels is crucial for optimizing the automated parking system.

[0148] First, the first training error is determined based on the prior information, parking path planning label, and prior confidence level of each historical transport vehicle. For each historical transport vehicle performing automatic parking tasks at the ski resort, the prior information represents whether there is a path intersection point with the automatic parking control strategy during the automatic parking process. Determining path intersection points is more complex at ski resorts because, in addition to regular vehicles and parking facilities, there may be factors such as skiers crossing the road and snowdrifts obstructing the view. The parking path planning label represents the probability that the vehicle matches a path intersection point, and the prior confidence level reflects the reliability of the vehicle's initial vehicle environment state vector. For example, if the prior information of a certain historical transport vehicle indicates the existence of a path intersection point, its parking path planning label has a probability of 0.6, and its prior confidence level is 0.7.

[0149] Determining the first training error requires following specific algorithmic logic. First, an intermediate result is derived based on the relationship between prior data and parking path planning labels. This intermediate result is related to the specific values ​​and interrelationships of the prior data and parking path planning labels, and is calculated according to the algorithm's internal rules. In a ski resort environment, snow accumulation affects sensors; for example, ultrasonic sensors may deviate in distance measurement under snow conditions. This will be reflected in the numerical relationship between the prior data and parking path planning labels, thus affecting the calculation of the intermediate result. Then, this intermediate result is calculated with the prior confidence level according to a prescribed algorithm to obtain the error correlation value for each historical transport vehicle. For example, if the intermediate result is a certain value, it is calculated with the prior confidence level of 0.7 according to the algorithm to obtain a specific error correlation value. Finally, these error correlation values ​​for all historical transport vehicles are comprehensively processed. The processing method is determined by the algorithm's internal logic, such as summation or weighted averaging, to determine the first training error.

[0150] Next, the fourth training error is determined based on the prior confidence of each historical transport vehicle and the number of historical transport vehicles. The prior confidence reflects information such as the reliability of the initial vehicle environment state vector for each historical transport vehicle. In the ski resort automated parking scenario, the initial vehicle environment state vector of the historical transport vehicles contains more unique factors. For example, the vehicle's position coordinates may be difficult to determine precisely due to snow-covered landmarks, and the vehicle's speed may be unstable due to snow-covered road conditions. For instance, there may be multiple historical transport vehicles, some with prior confidence values ​​of 0.6, 0.7, 0.8, etc. The number of historical transport vehicles is also an important factor in this process. If the prior confidence is generally high and the number of historical transport vehicles is large, or if the prior confidence is generally low and the number of historical transport vehicles is large, it will have different impacts on the fourth training error. The specific method for determining the fourth training error is based on the logic set within the algorithm. This may involve performing some mathematical operation on the prior confidence (such as summation, averaging, etc.) and making logical judgments in conjunction with the number of historical transport vehicles, thereby obtaining a fourth training error that can reflect the impact of this relationship on algorithm debugging.

[0151] Finally, the Transformer algorithm was debugged based on the first and fourth training errors to obtain the intelligent driving control processing algorithm. The first and fourth training errors are crucial for debugging the Transformer algorithm. The first training error reflects the algorithm's deviation in processing information related to automatic vehicle parking from the perspectives of prior knowledge of historical transport vehicles, parking path planning labels, and prior confidence levels. For example, in a ski resort environment, a large first training error may indicate significant problems in processing the logical relationships between prior knowledge of historical transport vehicles, parking path planning labels, and prior confidence levels. Due to the special conditions of ski resorts, such as snow interference with sensors and the unpredictability of skiers, these relationships may become more complex and difficult to process accurately.

[0152] The fourth training error reflects another type of algorithm bias from the perspective of the prior confidence of historical transport vehicles and the number of historical transport vehicles. During debugging, if the first training error is large, it may be necessary to adjust the calculation parts of the algorithm related to prior information, parking path planning labels, and prior confidence. For example, the weights of factors such as the distance to objects around the vehicle and vehicle speed when calculating parking path planning labels, or the calculation method of prior confidence, etc. In the ski resort scenario, the measurement of the distance to objects around the vehicle may be inaccurate due to snow accumulation, and the control of vehicle speed also needs to consider factors such as the friction of the snow. If the fourth training error is large, it may be necessary to adjust the parts of the algorithm related to prior confidence and the number of historical transport vehicles, such as adjusting the algorithm logic when considering the relationship between the sum of prior confidence and the number of vehicles. By comprehensively considering these two types of training errors and conducting a comprehensive debugging of the Transformer algorithm, a more accurate intelligent driving control processing algorithm for handling the automatic parking of the controlled transport vehicle in the ski resort is finally obtained.

[0153] The Transformer algorithm was debugged by combining the first and fourth training errors. Numerical examples show that the probability of parking path planning labels for historical transport vehicles and prior confidence levels are all involved. In the context of automated parking at ski resorts, this helps improve the accuracy of the intelligent driving control processing algorithm, enabling it to better handle complex situations such as snow affecting sensors, and environmental complexity caused by skiers and snowdrifts. This reduces parking errors, improves the reliability and efficiency of the parking system, and enhances its adaptability to different parking scenarios and vehicle states at ski resorts.

[0154] In summary, this invention, by acquiring various IoT sensor information, can comprehensively understand the surrounding environment of the ski resort transport vehicle. The fusion of ultrasonic, geomagnetic, visual, and lidar sensor information can accurately reflect the vehicle's status. The intelligent driving control processing algorithm determines the initial vehicle environmental state vector, providing an accurate basis for subsequent decisions. Determining parking path planning labels helps to assess the risk of path intersections with control strategies in advance. Based on these labels, the decision to issue an automatic parking control strategy can improve parking safety and efficiency, avoid unnecessary collision risks, optimize parking lot resource utilization, and adapt to the complex and ever-changing parking needs of ski resorts.

[0155] This invention provides a transport vehicle that is communicatively connected to an IoT sensing and control system. The transport vehicle is used to receive automatic parking control strategies issued by the IoT sensing and control system. The IoT sensing and control system includes at least one processor and a memory. The memory stores computer execution instructions. The at least one processor executes the computer execution instructions stored in the memory, causing the at least one processor to perform the above-described method.

[0156] Furthermore, Figure 2 This is a schematic diagram of the structure of an IoT sensing and control system 200 provided in an embodiment of the present invention. Figure 2 The IoT sensing and control system 200 shown includes a processor 210, which can call and run computer programs from a memory to implement the methods in the embodiments of the present invention.

[0157] Optionally, such as Figure 2 As shown, the IoT sensing and control system 200 may further include a memory 230. The processor 210 can retrieve and run computer programs from the memory 230 to implement the methods described in this embodiment of the invention.

[0158] The memory 230 can be a separate device independent of the processor 210, or it can be integrated into the processor 210.

[0159] Optionally, such as Figure 2 As shown, the IoT sensing and control system 200 may also include a transceiver 220. The processor 210 can control the transceiver 220 to interact with other devices. Specifically, it can send information or data to other devices or receive information or data sent by other devices.

[0160] Optionally, the IoT sensing and control system 200 can implement the corresponding processes of the storage engine or components (such as processing modules) in the storage engine or the device with the storage engine deployed in the various methods of the embodiments of the present invention. For the sake of brevity, these will not be described in detail here.

[0161] It should be understood that the processor in this embodiment of the invention may be an integrated circuit chip with signal processing capabilities.

[0162] It is understood that the memory in the embodiments of the present invention may be volatile memory or non-volatile memory, or may include both volatile and non-volatile memory. It should be noted that the memory of the systems and methods described herein is intended to include, but is not limited to, suitable types of memory.

[0163] Based on the above, a readable storage medium is provided, on which a program or instructions are stored, and when the program or instructions are executed by a processor, the steps of the above method are implemented.

[0164] The embodiments of the present invention have been described above with reference to the accompanying drawings. However, the embodiments of the present invention are not limited to the specific implementation methods described above. The specific implementation methods described above are merely illustrative and not restrictive. Those skilled in the art can make many other forms under the guidance of the embodiments of the present invention without departing from the spirit and scope of protection of the embodiments of the present invention, and all of these forms are within the protection scope of the embodiments of the present invention.

Claims

1. A method for automatic parking control in ski resorts based on Internet of Things (IoT) sensing, characterized in that, The method is applied to an IoT sensing and control system, and the method includes: The first IoT sensing information of the transport vehicle to be controlled is obtained, wherein the transport vehicle to be controlled is a transport vehicle in the ski resort area; the first IoT sensing information includes ultrasonic sensing information, geomagnetic sensing information, visual sensing information and lidar sensing information; The first IoT sensor information of the transport vehicle to be controlled is processed using an intelligent driving control processing algorithm to determine the initial vehicle environment state vector of the transport vehicle to be controlled. The intelligent driving control processing algorithm is used to process the initial vehicle environment state vector of the vehicle to be controlled in order to determine the parking path planning label of the vehicle to be controlled. The parking path planning label of the vehicle to be controlled represents the probability that the vehicle to be controlled has a path intersection point with the automatic parking control strategy during the automatic parking process. Based on the parking path planning label of the vehicle to be controlled, determine whether to issue the automatic parking control strategy to the vehicle to be controlled; The parking path planning label of the vehicle to be controlled includes a first parking path planning label and a second parking path planning label. The first parking path planning label represents the probability that the vehicle to be controlled will match the path intersection point when an automatic parking control strategy is issued to the vehicle to be controlled. The second parking path planning label represents the probability that the vehicle to be controlled will match the path intersection point when an automatic parking control strategy is not issued to the vehicle to be controlled. The step of determining whether to issue the automatic parking control strategy to the vehicle under control based on the parking path planning label of the vehicle under control includes: If the first and second discrimination mechanisms are met, the automatic parking control strategy is issued to the transport vehicle to be controlled. The first discrimination mechanism is met when the weight of the first parking path planning label is greater than the weight of the second parking path planning label. The second discrimination mechanism is met when the weight difference between the first parking path planning label and the second parking path planning label is greater than the set weight. If at least one of the first and second discrimination mechanisms is not met, then it is determined that the automatic parking control strategy will not be issued to the transport vehicle to be controlled.

2. The method as described in claim 1, characterized in that, The method further includes: The system acquires the second IoT sensor information of multiple historical transport vehicles and the prior information of each historical transport vehicle. The prior information of the historical transport vehicles indicates whether there are any path intersections between the historical transport vehicles and the automatic parking control strategy during the automatic parking process. The Transformer algorithm is used to mine the state features of the second IoT sensor information of the multiple historical transport vehicles to obtain the initial vehicle environment state vector of each historical transport vehicle. The Transformer algorithm is used to determine the parking path planning label of each historical transport vehicle based on the initial vehicle environment state vector of each historical transport vehicle. The parking path planning label of the historical transport vehicle represents the probability that the historical transport vehicle matches the path intersection point. Based on the initial vehicle environment state vector of each historical transport vehicle, the prior confidence of each historical transport vehicle is determined. The prior confidence of the historical transport vehicle is used to adjust the initial vehicle environment state vector of the historical transport vehicle to the target mode. Based on the prior information, parking path planning labels, and prior confidence scores of each historical transport vehicle, the Transformer algorithm is debugged to obtain an intelligent driving control processing algorithm. This intelligent driving control processing algorithm is used to determine the probability that the transport vehicle to be controlled will match the path intersection point.

3. The method as described in claim 2, characterized in that, The plurality of historical transport vehicles include a plurality of first transport vehicles to which the automatic parking control strategy has been issued and a plurality of second transport vehicles to which the automatic parking control strategy has not been issued; The step of using the Transformer algorithm to determine the parking path planning label for each historical transport vehicle based on the initial vehicle environment state vector includes: The Transformer algorithm is used to determine the parking path planning label of each of the first transport vehicles based on the first debugging annotation and the initial vehicle environment state vector of the plurality of first transport vehicles. The first debugging annotation represents the result of the automatic parking control strategy being issued. The Transformer algorithm is used to determine the parking path planning label of each of the second transport vehicles based on the second debugging annotation and the initial vehicle environment state vector of the plurality of second transport vehicles. The second debugging annotation represents the result of not issuing an automatic parking control strategy.

4. The method as described in claim 2, characterized in that, The step of determining the prior confidence level of each historical transport vehicle based on its initial vehicle environment state vector includes: Feature mapping is performed on the initial vehicle environment state vector of each historical transport vehicle to obtain the driving control decision mapping vector of each historical transport vehicle. Based on the driving control decision mapping vector of each historical transport vehicle, the prior confidence level of each historical transport vehicle is determined; The driving control decision mapping vector of the historical transport vehicle includes the local control decision mapping vector of the historical transport vehicle in multiple attention indicators; The step of determining the prior confidence level of each historical transport vehicle based on its driving control decision mapping vector includes: For any historical transport vehicle, the comparison result of the first mapping vector is determined based on the local control decision mapping vector of the historical transport vehicle in the first attention index and the local control decision mapping vector of each historical transport vehicle in the first attention index. The first attention index is any one of the multiple attention indices. Based on the local control decision mapping vector of any historical transport vehicle in the second attention index and the local control decision mapping vector of each historical transport vehicle in the second attention index, the comparison result of the second mapping vector is determined. The second attention index is any attention index other than the first attention index among the plurality of attention indices. Based on the first mapping vector comparison result and the second mapping vector comparison result, determine the local control decision mapping correlation degree between the first attention index and the second attention index for any historical transport vehicle; Based on the correlation degree of the local control decision mapping between each pair of attention indicators for each historical transport vehicle, the prior confidence of each historical transport vehicle is determined. The step of determining the first mapping vector comparison result based on the local control decision mapping vector of any historical transport vehicle in the first attention index and the local control decision mapping vector of each historical transport vehicle in the first attention index includes: Based on the basic confidence level of each historical transport vehicle and the local control decision mapping vector of each historical transport vehicle in the first attention index, the mean vector of the local control decision mapping of the first attention index is determined. Based on the basic confidence level of any historical transport vehicle and the local control decision mapping vector of any historical transport vehicle in the first attention index, determine the local control decision mapping reinforcement vector of any historical transport vehicle in the first attention index. The vector operation value between the local control decision mapping enhancement vector and the local control decision mapping mean vector of the first attention index for any historical transport vehicle is used as the comparison result of the first mapping vector.

5. The method as described in claim 2, characterized in that, Based on the prior information, parking path planning labels, and prior confidence levels of each historical transport vehicle, the Transformer algorithm is debugged to obtain the intelligent driving control processing algorithm, including: For any historical transport vehicle, based on the prior information of the historical transport vehicle and the parking path planning label of the historical transport vehicle, the error of the first path planning node corresponding to the historical transport vehicle is determined. Based on the error of the first path planning node corresponding to the historical transport vehicle and the prior confidence level of the historical transport vehicle, the weighted error of the first path planning node of the historical transport vehicle is determined. The first training error is determined based on the weighted error of the first path planning nodes of each historical transport vehicle. The Transformer algorithm is adjusted based on the first training error to obtain the intelligent driving control processing algorithm.

6. The method as described in claim 2, characterized in that, The plurality of historical transport vehicles include a plurality of first transport vehicles to which the automatic parking control strategy has been issued and a plurality of second transport vehicles to which the automatic parking control strategy has not been issued; Based on the prior information, parking path planning labels, and prior confidence levels of each historical transport vehicle, the Transformer algorithm is debugged to obtain the intelligent driving control processing algorithm, including: Based on the prior information, parking path planning labels, and prior confidence levels of each historical transport vehicle, the first training error is determined. The second training error is determined based on the comparison results of the initial state vectors between the initial vehicle environment state vectors of the plurality of first transport vehicles and the initial vehicle environment state vectors of the plurality of second transport vehicles. Based on the first training error and the second training error, the Transformer algorithm is debugged to obtain the intelligent driving control processing algorithm.

7. The method as described in claim 2, characterized in that, Based on the prior information, parking path planning labels, and prior confidence levels of each historical transport vehicle, the Transformer algorithm is debugged to obtain the intelligent driving control processing algorithm, including: Based on the prior information, parking path planning labels, and prior confidence levels of each historical transport vehicle, the first training error is determined. Hyperparameter data is determined based on the algorithm weights of the Transformer algorithm, and the hyperparameter data characterizes the long-distance dependency performance of the Transformer algorithm; Based on the hyperparameter data and the number of historical transport vehicles, the third training error is determined; Based on the first training error and the third training error, the Transformer algorithm is debugged to obtain the intelligent driving control processing algorithm.

8. The method as described in claim 2, characterized in that, Based on the prior information, parking path planning labels, and prior confidence levels of each historical transport vehicle, the Transformer algorithm is debugged to obtain the intelligent driving control processing algorithm, including: Based on the prior information, parking path planning labels, and prior confidence levels of each historical transport vehicle, the first training error is determined. The fourth training error is determined based on the prior confidence level of each historical transport vehicle and the number of historical transport vehicles. Based on the first training error and the fourth training error, the Transformer algorithm is debugged to obtain the intelligent driving control processing algorithm.

9. A transport vehicle, characterized in that, The transport vehicle is communicatively connected to the IoT sensing and control system; the transport vehicle is used to receive automatic parking control strategies issued by the IoT sensing and control system; the IoT sensing and control system includes at least one processor and a memory; the memory stores computer execution instructions; the at least one processor executes the computer execution instructions stored in the memory, causing the at least one processor to perform the method according to any one of claims 1-8.

Images