Parking method and device based on dynamic semantic field, medium and equipment

Abstract

The invention discloses a parking method and device based on a dynamic semantic field, a medium and equipment, and the method comprises the steps that the dynamic semantic field at the current moment is obtained in response to a parking instruction, the dynamic semantic field is a model which describes static environment characteristics of a target parking environment and real-time traffic flow through semantic characteristics, and the static environment characteristics of the target parking environment and the real-time traffic flow are obtained; according to the dynamic semantic field, a plurality of candidate parking paths are determined, a target parking path is determined from the plurality of candidate parking paths based on a preset strategy, and a parking scheme is generated and executed according to the target parking path. By constructing a parking environment digital model fusing a static environment and dynamic traffic information, upgrading from a static map to dynamic situation awareness is realized. Based on the dynamic semantic field, each vehicle can be regarded as an edge calculation node, respective parking paths are generated in real time, high-frequency data are processed through distributed edge calculation, and the real-time performance, stability and system expandability of parking decision making are effectively improved.

Description

Technical Field

[0001] This application relates to the field of intelligent driving, and in particular to a parking method, apparatus, medium and device based on a dynamic semantic field. Background Technology

[0002] With the development of intelligent driving, automatic parking has gradually become an important breakthrough in the construction of a smart transportation ecosystem. It can reduce the time spent by people searching for parking spaces and making repeated adjustments through accurate and efficient parking space recognition and parking, thereby improving the turnover rate of the parking environment and alleviating congestion.

[0003] To achieve automatic parking, existing technologies typically employ memory parking technology. This involves the driver manually navigating the parking route during the initial parking maneuver, then memorizing the route at the entrance of the parking garage or parking environment, and finally remembering the route after parking so that subsequent automatic parking can be achieved based on the memorized route.

[0004] However, memory parking relies on preset environmental features and path parameters for positioning and navigation, which is easily affected by dynamic changes in the parking environment, causing the automatic parking function to fail and the automatic parking task to be unable to be performed. Summary of the Invention

[0005] In view of this, the embodiments of the present invention aim to provide a parking method, apparatus, medium and device based on dynamic semantic field, so as to solve the problem that memory parking relies on preset environmental features and path parameters for positioning and navigation, which is easily affected by dynamic changes in the parking environment, resulting in the failure of automatic parking function and inability to perform automatic parking tasks.

[0006] This invention provides a parking method based on a dynamic semantic field, the method comprising: In response to a parking instruction, the dynamic semantic field at the current moment is obtained, wherein the dynamic semantic field is a model that describes the static environmental features of the target parking environment and the real-time traffic flow through semantic features. Based on the dynamic semantic field, multiple candidate parking paths are determined; Based on a preset strategy, a target parking path is determined from the multiple candidate parking paths; Based on the target parking path, a parking plan is generated, which guides the vehicle to park in the parking space.

[0007] In some embodiments of this application, the step of determining the target parking path from the plurality of candidate parking paths based on a preset strategy specifically includes: For each of the candidate parking paths, the total cost required to complete the candidate parking path is estimated, wherein the total cost includes at least one of distance cost, time cost, probability of successful parking, and probability of conflict with other vehicles; The target parking path is determined from the plurality of candidate parking paths based on the total cost corresponding to each candidate parking path.

[0008] In some embodiments of this application, the method further includes: When the target parking path conflicts with the parking path of other vehicles, a coordination request is sent to the server of the target parking environment. Obtain the passage order returned by the server, wherein the passage order is determined by the server based on the coordination request and preset rules; Based on the traffic flow, the parking plan is adjusted and the adjusted parking plan is executed.

[0009] In some embodiments of this application, the method further includes: Based on the parking scheme and key indicators used to characterize the execution results of the parking scheme, the preset strategy is optimized, wherein the key indicators include at least one of time consumption, number of conflicts, and success rate.

[0010] In some embodiments of this application, the step of optimizing the preset strategy based on the parking scheme and key indicators used to characterize the execution result of the parking scheme specifically includes: Update the reward function based on the aforementioned key indicators; In a pre-built digital twin training environment, the parameters of the preset strategy are adjusted based on the updated reward function; The performance of the adjusted preset strategy is evaluated. When the evaluation result is a performance improvement, the preset strategy is updated according to the adjusted parameters. If the evaluation result indicates no performance improvement, the parameters of the preset strategy are readjusted until the evaluation result indicates a performance improvement.

[0011] In some embodiments of this application, the method further includes: Collect environmental data and its own status data; The environmental data and / or the state data are sent to the server of the target parking environment, wherein the environmental data and / or the state data are used to update the dynamic semantic field.

[0012] In some embodiments of this application, the method further includes: When parking is abnormal, determine the cause of the abnormality, wherein the cause of the abnormality includes at least one of sensor failure, communication interruption, path blockage, and vehicle system failure; Based on the cause of the anomaly, the corresponding emergency plan is activated, and a recovery plan is determined according to the cause of the anomaly. The recovery plan is used to repair the parking anomaly. Recheck for any anomalies; if so, determine a new recovery plan. If not, a new target parking path is determined, and a new parking plan is generated and executed based on the new target parking path.

[0013] In some embodiments of this application, the method is applied to an in-vehicle intelligent agent.

[0014] Secondly, embodiments of this application provide a parking method based on a dynamic semantic field, the method being applied to a cloud server in a parking environment, the method comprising: In response to the acquisition instruction, the most recently updated dynamic semantic field is determined, wherein the dynamic semantic field is a model that describes the static environmental features of the target parking environment and the real-time traffic flow through semantic features. Return the most recently updated dynamic semantic field.

[0015] In some embodiments of this application, the step of updating the dynamic semantic field specifically includes: Receive environmental data and / or vehicle status data, and perform data preprocessing on the environmental data and / or vehicle status data, wherein the data preprocessing includes at least one of outlier filtering, coordinate system I, and timestamp alignment; The preprocessed environmental data and / or vehicle status data are converted into semantic features; Update the dynamic semantic field based on the semantic features.

[0016] In some embodiments of this application, the step of updating the dynamic semantic field based on the semantic features specifically includes: Based on the semantic features, the dynamic semantic field is pre-updated; The pre-update results of the dynamic semantic field are evaluated according to the preset evaluation criteria. When the evaluation result of the pre-update result is qualified, the updated dynamic semantic field is determined; When the evaluation result of the pre-update result is unqualified, a re-acquisition instruction is returned, and the environmental data and / or the vehicle status data are reacquired to re-update the dynamic semantic field until the evaluation result of the pre-update result is qualified.

[0017] Thirdly, embodiments of this application provide a parking device based on a dynamic semantic field, comprising: The acquisition module is used to acquire the dynamic semantic field at the current moment in response to the parking instruction. The dynamic semantic field is a model that describes the static environmental features of the target parking environment and the real-time traffic flow through semantic features. The candidate module is used to determine multiple candidate parking paths based on the dynamic semantic field. The selection module is used to determine the target parking path from the multiple candidate parking paths based on a preset strategy. The execution module is used to generate a parking plan based on the target parking path, and the parking plan is used to guide the vehicle to park in the parking space.

[0018] Fourthly, embodiments of this application provide a parking device based on a dynamic semantic field, the device being applied to a cloud server in a parking environment, comprising: The response module is used to determine the most recently updated dynamic semantic field in response to the acquired instruction. The dynamic semantic field is a model that describes the static environmental features of the target parking environment and the real-time traffic flow through semantic features. The sending module is used to return the most recently updated dynamic semantic field.

[0019] Fifthly, embodiments of this application provide a vehicle, the vehicle including: a memory and a processor; the memory is used to store related program code; the processor is used to call the program code to execute the above-described method.

[0020] Sixthly, embodiments of this application provide an electronic device, the electronic device comprising: a memory and a processor; the memory being used to store related program code; and the processor being used to call the program code to execute the above-described method.

[0021] In a seventh aspect, embodiments of this application provide a computer-readable storage medium for storing a computer program for performing the methods described above.

[0022] The above-mentioned technical solution adopted in this application can achieve the following beneficial effects: This application provides a parking method based on a dynamic semantic field, which upgrades from static maps to dynamic situational awareness by constructing a digital model of the parking environment that integrates static and dynamic traffic information. Based on this dynamic semantic field, each vehicle can be regarded as an edge computing node, generating its own parking path in real time. High-frequency data is processed through distributed edge computing, effectively improving the real-time performance, stability, and scalability of parking decisions. Attached Figure Description

[0023] The accompanying drawings, which are included to provide a further understanding of this application and form part of this application, illustrate exemplary embodiments of this application and are used to explain this application, but do not constitute an undue limitation of this application. In the drawings...

[0024] Figure 1 This is a schematic diagram illustrating the process of a parking method based on a dynamic semantic field, as provided in an embodiment of this application.

[0025] Figure 2 This is a schematic diagram of a target parking environment provided in an embodiment of this application.

[0026] Figure 3 This is a schematic diagram illustrating the process of a parking method based on a dynamic semantic field, as provided in an embodiment of this application.

[0027] Figure 4 This is a schematic diagram of a parking solution system provided in an embodiment of this application.

[0028] Figure 5 A schematic diagram of a parking device based on a dynamic semantic field provided in this application.

[0029] Figure 6 A schematic diagram of a parking device based on a dynamic semantic field provided in this application.

[0030] Figure 7 A corresponding to the present application Figure 1 and Figure 2 A schematic diagram of the structure of an electronic device. Detailed Implementation

[0031] To make the objectives, technical solutions, and advantages of this application clearer, the technical solutions of this application will be clearly and completely described below in conjunction with specific embodiments and corresponding drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0032] The terminology used in the embodiments of this invention is for the purpose of describing particular embodiments only and is not intended to limit the invention. The singular forms “a,” “the,” and “the” as used in the embodiments of this invention and the appended claims are also intended to include the plural forms unless the context clearly indicates otherwise.

[0033] It should be understood that the term "and / or" used in this article is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, and B existing alone. Additionally, the character " / " in this article generally indicates that the preceding and following related objects have an "or" relationship.

[0034] Depending on the context, the word "if" as used here can be interpreted as "when," "when," "in response to determination," or "in response to detection." Similarly, depending on the context, the phrase "if determination" or "if detection (of the stated condition or event)" can be interpreted as "when determination," "in response to determination," "when detection (of the stated condition or event)," or "in response to detection (of the stated condition or event)."

[0035] With the development of intelligent driving, automatic parking has gradually become an important breakthrough in the construction of a smart transportation ecosystem. It can reduce the time spent by people searching for parking spaces and making repeated adjustments through accurate and efficient parking space recognition and parking, thereby improving the turnover rate of the parking environment and alleviating congestion.

[0036] Current autonomous parking systems typically employ memory parking technology. This means that when parking for the first time, the driver manually drives the vehicle along the parking route, then memorizes the route at the entrance of the parking garage or parking environment, and completes the route memorization after parking, so that subsequent automatic parking can be achieved based on the memorized route.

[0037] However, memory parking has poor environmental adaptability and is easily affected by dynamic changes in the parking environment. If temporary obstacles appear on the path, parking may be interrupted or even fail, or even cause a collision risk.

[0038] Furthermore, in multi-vehicle collaboration scenarios, vehicles may get stuck in congestion during parking because they cannot obtain real-time traffic flow changes; at the same time, the lack of collaboration mechanisms between vehicles can easily lead to path conflicts and increase the risk of collisions.

[0039] On the other hand, existing memory parking systems rely on preset paths and fixed rules, lacking self-learning and upgrade capabilities, which limits the adaptability of automatic parking and affects the user's parking experience. Therefore, this application provides a parking method based on dynamic semantic fields.

[0040] like Figure 1 As shown, Figure 1 This is a schematic diagram of a parking method based on a dynamic semantic field provided in an embodiment of this application, including the following steps.

[0041] S100: In response to a parking instruction, obtain the dynamic semantic field at the current moment, wherein the dynamic semantic field is a model that describes the static environmental features of the target parking environment and the real-time traffic flow through semantic features.

[0042] In one or more embodiments of this application, the specific device executing the parking method is not limited, such as a mobile terminal or server that can interact with or control the vehicle. However, since subsequent steps involve parking path generation and parking scheme execution, which are generally executed by the on-board processor, this application describes the parking method using an on-board processor as the executing entity. The on-board processor can be a microcontroller unit (MCU), a system-on-a-chip (SoC), a microprocessor unit (MPU), a multi-sensor fusion chip, etc. This application does not impose any restrictions and can be configured according to actual needs.

[0043] Since dynamic traffic flow in the target parking environment often occurs locally, traditional parking lot models typically acquire dynamic traffic data and then globally reconstruct the parking lot model during updates, resulting in high computational cost and long processing time. Therefore, to achieve efficient parking, this onboard processor can respond to parking commands by first acquiring the dynamic semantic field of the target parking environment at the current moment.

[0044] The dynamic semantic field refers to a model that describes the static environmental features and real-time traffic flow features of the target parking environment through semantic features. It converts real-time traffic flow data into semantic features to locally update the dynamic semantic field, thereby representing the real-time traffic flow of the target parking environment.

[0045] As an example, such as Figure 2 As shown, Figure 2 This is a schematic diagram of a target parking environment provided in an embodiment of this application. The target parking environment includes two rows of parking spaces, each with five spaces, surrounded by a rectangular ring road, designated as Road A, Road B, Road C, and Road D. At the current moment, a car E is traveling at 30 km / h on Road D towards the second parking space in the second row. There are two cars in parking spaces 2 and 4 respectively. Therefore, the dynamic semantic field of this target parking environment at the current moment can be: Static semantic layer: Parking space status: Parking space 1: Vacant parking space, size = standard, unobstructed; Parking space 2: Occupies parking space, vehicle type = sedan; Parking space 3: Vacant parking space, size = standard; Parking space 4: Occupies parking space, vehicle type = sedan; Parking space 5: Vacant parking space, size = standard, near the exit; Parking space 6: Available parking space, size = standard; Parking space 7: Available parking space, size = standard; Parking space 8: Available parking space, size = standard; Parking space 9: Vacant parking space, size = standard; Parking space 10: Available parking space, size = standard; the way: Road A: East-west main road, width = 6 meters, one-way traffic (east → west); Road B: North-South auxiliary road, 4 meters wide, one-way traffic (North → South). Road C: East-west service road, 5 meters wide, two-way traffic; Road D: North-south entrance road, width = 4 meters, one-way traffic (south → north); Lane 1 (between two rows of parking spaces): Internal passageway, 3 meters wide, for two-way traffic; Lane 2 (in front of the south parking space): Parking service lane, width = 2.5 meters, two-way traffic; Infrastructure: Entrance: Main entrance, Location = South end of road D, Status = Open; Exit: Main exit, Location = North end of Road B, Status = On; Sidewalk: Pedestrian walkway, location = north side of road A, width = 2 meters.

[0046] Traffic flow characteristics: Road D speed field (southbound → northbound): [Entrance] 0km / h → 20km / h → 30km / h (E vehicle) → 15km / h → [intersection]; Internal traffic density field: Lane 1: Vacancy rate = 100%; Lane 2: Vacancy rate = 100%. The aforementioned dynamic semantic field can rapidly update traffic flow features based on real-time traffic flow updates, thereby achieving real-time characterization of the traffic flow in the target parking environment. It should be noted that the aforementioned dynamic semantic field is merely one embodiment provided in this application. In one or more embodiments of this application, the specific description of the static environmental features and real-time traffic flow features of the target parking environment within the dynamic semantic field is not limited. It can be determined based on the actual semantic conversion tool used, and this application does not impose any restrictions on this. Furthermore, in one or more embodiments of this application, the specific device used to construct the dynamic semantic field is not limited. It can be constructed in real-time by the server of the target parking environment based on sensors, cameras, and other devices. Alternatively, it can be updated by the cloud server of the target parking environment based on static environmental information and dynamic traffic information obtained from each vehicle. Each vehicle can also update the dynamic semantic field based on static environmental information and dynamic traffic information sent by other vehicles, and / or static environmental information and dynamic traffic information collected by the parking environment server. This application does not impose any restrictions on this, and it can be set according to actual needs.

[0047] Furthermore, in one or more embodiments of this application, the parking instruction is not limited to the device from which it originates. It can be a parking instruction generated by the user clicking on the in-vehicle device when operating the vehicle, or a parking instruction issued by the user through a terminal or remote server, or a parking instruction issued by a device in the target parking environment after the vehicle to be parked enters the target parking environment. There are many possible solutions, which are not listed one by one in this application, and can be set according to actual needs.

[0048] Furthermore, in one or more embodiments of this application, the specific type of parking instruction is not limited. It can be a series of electrical signals, a series of digital signals, or a series of codes. There are many possible forms, which will not be listed one by one in this application. The instruction can be set according to actual needs.

[0049] Furthermore, the target parking environment mentioned in the application may refer to a parking lot, a roadside parking space, or other environments suitable for long-term or temporary parking. This application does not impose any restrictions on this and the environment can be set according to actual needs.

[0050] Of course, in one or more embodiments of this application, the specific method by which the vehicle-mounted server obtains the dynamic semantic field is not limited. It may be obtained from a server in the target parking environment or a cloud server after entering the target parking environment. It may also be obtained from software in response to user operations on the terminal. Other methods may also be used; since there are many options, this application does not list them all, and the method can be set according to actual needs.

[0051] S102: Determine multiple candidate parking paths based on the dynamic semantic field.

[0052] After acquiring the dynamic semantic field, the onboard server can determine each candidate parking path based on this field, so as to select the target parking path from the candidate parking paths to achieve parking. The candidate parking path is a set of continuous motion trajectories generated from the vehicle's current parking or driving position to the target parking space, forming a specified parking posture, such as the vehicle body being centered and the direction being straight, based on the vehicle's current position, the spatial parameters of the target parking space (such as size, orientation, and coordinates), and surrounding environmental constraints (such as obstacles, road boundaries, and traffic rules). Each candidate parking path has at least one different parking space or parking path.

[0053] It should be noted that, in one or more embodiments of this application, the specific method used by the vehicle processor to determine each candidate parking path is not limited. The vehicle processor can determine each candidate parking path by traversing each path corresponding to each parking space based on the dynamic semantic field, or it can generate each candidate parking path based on a pre-trained algorithm. Other methods can also be used to determine each candidate parking path; this application does not limit this. Since there are many possible methods, they are not listed here, and the appropriate method can be set according to actual needs.

[0054] S104: Based on a preset strategy, determine the target parking path from the multiple candidate parking paths.

[0055] To improve parking safety and efficiency and ensure vehicles can park smoothly in complex environments, the onboard processor can determine the target parking path from multiple candidate parking paths based on a preset strategy.

[0056] Of course, in one or more embodiments of this application, the specific content of the preset strategy is not limited. For example, it may select the parking path with the lowest collision probability from the determined candidate parking paths as the target parking path. Alternatively, it may select the target parking path based on minimizing the total parking time. Or it may select the path that best conforms to the vehicle's kinematic constraints based on the vehicle model itself. Or it may determine the target parking path based on the user's preference. There are many possible methods, which will not be listed here. There is no limitation on this, and it can be set according to actual needs.

[0057] Furthermore, it should be noted that in one or more embodiments of this application, the method by which the preset strategy is driven is not limited. For example, it could be achieved by pre-training a model to learn the preset strategy and determine the target parking path. Alternatively, it could be based on preset logical rules to determine the target parking path. Or, it could be determined from pre-stored typical scenarios including parking space dimensions and starting poses. Since there are many possible methods, this application does not list them all; the method can be set according to actual needs.

[0058] S106: Generate a parking plan based on the target parking path, the parking plan being used to guide the vehicle into the parking space.

[0059] Once the target parking path is determined, the on-board processor can generate a parking plan based on the target parking path to guide the vehicle into the parking space. Of course, in one or more embodiments of this application, the specific method used by the on-board processor to generate the parking plan is not limited.

[0060] As one embodiment, the onboard processor can perform spatiotemporal parameterization of the motion trajectory based on the determined target parking path and parking space information, identify several discrete points, and then assign speed, acceleration, and steering angle change rate to each point on the path, forming a smooth and trackable trajectory in the time dimension. Then, based on this trajectory, a discrete control sequence is generated, including steering wheel angle, vehicle speed, or acceleration / braking commands, etc., and finally, the vehicle is controlled to complete parking based on the control sequence.

[0061] Of course, the control sequence described above can also be replaced by a series of voice or text commands to guide the user in completing the parking process. There are many options available, which will not be listed here; the specific configuration should be determined based on actual needs.

[0062] based on Figure 1 The parking method shown here is based on a dynamic semantic field. By constructing a digital model of the parking environment that integrates static and dynamic traffic information, it upgrades from static maps to dynamic situational awareness. Based on this dynamic semantic field, each vehicle can be regarded as an edge computing node, generating its own parking path in real time. High-frequency data is processed through distributed edge computing, effectively improving the real-time performance, stability, and scalability of parking decisions.

[0063] Furthermore, relying solely on a single objective when determining the target parking path may result in parking solutions with poor performance in other aspects, failing to effectively improve user experience and parking efficiency. For example, using the minimum collision probability as the selection strategy might significantly improve the driving safety of the selected target parking path, but it could also lead to a longer parking path. Therefore, this onboard processor uses multi-objective decision-making to determine the target parking path.

[0064] Specifically, for each candidate parking path, the total cost required to complete the candidate parking path is estimated. The total cost includes at least one of the following: distance cost, time cost, probability of successful parking, and probability of conflict with other vehicles. Then, based on the total cost corresponding to each candidate parking path, the target parking path is determined from multiple candidate parking paths.

[0065] It should be noted that the distance cost mentioned above refers to the total distance traveled from the vehicle's current location to the target parking space. The time cost refers to the total time required for the vehicle to park in the target parking space from its current location. The parking success probability refers to the assessment of the likelihood of safely and accurately completing the parking operation using the candidate parking path. The probability of conflict with other vehicles refers to the estimated likelihood of mutual interference or collision between the vehicle and other surrounding vehicles, including vehicles that are moving, stopped, or parking, during the parking process.

[0066] In one or more embodiments of this application, the method by which the vehicle processor determines each parameter of the total cost is not limited. It may be determined by simulating each parameter through a preset vehicle model, or it may be calculated based on the dynamic semantic field, or it may be obtained by simulating the target parking environment through a server. This application does not limit this and can set it according to actual needs.

[0067] Furthermore, in one or more embodiments of this application, the specific method used by the vehicle processor to calculate the total cost is not limited. It can be a summation of each parameter, with the summation result used as the total cost. Alternatively, it can be a weighted summation of each parameter, with the weighted summation result used as the total cost, depending on actual needs. The weight of each parameter can be set according to user preferences or actual needs, and the importance of each parameter is positively correlated with its weight.

[0068] In addition, the parking method provided in this application can be applied to multi-vehicle cooperative scenarios. However, in such scenarios, path conflicts may occur during the parking process, that is, the target parking paths of at least two vehicles may overlap at some point in the future, during a period of time, or in the current period of time, thereby leading to a collision risk or forcing one of them to change its intended behavior.

[0069] To prevent vehicle collisions or scrapes due to parking path conflicts, the onboard processor, after determining the target parking path, estimates the probability of a conflict based on a dynamic semantic field. When the conflict probability exceeds a threshold, it sends a coordination request to the cloud server of the target parking environment and obtains the passage order returned by the cloud server. This passage order is determined by the cloud server based on the coordination request and preset rules. Then, based on this passage order, it adjusts or generates a parking plan and executes it.

[0070] Then, if during the parking process, the onboard processor detects a conflict between a vehicle and its own path based on the sensors, it can also send a coordination request to the cloud server of the target parking environment, obtain the passage order returned by the cloud server, adjust the parking plan and execute it.

[0071] It should be noted that, in one or more embodiments of this application, the specific content of the preset rule is not limited. It may include first-come-first-served, turning yields to straight-going traffic, right-turn yields to left-turn, etc. It can be set according to actual needs, and this application does not limit it.

[0072] Furthermore, with the continuous growth of vehicle ownership and the increasing scarcity of urban space, parking environments are becoming increasingly complex. To adapt to different parking environments, this onboard processor can record the parking plan and key metrics characterizing its execution results after each parking task or execution, such as time consumption, number of conflicts, and success rate. This allows for subsequent optimization and training of preset strategies based on the recorded parking plans and key metrics. Additionally, after parking is completed, the processor can selectively store the parking plan and corresponding key metrics based on user feedback for further optimization of preset strategies.

[0073] Of course, in one or more embodiments of this application, the optimization method of the preset strategy is not limited, and the preset strategy can be optimized by reinforcement learning.

[0074] Specifically, adjusting the parameters of the preset strategy is defined as an action in reinforcement learning, the parking scheme and corresponding key indicators are the states of reinforcement learning, successful parking with correct posture is the final success reward, collision, timeout, human intervention are the final failure penalties, and key indicators such as time consumption, number of conflicts, and success probability are the process rewards to guide the optimization of the preset strategy.

[0075] Furthermore, in order to better optimize the preset strategy, the vehicle processor can also optimize the preset strategy through digital twin reinforcement learning.

[0076] Specifically, the reward function is updated based on the aforementioned key indicators. Then, in a pre-built digital twin training environment, the parameters of the preset strategy are adjusted based on the updated reward function. The performance of the adjusted preset strategy is then evaluated. If the evaluation result indicates performance improvement, the preset strategy is updated according to the adjusted parameters. If the evaluation result indicates no performance improvement, the parameters of the preset strategy are readjusted until the evaluation result indicates performance improvement.

[0077] It should be noted that, in one or more embodiments of this application, the specific method used by the vehicle processor to evaluate the performance of the adjusted preset strategy is not limited. It can be that the adjusted preset strategy is simulated and run in a digital twin environment, the running results are recorded, and the running results are compared with the running results of historical preset strategies to determine the performance of the adjusted preset strategy.

[0078] Furthermore, due to the limited computing resources of the onboard processor, in order to save computing resources, the onboard processor can also upload the stored parking path and key indicators to the main server. After the main server completes the update of the preset policy, the updated preset policy is loaded from the main server.

[0079] To protect user information security, a federated learning approach can also be used. This involves updating the preset policy locally on the in-vehicle server, uploading the updated parameters to the central server, and then aggregating the parameters returned by each vehicle on the central server. The in-vehicle processor then retrieves the aggregated parameters from the central server to update the preset policy. This allows for continuous self-optimization of the preset policy, improving the user's parking experience.

[0080] Furthermore, to obtain a more accurate dynamic semantic field, the onboard processor can also upload real-time environmental data and / or its own state data collected by sensors to a server in the target parking environment. This allows the server to update the dynamic semantic field more accurately based on real-time traffic flow information. The environmental data refers to static data about the environment within the preset range where the vehicle is located, while the vehicle's own state data includes information such as the vehicle's model, speed, and acceleration to indicate its real-time status, as well as data on the vehicle's target parking path and reserved parking space.

[0081] It should be noted that, in one or more embodiments of this application, the specific frequency at which the vehicle-mounted processor collects and uploads environmental data and / or status data is not limited. It can upload data according to a preset cycle, or it can upload the collected environmental data and / or status data to the cloud server only when the environment or its own status data undergoes significant changes. For example, after the vehicle-mounted processor determines a target parking path, it sends the determined target parking path and the corresponding parking space occupancy information to the server, and the server updates the dynamic semantic field to complete the parking space reservation. It can also respond to a collection command issued by the server and upload the collected environmental data and status data to the server. There are many possible solutions, which are not listed here; the specific settings can be configured according to actual needs.

[0082] Furthermore, in one or more embodiments of this application, the specific method used by the vehicle processor to update the dynamic semantic field is not limited. The update method will be described in detail later in this application, and will not be repeated here.

[0083] Furthermore, if an anomaly occurs during the execution of the parking plan, in order to avoid affecting automatic parking, the onboard processor can, upon confirming the anomaly, activate a historical parking plan, determine the cause of the anomaly, and, based on the cause, initiate a corresponding recovery plan. The cause of the anomaly can include at least one of sensor malfunction, communication interruption, path congestion, or onboard system malfunction. The recovery plan is used to correct the parking anomaly. Then, the processor re-checks for any remaining anomalies. If an anomaly is detected, a new recovery plan is determined and activated; otherwise, a new target parking path is determined, and a new parking plan is generated and executed based on the new target parking path.

[0084] Specifically, if the vehicle processor detects a parking anomaly, such as interruption or pause, unexpected movement, deviation from the parking path, or human-machine interaction warning, it can first determine the cause of the anomaly. The cause of the anomaly includes at least one of sensor failure, communication interruption, path blockage, or vehicle system failure. Then, based on the cause of the anomaly, the corresponding emergency plan is activated, and a recovery plan is determined according to the cause of the anomaly. The recovery plan is used to repair the parking anomaly.

[0085] It should be noted that, in one or more embodiments of this application, the method by which the vehicle processor determines the cause of the parking anomaly is not limited. It can use the processor's built-in fault diagnosis tree or expert rule base, taking the anomaly as input, and perform logical reasoning. Alternatively, it can invoke a pre-trained diagnostic model to determine the cause of the fault; this application does not impose any restrictions on this. Furthermore, the vehicle processor can perform fault detection in parallel for each anomaly cause, such as communication detection and sensor detection, or it can perform detection sequentially for each anomaly. The configuration can be tailored to actual needs.

[0086] Then, when activating the corresponding emergency plan, this application does not limit the specific content of the emergency plan. For example, it may continue parking based on historical parking paths, or it may determine a safe area and drive into it based on the most recently collected environmental information. Other automatic parking technologies may also be used, such as memory parking or parking based on real-time environmental perception by onboard sensors. The appropriate parking technology can be selected according to the actual cause of the anomaly to continue the parking task.

[0087] In addition, this onboard processor can determine corresponding emergency plans based on the cause of the anomaly. For example, if the anomaly is caused by a faulty onboard sensor, the processor can repair it through sensor redundancy switching. If the anomaly is caused by a communication interruption with the cloud server or a disruption of the vehicle's internal communication bus, local emergency planning can be used, generating a path to park the vehicle in a parking space or safe area based solely on the processor's basic built-in rules and residual sensor information. If the anomaly is caused by a blocked path, the processor can dynamically replan the parking path and re-determine the target parking route. If the anomaly is caused by a malfunction in the onboard system, it can automatically activate a safety mode, such as slowing down or pulling over.

[0088] While activating the emergency response plan, the onboard processor can also determine the corresponding recovery plan based on the cause of the anomaly to repair it and restore efficient parking. After the recovery plan is completed, the onboard processor can re-check whether the anomaly still exists. If so, it will repeat the repair steps; if not, it can regenerate the target parking path based on the vehicle's current location and generate a new parking plan based on the new target parking path for efficient parking.

[0089] It should be noted that, in one or more embodiments of this application, the specific recovery scheme adopted by the vehicle processor to repair the anomaly is not limited, such as sensor switching, software update, physical cleaning, circuit inspection, etc. It can be repaired manually or automatically by the vehicle, and this application does not impose any restrictions on this.

[0090] Based on the aforementioned parking technology's multiple redundancy and degradation testing strategies, vehicle parking efficiency is improved and successful. Even when the vehicle's parking system experiences partial failures, vehicle safety can still be ensured, and autonomous parking can be achieved to a certain extent.

[0091] Furthermore, with the development of computer technology, intelligent agents, due to their ability to transcend human rules and their capacity for continuous learning, are gradually being widely applied in various fields. To achieve intelligent parking, the executor of the aforementioned parking method can also be an onboard intelligent agent. By combining the intelligent agent with preset strategies, intelligent parking can be achieved with high efficiency and a high success rate. Specific implementation methods can be found in the above description and will not be repeated here.

[0092] Current intelligent parking systems are mostly based on servers within the parking environment. They generate and distribute parking paths for each vehicle based on an environmental model within the parking lot, representing centralized intelligent parking scheduling. This relies on the powerful computing capabilities of the servers within the parking environment to allocate parking paths that maximize group efficiency for vehicles. However, their anti-interference capabilities are relatively weak, and they are heavily dependent on the computing power of the parking environment's servers. Therefore, this application also discloses a parking method based on a dynamic semantic field. This method is applied to a parking lot server, which updates the dynamic semantic field and provides each vehicle with a dynamic semantic field for determining its target parking path.

[0093] like Figure 3 As shown, Figure 3 A schematic diagram of the process of a parking method based on a dynamic semantic field provided in this application embodiment includes: S300: In response to the acquisition instruction, determine the most recently updated dynamic semantic field, which is a model that describes the static environmental features of the target parking environment and the real-time traffic flow through semantic features.

[0094] S302: Return the most recently updated dynamic semantic field.

[0095] It should be noted that, in one or more embodiments of this application, the specific device executing the parking method is not limited; it can be a physical server of the target parking environment or a cloud server. It can be composed of a single server or a distributed set of servers; this application does not impose any restrictions on this.

[0096] When the server receives a retrieval command, it can determine the most recently updated dynamic semantic field. Then, it returns that most recently updated dynamic semantic field as the semantic field at the current moment.

[0097] Of course, in one or more embodiments of this application, there is no limitation on which device issues the acquisition instruction. It can be issued by the vehicle to be parked or by the user terminal; this application does not impose any restrictions on this. It can be set according to actual needs. Similarly, in one or more embodiments of this application, there is no limitation on which device the server returns the dynamic semantic field to. It can be returned to the device that sent the acquisition instruction or to the device specified by the acquisition instruction.

[0098] based on Figure 3 The parking method shown upgrades from static maps to dynamic situational awareness by constructing a digital model of the parking environment that integrates static and dynamic traffic information. Based on this dynamic semantic field, each vehicle can be regarded as an edge computing node, generating its own parking path in real time. High-frequency data is processed through distributed edge computing, effectively improving the real-time performance, stability, and scalability of parking decisions.

[0099] Furthermore, the construction of the dynamic semantic field relies solely on real-time perceived data of the target parking environment, and its accuracy is limited by the performance of the sensors in the target parking environment and the accuracy of the environmental recognition algorithm. This makes it difficult to improve the accuracy of the dynamic semantic field during subsequent updates, and it may even further decrease due to error accumulation. Therefore, the server can also receive environmental data and / or status data collected from each vehicle in the parking lot in real time. Then, the received environmental data and / or status data are converted into semantic features to update the dynamic semantic field.

[0100] It should be noted that, in one or more embodiments of this application, the specific frequency at which the server updates the dynamic semantic field is not limited. It can be set according to actual needs. For example, the dynamic semantic field can be updated immediately after receiving environmental data or vehicle status data, or the dynamic semantic field can be updated according to a preset period, or the dynamic semantic field can be updated when the amount of received data reaches a preset size. There are many possible methods, which will not be listed one by one in this application.

[0101] Furthermore, to ensure the quality of dynamic semantic field updates, the server can preprocess environmental data and / or vehicle status data during the update process, such as outlier filtering, coordinate system unification, and timestamp alignment. Then, the dynamic semantic field is updated based on the preprocessed environmental data and / or vehicle status data.

[0102] Furthermore, to mitigate the impact of fluctuations in the update quality of the dynamic semantic field on user parking quality, the server can pre-update the dynamic semantic field based on the semantic features corresponding to the received environmental data and / or vehicle status data before updating it. The pre-updated semantic field is then evaluated. If the evaluation result is satisfactory, the updated dynamic semantic field is determined. If the evaluation result is unsatisfactory, a re-acquisition command is returned, and environmental data and / or vehicle status data are reacquired to re-update the dynamic semantic field until the evaluation result of the pre-update result is satisfactory. This ensures the update quality of the dynamic semantic field.

[0103] It should be noted that, in one or more embodiments of this application, the specific method used by the server to evaluate the dynamic semantic field is not limited. For example, it can evaluate the dynamic semantic field based on data collected by sensors in the parking environment, or it can pre-train an evaluation model to assess the confidence level of the corresponding environmental data and traffic flow data of the target parking environment after conversion into a dynamic semantic field, and determine the evaluation result based on the confidence level. There are many possible solutions, which are not listed here; the specific solution can be set according to actual needs.

[0104] Then, since the acquired dynamic semantic field also includes traffic flow data, which generally contains the location information and vehicle information of other vehicles, the server can also set access permissions to protect user privacy and system security. That is, when receiving data sent by vehicles, returning dynamic semantic fields to vehicles, or when vehicles want to establish communication with the server, authentication can be performed first. When authentication is successful, data transmission or the establishment of communication relationship can then proceed.

[0105] In addition, to protect user privacy and data security, steps such as determining the parking path can also be performed by the server, with the onboard processor only used to issue parking commands and execute parking plans.

[0106] Specifically, such as Figure 4 As shown, Figure 4 This is a schematic diagram of a parking solution system provided in an embodiment of this application. The parking system includes at least two vehicles, M and N, to be parked, and a cloud server. It includes: S400: Vehicle M to be parked sends a parking path request to the cloud server; S402: The cloud server queries environmental data and traffic flow data based on dynamic semantic field and determines the target parking path based on preset strategies; S404: Based on historical path planning, the cloud server determines that the target parking path conflicts with the parking path of vehicle N, and the vehicle M to be parked meets the prior rules. Then, it generates a coordination request and sends it to the vehicle N to be parked. S406: Vehicle N waiting to park modifies its parking plan according to the coordination request and returns confirmation to give way; S408: The cloud server determines the target parking path and returns to the vehicle M to be parked; S410: The vehicle M to be parked executes the parking scheme corresponding to the target parking path, and returns a confirmation that parking is complete after execution; S412: The cloud server records the collaborative data and key parking metrics for this operation, and updates the preset and collaborative strategies.

[0107] Based on the parking method based on dynamic semantic fields provided in one or more embodiments of this application, this application also provides a corresponding parking device based on dynamic semantic fields, such as... Figure 5 As shown.

[0108] Figure 5 This application provides a schematic diagram of a parking device based on a dynamic semantic field. The device is applied to a receiving vehicle, which includes an onboard camera, specifically comprising: The acquisition module 500 is used to acquire the dynamic semantic field at the current moment in response to the parking instruction. The dynamic semantic field is a model that describes the static environmental features of the target parking environment and the real-time traffic flow through semantic features. Candidate module 501 is used to determine multiple candidate parking paths based on the dynamic semantic field; Selection module 502 is used to determine a target parking path from the plurality of candidate parking paths based on a preset strategy; The execution module 503 is used to generate a parking plan based on the target parking path, the parking plan being used to guide the vehicle into the parking space.

[0109] Optionally, the selection module 502 is specifically used to estimate the total cost required to complete each candidate parking path, wherein the total cost includes at least one of distance cost, time cost, probability of successful parking, and probability of conflict with other vehicles; and to determine a target parking path from the plurality of candidate parking paths based on the total cost corresponding to each candidate parking path.

[0110] Optionally, the device further includes a coordination module 504, specifically configured to send a coordination request to the server of the target parking environment when the target parking path conflicts with the parking paths of other vehicles; obtain the passage order returned by the server, wherein the passage order is determined by the server based on the coordination request and preset rules; adjust the parking plan based on the passage order, and execute the adjusted parking plan.

[0111] Optionally, the execution module 503 is specifically used to optimize the preset strategy based on the parking scheme and key indicators used to characterize the execution result of the parking scheme, wherein the key indicators include at least one of time consumption, number of conflicts, and success rate.

[0112] Optionally, the execution module 503 is specifically configured to update the reward function based on the key indicators; adjust the parameters of the preset strategy based on the updated reward function in a pre-built digital twin training environment; evaluate the performance of the adjusted preset strategy; when the evaluation result indicates performance improvement, update the preset strategy according to the adjusted parameters; when the evaluation result indicates no performance improvement, readjust the function of the preset strategy until the evaluation result indicates performance improvement.

[0113] Optionally, the device further includes a data acquisition module 505, specifically used to acquire environmental data and its own state data; and to send the environmental data and / or the state data to the server, wherein the environmental data and / or the state data are used to update the dynamic semantic field.

[0114] Optionally, the execution module 503 is specifically used to determine the cause of the parking anomaly when the parking anomaly occurs, wherein the cause of the anomaly includes at least one of sensor failure, communication interruption, path blockage, and vehicle system failure; based on the cause of the anomaly, activate the corresponding emergency plan, and determine a recovery plan according to the cause of the anomaly, wherein the recovery plan is used to repair the parking anomaly; re-detect whether there is an anomaly, and if so, re-determine the recovery plan; if not, re-determine a new target parking path, and regenerate and execute a new parking plan according to the new target parking path.

[0115] Based on the parking method based on dynamic semantic fields provided in one or more embodiments of this application, this application also provides a corresponding parking device based on dynamic semantic fields, such as... Figure 6 As shown.

[0116] Figure 6 This application provides a schematic diagram of a parking device based on a dynamic semantic field. The device is applied to a cloud server in a parking environment and specifically includes: The response module 600 is used to determine the most recently updated dynamic semantic field in response to the acquisition instruction. The dynamic semantic field is a model that describes the static environmental features of the target parking environment and the real-time traffic flow through semantic features. The sending module 601 is used to return the most recently updated dynamic semantic field.

[0117] Optionally, the device further includes an update module 602, specifically configured to receive environmental data and / or vehicle status data, perform data preprocessing on the environmental data and / or vehicle status data, wherein the data preprocessing includes at least one of outlier filtering, coordinate system one, and timestamp alignment; convert the preprocessed environmental data and / or vehicle status data into semantic features; and update the dynamic semantic field according to the semantic features.

[0118] Optionally, the update module 602 is specifically configured to: pre-update the dynamic semantic field according to the semantic features; evaluate the pre-update result of the dynamic semantic field according to a preset evaluation standard; determine the updated dynamic semantic field when the evaluation result of the pre-update result is qualified; and return a re-acquisition instruction when the evaluation result of the pre-update result is unqualified, and reacquire the environmental data and / or the vehicle status data to re-update the dynamic semantic field until the evaluation result of the pre-update result is qualified.

[0119] This application also provides a computer-readable storage medium storing a computer program that can be used to execute the above-described... Figure 1 or Figure 2A parking method based on dynamic semantic fields is provided.

[0120] This application also provides a vehicle including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it can be used to perform the aforementioned... Figure 1 or Figure 2 A parking method based on dynamic semantic fields is provided.

[0121] The various embodiments in this specification are described in a progressive manner. Similar or identical parts between embodiments can be referred to mutually. Each embodiment focuses on describing the differences from other embodiments. In particular, for apparatus or device embodiments, since they are basically similar to method embodiments, the description is relatively simple; relevant parts can be referred to the descriptions in the method embodiments. The professional and apparatus embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without creative effort.

[0122] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties. Furthermore, the collection, use and processing of the relevant data must comply with the relevant laws, regulations and standards of the relevant countries and regions, and corresponding operation entry points are provided for users to choose to authorize or refuse.

[0123] This application also provides an electronic device, such as Figure 7 As shown. The electronic device 700 provided in this application includes a memory 710, a processor 720, and an input / output interface 730. The memory 710, processor 720, and input / output interface 730 are connected via internal connection paths. The memory 710 stores instructions, and the processor 720 executes the instructions stored in the memory 710 to control the input / output interface 713 to receive input data and information, and output operation results and other data.

[0124] It should be understood that in the embodiments of this application, the processor 720 may be a general-purpose central processing unit (CPU), GPU, FPGA, microprocessor, application-specific integrated circuit (ASIC), or one or more integrated circuits to execute related programs in order to implement the technical solutions provided in the embodiments of this application.

[0125] The memory 710 may include a processor 720, a video display adapter 711, a disk drive 712, an input / output interface 713, a network interface 714, and the memory itself. The processor 720, video display adapter 711, disk drive 712, input / output interface 713, network interface 714, and memory 710 can communicate with each other via a communication bus 730.

[0126] The processor 720 can be implemented using a general-purpose CPU, microprocessor, application-specific integrated circuit (ASIC), or one or more integrated circuits to execute relevant programs in order to implement the technical solution provided in this application.

[0127] The memory 710 can be implemented in the form of ROM (Read Only Memory), RAM (Random Access Memory), static storage device, dynamic storage device, etc. The memory 710 can store the operating system 721 for controlling the operation of the electronic device 700, and the basic input / output system (BIOS) 722 for controlling the low-level operations of the electronic device 700. Additionally, it can store a web browser 723, a data storage management system 724, and a media file playback device 800, etc. The aforementioned media file playback device 800 can be the application program that specifically implements the aforementioned steps in this embodiment. In summary, when the technical solution provided in this application is implemented through software or firmware, the relevant program code is stored in the memory 710 and is called and executed by the processor 720.

[0128] Input / output interface 713 is used to connect input / output modules to realize information input and output. Input / output modules can be configured as components in the device (not shown in the figure) or externally connected to the device to provide corresponding functions. Input devices may include keyboards, mice, touch screens, microphones, various sensors, etc., and output devices may include displays, speakers, vibrators, indicator lights, etc.

[0129] Network interface 714 is used to connect a communication module (not shown in the figure) to enable communication between this device and other devices. The communication module can communicate via wired means (such as USB, Ethernet cable, etc.) or wireless means (such as mobile network, WIFI, Bluetooth, etc.).

[0130] Bus 730 includes a pathway for transmitting information between various components of the device, such as processor 720, video display adapter 711, disk drive 712, input / output interface 713, network interface 714, and memory 710.

[0131] It should be noted that although the above-described device only shows the processor 720, video display adapter 711, disk drive 712, input / output interface 713, network interface 714, memory 710, bus 730, etc., in specific implementations, the device may also include other components necessary for normal operation. Furthermore, those skilled in the art will understand that the above-described device may only include the components necessary for implementing the solution of this application, and does not necessarily include all the components shown in the figures.

[0132] All of the above-mentioned optional technical solutions can be combined in any way to form optional embodiments of this application, and will not be described in detail here.

[0133] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0134] Those skilled in the art will understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0135] In the several embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or units may be electrical, mechanical, or other forms.

[0136] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0137] In addition, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0138] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program verification codes, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0139] It should be noted that in the description of this application, the terms "first," "second," "third," etc., are used for descriptive purposes only and should not be construed as indicating or implying relative importance. Furthermore, in the description of this application, unless otherwise stated, "a plurality of" means two or more.

[0140] The above description is merely a preferred embodiment of this application and is not intended to limit this application. Any modifications or equivalent substitutions made within the spirit and principles of this application should be included within the protection scope of this application.

Claims

1. A parking method based on a dynamic semantic field, characterized in that, include: In response to a parking instruction, the dynamic semantic field at the current moment is obtained, wherein the dynamic semantic field is a model that describes the static environmental features of the target parking environment and the real-time traffic flow through semantic features. Based on the dynamic semantic field, multiple candidate parking paths are determined; Based on a preset strategy, a target parking path is determined from the multiple candidate parking paths; Based on the target parking path, a parking plan is generated, which guides the vehicle to park in the parking space.

2. The method as described in claim 1, characterized in that, The step of determining the target parking path from the multiple candidate parking paths based on a preset strategy specifically includes: For each of the candidate parking paths, the total cost required to complete the candidate parking path is estimated, wherein the total cost includes at least one of distance cost, time cost, probability of successful parking, and probability of conflict with other vehicles; The target parking path is determined from the plurality of candidate parking paths based on the total cost corresponding to each candidate parking path.

3. The method as described in claim 1, characterized in that, The method further includes: When the target parking path conflicts with the parking path of other vehicles, a coordination request is sent to the server of the target parking environment. Obtain the passage order returned by the server, wherein the passage order is determined by the server based on the coordination request and preset rules; Based on the traffic flow, the parking plan is adjusted and the adjusted parking plan is executed.

4. The method as described in claim 1, characterized in that, The method further includes: Based on the parking scheme and key indicators used to characterize the execution results of the parking scheme, the preset strategy is optimized, wherein the key indicators include at least one of time consumption, number of conflicts, and success rate.

5. The parking method as described in claim 4, characterized in that, The step of optimizing the preset strategy based on the parking scheme and key indicators used to characterize the execution result of the parking scheme specifically includes: Update the reward function based on the aforementioned key indicators; In a pre-built digital twin training environment, the parameters of the preset strategy are adjusted based on the updated reward function; The performance of the adjusted preset strategy is evaluated. When the evaluation result is a performance improvement, the preset strategy is updated according to the adjusted parameters. If the evaluation result indicates no performance improvement, the parameters of the preset strategy are readjusted until the evaluation result indicates a performance improvement.

6. The method as described in claim 1, characterized in that, The method further includes: Collect environmental data and its own status data; The environmental data and / or the state data are sent to the server of the target parking environment, wherein the environmental data and / or the state data are used to update the dynamic semantic field.

7. The method as described in claim 1, characterized in that, The method further includes: When parking is abnormal, determine the cause of the abnormality, wherein the cause of the abnormality includes at least one of sensor failure, communication interruption, path blockage, and vehicle system failure; Based on the cause of the anomaly, the corresponding emergency plan is activated, and a recovery plan is determined according to the cause of the anomaly. The recovery plan is used to repair the parking anomaly. Recheck for any anomalies; if so, determine a new recovery plan. If not, a new target parking path is determined, and a new parking plan is generated and executed based on the new target parking path.

8. A parking method based on a dynamic semantic field, characterized in that, The method is applied to a server in a parking environment, including: In response to the acquisition instruction, the most recently updated dynamic semantic field is determined, wherein the dynamic semantic field is a model that describes the static environmental features of the target parking environment and the real-time traffic flow through semantic features. Return the most recently updated dynamic semantic field.

9. The method as described in claim 8, characterized in that, The steps for updating the dynamic semantic field specifically include: Receive environmental data and / or vehicle status data, and perform data preprocessing on the environmental data and / or vehicle status data, wherein the data preprocessing includes at least one of outlier filtering, coordinate system I, and timestamp alignment; The preprocessed environmental data and / or vehicle status data are converted into semantic features; Update the dynamic semantic field based on the semantic features.

10. The method as described in claim 9, characterized in that, The step of updating the dynamic semantic field based on the semantic features specifically includes: Based on the semantic features, the dynamic semantic field is pre-updated; The pre-update results of the dynamic semantic field are evaluated according to the preset evaluation criteria. When the evaluation result of the pre-update result is qualified, the updated dynamic semantic field is determined; When the evaluation result of the pre-update result is unqualified, a re-acquisition instruction is returned, and the environmental data and / or the vehicle status data are reacquired to re-update the dynamic semantic field until the evaluation result of the pre-update result is qualified.

Images