Method for dispatching automatically parked vehicles in a parking lot, storage medium and electronic device

Abstract

The application discloses a kind of automatic parking vehicle scheduling method in parking lot, storage medium and electronic equipment, the method includes: using pre-trained global optimization model according to the dispatching request of vehicle to be dispatched and dispatching strategy suggestion generates global dispatching plan, and global dispatching plan is detected to generate dispatching instruction set;According to the resource reservation table of dispatching instruction set, according to resource reservation table, dynamic road right rule and the macroscopic path corresponding to the vehicle to be dispatched generates structured guidance information, and structured guidance information is sent to the vehicle to be dispatched, to guide the vehicle to be dispatched and park in target parking stall.The method ensures the safety of dispatching operation, avoids vehicle traffic deadlock, significantly improves the throughput of parking lot overall traffic, task response speed and site parking stall resource utilization.

Description

Technical Field

[0001] This invention relates to the field of parking scheduling technology, and in particular to an automatic parking vehicle scheduling method, storage medium, and electronic device in a parking lot. Background Technology

[0002] With the popularization of autonomous driving technology and the continuous growth of urban parking demand, the simultaneous operation of multiple Automated Valet Parking (AVP) vehicles in high-density parking lots has become a reality. In this environment, the real-time competition of vehicles for limited resources such as parking spaces and routes has led to potential driving conflicts, system deadlocks, and overall inefficiency. Traditional solutions either employ a completely centralized scheduling approach, which faces computational scalability bottlenecks and single-point failure risks; or rely on fully distributed autonomous negotiation among vehicles, which is prone to getting stuck in local deadlocks and makes it difficult to achieve global optimization.

[0003] Therefore, how to design a collaborative scheduling system that balances efficiency, safety, and scalability has become a key technical challenge that urgently needs to be overcome in the field of intelligent transportation. Summary of the Invention

[0004] This invention aims to at least partially solve one of the technical problems in related technologies. Therefore, one objective of this invention is to propose an automatic parking vehicle scheduling method in a parking lot, which, while ensuring safe scheduling operations and avoiding vehicle traffic deadlocks, significantly improves the overall throughput, task response speed, and parking space resource utilization rate of the parking lot.

[0005] A second objective of this invention is to provide a computer-readable storage medium.

[0006] The third objective of this invention is to provide an electronic device.

[0007] The fourth objective of this invention is to provide a method for parking a vehicle.

[0008] To achieve the above objectives, a first aspect of the present invention proposes an automatic parking vehicle scheduling method in a parking lot. The method includes: generating a global scheduling plan based on the scheduling request and scheduling strategy suggestion of the vehicle to be scheduled using a pre-trained global optimization model; performing conflict detection on the global scheduling plan to generate a scheduling instruction set; wherein the scheduling strategy suggestion is determined by the current resource reservation status of the parking lot, semantic scene description, and resource prediction results; the resource prediction results are determined by the current global dynamic map and historical traffic data of the parking lot; the global scheduling plan includes the target parking space and macro-path corresponding to the vehicle to be scheduled; generating a resource reservation table based on the scheduling instruction set; generating structured guidance information based on the resource reservation table, dynamic right-of-way rules, and the macro-path corresponding to the vehicle to be scheduled; and sending the structured guidance information to the vehicle to be scheduled to guide the vehicle to be scheduled to park in the target parking space; wherein the dynamic right-of-way rules are determined by the real-time traffic data in the parking lot and the global scheduling plan.

[0009] According to an embodiment of the present invention, when a scheduling request for a vehicle to be scheduled is received, a pre-trained global optimization model is guided by a scheduling strategy suggestion determined by the current resource reservation status, semantic scene description, and resource prediction results to allocate a target parking space and a macro-path for at least one vehicle to be scheduled, thereby obtaining a global scheduling plan. Conflict detection is performed on the global scheduling plan to generate a conflict-free scheduling instruction set containing a time dimension. Based on the scheduling instruction set, resources in the parking lot are reserved on a timeline basis. Structured guidance information is generated based on the resource reservation table, dynamic right-of-way rules, and the macro-path corresponding to the vehicle to be scheduled, and the structured guidance information is sent to the vehicle to be scheduled to guide it to park in the target parking space.

[0010] In addition, the automatic parking vehicle scheduling method in a parking lot proposed according to the above embodiments of the present invention may also have the following additional technical features: According to an embodiment of the present invention, before generating a global scheduling plan based on the scheduling request and scheduling strategy suggestion of the vehicle to be scheduled using a pre-trained global optimization model, the method includes: performing time and space alignment processing on the parking lot-side perception data stream and the vehicle-side perception data stream, and then converting them to the parking lot digital coordinate system to obtain parking lot-side converted data and vehicle-side converted data; performing association and fusion processing on the same physical entity based on the parking lot-side converted data and the vehicle-side converted data to obtain fused perception data; establishing the global dynamic map based on the fused data, and generating the semantic scene description based on the fused perception data using a pre-trained large language model.

[0011] According to an embodiment of the present invention, the method includes: calculating real-time traffic data of each lane in the parking lot based on the global dynamic map; predicting resources in the parking lot based on the real-time traffic data of each lane and the historical traffic data to obtain resource prediction results; and generating a scheduling strategy suggestion based on the current resource reservation status, the semantic scene description, and the resource prediction results, wherein the scheduling strategy suggestion includes at least one scheduling strategy and its corresponding global key performance indicator.

[0012] According to one embodiment of the present invention, the step of performing conflict detection on the global scheduling plan to generate a scheduling instruction set includes: performing conflict detection on the global scheduling plan in terms of time and space dimensions; if there is no conflict, generating the scheduling instruction set according to the global scheduling plan; if there is a conflict, coordinating the global scheduling plan using a preset adjudication mechanism, and generating the scheduling instruction set according to the coordinated global scheduling plan.

[0013] According to one embodiment of the present invention, generating a resource reservation table based on the scheduling instruction set includes: reserving resources in the parking lot on a timeline according to the scheduling instruction set, generating the resource reservation table, updating the resource reservation table based on the actual driving information and predicted driving information of the vehicles to be scheduled, and / or updating the resource reservation table based on a reservation update request.

[0014] According to one embodiment of the present invention, generating structured guidance information based on the resource reservation table, dynamic right-of-way rules, and the macro-path corresponding to the vehicle to be dispatched includes: generating dynamic right-of-way rules based on real-time traffic data in the parking lot and the global dispatch plan; allocating a passage corridor to the vehicle to be dispatched based on the dynamic right-of-way rules and the macro-path corresponding to the vehicle to be dispatched; and packaging the passage corridor, the dynamic right-of-way rules, and the resource reservation table to obtain the structured guidance information.

[0015] According to one embodiment of the present invention, the step of generating a global scheduling plan based on the scheduling request and scheduling strategy suggestion of the vehicle to be scheduled using a pre-trained global optimization model includes: performing intent recognition and intent extraction on the scheduling request to obtain structured demand features, and simultaneously querying the corresponding target vehicle capability model from a pre-established vehicle capability model library based on the vehicle type information of the vehicle to be scheduled; inputting the structured demand features, the target vehicle capability model and the scheduling strategy suggestion into the pre-trained global optimization model to obtain the global scheduling plan.

[0016] According to one embodiment of the present invention, the vehicle to be scheduled is parsed based on the structured guidance information to generate a fine path, and the vehicle to be scheduled is called from a pre-established behavior parking strategy library to park in the target parking space based on the fine path.

[0017] To achieve the above objectives, a second aspect of the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein when the computer program is executed by a processor, it implements the automatic parking vehicle scheduling method in a parking lot as proposed in the first aspect of the present invention.

[0018] To achieve the above objectives, a third aspect of the present invention provides an electronic device, including a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, it implements the automatic parking vehicle scheduling method in a parking lot as proposed in the first aspect of the present invention.

[0019] To achieve the above objectives, a fourth aspect of the present invention provides a vehicle parking method, the method comprising: generating a scheduling request and sending the scheduling request to an automatic parking vehicle scheduling system in a parking lot; and, in response to structured guidance information fed back by the automatic parking vehicle scheduling system in the parking lot based on the scheduling request, parking the vehicle in a target parking space, wherein the structured guidance information is generated by the automatic parking vehicle scheduling method in the parking lot as proposed in the first aspect of the present invention.

[0020] Additional aspects and advantages of the invention will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention. Attached Figure Description

[0021] Figure 1 This is a flowchart of an embodiment of the automatic parking vehicle scheduling method in a parking lot according to the present invention; Figure 2 This is a flowchart of a specific embodiment of the present invention for establishing a global dynamic map and generating a semantic scene description; Figure 3 This is a flowchart illustrating the generation of scheduling strategy suggestions according to a specific embodiment of the present invention; Figure 4 This is a flowchart illustrating the generation of structured guidance information according to a specific embodiment of the present invention; Figure 5 This is a flowchart illustrating the generation of a global scheduling plan according to a specific embodiment of the present invention; Figure 6 This is a schematic diagram of an automatic parking vehicle dispatching system in a parking lot according to an embodiment of the present invention; Figure 7 This is a structural block diagram of an electronic device according to an embodiment of the present invention; Figure 8 This is a flowchart of a vehicle parking method according to an embodiment of the present invention. Detailed Implementation

[0022] Embodiments of the present invention are described in detail below, examples of which are illustrated in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and intended to explain the present invention, and should not be construed as limiting the present invention.

[0023] Mainstream commercial smart parking lots employ an automated parking model combining "reservation navigation + vehicle-specific intelligence." This type of automated parking system typically provides a user application that allows drivers to pre-select and reserve an available parking space, then drive their vehicle to the handover area at the parking lot entrance. Once the vehicle activates the automated parking function, it relies on onboard sensors (such as cameras and radar) and pre-downloaded high-precision maps to independently plan its route and avoid obstacles until it reaches the assigned parking space. The parking lot primarily uses cameras and geomagnetic sensors to detect parking space occupancy and provides vehicle location guidance to users through electronic signs or an app.

[0024] However, the core limitation of this solution lies in its essence: it is essentially a "simple superposition of multiple independent intelligent agents" rather than a "cooperative system." Parking space allocation is based on a static, real-time "empty / occupied" state, failing to proactively coordinate the driving paths of multiple vehicles and the future parking space status, easily leading to vehicle concentration in popular areas and causing congestion. There is a lack of effective coordination between vehicles; their obstacle avoidance logic only reacts to static obstacles and sudden dynamic obstacles. When multiple vehicles are simultaneously operating at complex intersections or narrow road sections, they are highly susceptible to deadlocks due to rule conflicts or mutual refusal to yield, and the system lacks an efficient arbitration mechanism to quickly resolve such conflicts.

[0025] To address the aforementioned problems, embodiments of the present invention provide a method for scheduling automatic parking vehicles in a parking lot, a storage medium, and an electronic device. The method for scheduling automatic parking vehicles in a parking lot, a storage medium, and an electronic device according to embodiments of the present invention will be described in detail below with reference to the accompanying drawings and specific implementation details.

[0026] Figure 1 This is a flowchart of an embodiment of the automatic parking vehicle scheduling method in a parking lot according to the present invention. Figure 1 As shown, the automatic parking vehicle scheduling method in a parking lot may include: S101: Using a pre-trained global optimization model, a global scheduling plan is generated based on the scheduling requests and scheduling strategy suggestions of the vehicles to be scheduled. Conflict detection is then performed on the global scheduling plan to generate a scheduling instruction set. The scheduling strategy suggestions are determined by the current resource reservation status of the parking lot, semantic scene description, and resource prediction results. The resource prediction results are determined by the current global dynamic map of the parking lot and historical traffic data. The global scheduling plan includes the target parking space and macro-path corresponding to the vehicles to be scheduled.

[0027] The scheduling requests in this embodiment of the invention include parking requests and rescheduling requests.

[0028] For example, when the parking lot's site perception network detects that an automated parking vehicle has entered, it records the vehicle as a vehicle to be dispatched, and initiates the automated parking vehicle dispatching method in this embodiment of the invention based on the parking request sent by the vehicle to be dispatched, allocating a target parking space and a macro path for it, and guiding it to park in the target parking space.

[0029] For example, when a vehicle that has entered the parking area requests scheduling again, i.e., when a rescheduling request is triggered, the automatic parking vehicle scheduling method in this embodiment of the invention is started according to the rescheduling request, a target parking space and a macro path are allocated to it, and it is guided to park in the target parking space.

[0030] It should be noted that, in this embodiment of the invention, multiple vehicles to be dispatched can be simultaneously scheduled, target parking spaces and macro-paths can be assigned to them, and they can be guided to park in the target parking spaces.

[0031] Specifically, upon receiving a dispatch request from a vehicle, the pre-trained global optimization model performs decision analysis and generates a dispatch plan based on the dispatch request and suggested dispatch strategies. It should be noted that the suggested dispatch strategies are determined by the parking lot's current resource reservation status, semantic scene description, and resource prediction results. The resource prediction results are determined by the parking lot's current global dynamic map and historical traffic data. Therefore, when the pre-trained global optimization model receives a dispatch request, the suggested dispatch strategies can heuristically guide the model to output a global dispatch plan. Conflict detection is then performed on the global dispatch plan, and based on the conflict detection results, a conflict-free set of dispatch instructions, including a time dimension, is generated.

[0032] In practice, a multi-objective co-optimizer configured with a pre-trained global optimization model can be used to generate a global scheduling plan based on the scheduling requests and scheduling strategy suggestions of the vehicles to be scheduled. Specifically, the multi-objective co-optimizer is activated when a new parking request arrives or triggers a rescheduling. It transforms the parking lot model into a spatiotemporal resource network. The pre-trained global optimization model has the primary objective of minimizing the total estimated completion time of all vehicles, and the secondary objectives of maximizing parking space utilization and balancing the load of each area. Constraints include: matching vehicle physical size with parking spaces, requiring vehicles to be charged to charging spaces, user preferences (such as proximity to elevators) as soft constraints, and vehicle dynamics constraints (such as turning radius limiting feasible paths). The multi-objective co-optimizer, guided by scheduling strategy suggestions, outputs a global scheduling plan, assigning a target parking space to each vehicle to be scheduled, and planning a macroscopic path (represented by a series of key node sequences) from the current location to that parking space.

[0033] S102, a resource reservation table is generated based on the dispatch instruction set. Structured guidance information is generated based on the resource reservation table, dynamic right-of-way rules, and the macro-path corresponding to the vehicle to be dispatched. The structured guidance information is then sent to the vehicle to be dispatched to guide it to park in the target parking space. The dynamic right-of-way rules are determined by real-time traffic data in the parking lot and the global dispatch plan.

[0034] Specifically, to transform abstract scheduling instructions into concrete, exclusive resource occupancy commitments, this embodiment of the invention reserves resources within the parking lot on a timeline based on the scheduling instruction set, obtaining a resource reservation table. To transform macroscopic route plans and resource reservations into dynamic traffic rules and guidance suggestions that can be understood and executed by vehicles, this embodiment of the invention generates structured guidance information based on the resource reservation table, dynamic right-of-way rules, and the macroscopic route corresponding to the vehicle to be scheduled. The structured guidance information is then sent to the vehicle to be scheduled to guide it into the target parking space.

[0035] The automatic parking vehicle scheduling method in this embodiment of the invention, upon receiving a scheduling request from a vehicle to be scheduled, uses a scheduling strategy suggestion determined by the current resource reservation status, semantic scene description, and resource prediction results to guide a pre-trained global optimization model to allocate a target parking space and a macro-path for at least one vehicle to be scheduled, thereby obtaining a global scheduling plan. Conflict detection is performed on the global scheduling plan to generate a conflict-free scheduling instruction set containing a time dimension. Based on the scheduling instruction set, resources in the parking lot are reserved on a timeline basis. Structured guidance information is generated based on the resource reservation table, dynamic right-of-way rules, and the macro-path corresponding to the vehicle to be scheduled, and this structured guidance information is sent to the vehicle to be scheduled to guide it into the target parking space.

[0036] The automatic parking vehicle scheduling method in the parking lot embodiment of the present invention adopts a closed-loop decision-making process of prediction-optimization-reservation, and is supported by high-order reasoning and decision-making by a large model. It can resolve path conflicts and coordinate scheduling for multiple automatic parking vehicles in the parking lot. While ensuring the safety of scheduling operation and avoiding vehicle passage deadlock, it significantly improves the overall throughput, task response speed and parking space resource utilization of the parking lot.

[0037] This invention avoids vehicle access deadlock by preventing multiple automated parking vehicles from creating a circular dependency on limited access resources such as parking lot lanes, intersections, and parking spaces. It should be noted that a deadlock occurs when the access resources required by a vehicle in the parking lot are continuously occupied by other vehicles, preventing all vehicles from obtaining the necessary passage and thus preventing them from parking or exiting. This permanent stagnation of access within the parking lot constitutes a deadlock.

[0038] This invention aims to achieve automated parking vehicle scheduling by performing full-element perception and collaborative positioning of the parking lot. It can practically and accurately acquire unified status information of all dynamic and static elements within the parking lot in real time, providing a reliable data foundation for the entire automated parking vehicle scheduling system.

[0039] In one embodiment of the present invention, such as Figure 2 As shown, before generating a global scheduling plan based on the scheduling requests and scheduling strategy suggestions of the vehicles to be scheduled using a pre-trained global optimization model, the automatic parking vehicle scheduling method in a parking lot may include: S201: After performing time and space alignment processing on the field-end perception data stream and the vehicle-end perception data stream in the parking lot, the data stream is converted into the parking lot digital coordinate system to obtain field-end conversion data and vehicle-end conversion data.

[0040] Implementably, a site-side sensing network can be deployed within the parking lot to acquire site-side sensing data streams. For example, multiple types of sensor nodes can be systematically deployed at locations such as parking lot pillars and rooftops to create overlapping coverage.

[0041] The deployed sensor nodes include, but are not limited to: visual perception nodes: using wide-angle cameras to identify vehicle outlines, license plates, and directions of movement through machine vision algorithms; radar perception nodes: using millimeter-wave radar or lidar to accurately detect the position and speed of objects, unaffected by lighting conditions; and area status sensors: embedding geomagnetic or ultrasonic parking space detectors within parking spaces to directly detect the occupancy status of parking spaces. All nodes are connected to the edge computing unit via wired and / or wireless networks.

[0042] In a feasible manner, roadside communication units (RSUs) deployed in parking lots can establish communication connections with vehicles entering the parking lot (vehicles with V2X (Vehicle to Everything) communication capabilities) to obtain vehicle-side perception data streams from each vehicle in the parking lot. Specifically, after establishing a connection with connected vehicles with V2X communication capabilities through the roadside communication units (RSUs) deployed in the parking lot, the system requests the vehicles to report their own status information (including their own high-precision positioning, speed, and heading angle obtained based on the fusion of onboard sensors) and the local perception results of the surrounding environment obtained by the onboard sensors (such as a list of surrounding obstacles) according to a preset protocol.

[0043] It should be noted that for vehicles that are not connected to the network, their status can be independently detected and tracked by the field-side perception network, and a temporary ID can be assigned to ensure full coverage without blind spots.

[0044] Specifically, the system receives the parking lot-side sensing data stream and the vehicle-side sensing data stream, and performs time and space alignment processing on the parking lot-side sensing data stream and the vehicle-side sensing data stream of each vehicle. Since the parking lot-side sensing data stream and the vehicle-side sensing data stream are located in different coordinate systems, this embodiment of the invention unifies the aligned parking lot-side sensing data stream and the vehicle-side sensing data stream into the high-precision digital map coordinate system of the parking lot, obtaining parking lot-side converted data and vehicle-side converted data.

[0045] S202, Based on the field-end conversion data and vehicle-end conversion data, the same physical entity is associated and fused to obtain fused perception data.

[0046] In practice, target association algorithms (such as tracking based on the Hungarian algorithm and Kalman filtering) can be used to associate and fuse perception data from different sources targeting the same physical entity (such as a vehicle) to generate the optimal state estimate of the same physical entity (fused perception data).

[0047] S203: A global dynamic map is built based on the fused data, and a semantic scene description is generated based on the fused perception data using a pre-trained large language model.

[0048] Specifically, a global dynamic map can be built based on the fused data, comprising a static layer, a dynamic layer, and an event layer. The static layer can include lane lines, parking lines, curbs, speed bumps, charging station locations, etc.; the dynamic layer can include the unique ID, precise outline, location, speed, acceleration, and heading of all vehicles and pedestrians; and the event layer can include congested areas, slow-moving areas, and temporary obstacles (such as fallen objects).

[0049] In this embodiment of the invention, the global dynamic map is updated in real time using the above-described method based on the parking lot-side perception data stream and the vehicle-side perception data stream.

[0050] To better enable the pre-trained global optimization model to allocate target parking spaces and macroscopic paths for vehicles to be scheduled, this embodiment of the invention introduces a pre-trained large language model, such as a large language model (LLM) fine-tuned for parking scenarios or a visual language model (VLM), as a "perceptual semantic understander".

[0051] It should be noted that traditional perception systems output data such as coordinates and speed. However, this invention, through the introduction of a pre-trained large language model, analyzes and fuses the perception data stream to generate semantic scene descriptions. For example, fused perception data, such as "object A at coordinates (X,Y) with speed V," is semantically understood and output as: "A shopping cart has been left in the center of the main passage in area B, posing a moderate level of obstruction to traffic" or "A potential four-car collision is forming at the intersection in area C." These semantic scene descriptions provide high-level semantic information, offering richer and more interpretable context for subsequent decisions regarding target parking spaces and macroscopic paths, thus solving the problem of simple rule-based systems struggling to globally schedule contingency plans.

[0052] In one embodiment of the present invention, such as Figure 3 As shown, the automatic parking vehicle dispatching method in a parking lot may also include: S301 calculates real-time traffic data for each lane in the parking lot based on the global dynamic map, and predicts the resources in the parking lot based on the real-time traffic data and historical traffic data of each lane, thus obtaining the resource prediction results.

[0053] Specifically, the system analyzes the global dynamic map in real time to calculate real-time traffic data for each lane within the parking lot. This real-time traffic data includes at least vehicle density, average speed, and queue lengths at key nodes (such as intersections). Based on historical and real-time traffic data, time series analysis or lightweight machine learning models are used to predict the resource (parking spaces, lanes) situation within the parking lot for the next 1-5 minutes, yielding resource prediction results. These predictions may include: the occupancy / release probability of each parking space; the estimated travel time for each main path; and the probability of potential conflict points occurring within the next 1-5 minutes.

[0054] S302, Generate scheduling strategy recommendations based on the current resource reservation status, semantic scenario description and resource prediction results. The scheduling strategy recommendations include at least one scheduling strategy and its corresponding global key performance indicators.

[0055] Specifically, a large language model enhanced with parking scenario capabilities can be used to gain a deeper understanding of parking lot operation rules and traffic flow theory. The enhanced large language model, based on the current resource reservation status, semantic scenario description, and resource prediction results, explains complex situations (e.g., determining whether the root cause of congestion is "a car looking for a parking space" or "multiple vehicles having path conflicts"), provides qualitative analysis, and generates high-level scheduling strategy suggestions.

[0056] It should be noted that, in this embodiment of the invention, the enhanced large language model does not directly output specific scheduling instructions, but instead generates multiple selectable strategy descriptions, such as "Scheduling strategy A: Activate tidal flow lanes in zone D to prioritize the evacuation of outbound traffic; Scheduling strategy B: Temporarily restrict inbound traffic in zone E and guide it to zone F." The impact of each scheduling strategy is evaluated: Utilizing its powerful reasoning capabilities, the model quickly predicts the potential impact of different scheduling strategies on global KPIs (Key Performance Indicators), such as average stopping time and congestion point elimination speed, providing high-quality decision input and explanation for the multi-objective collaborative optimizer.

[0057] In one embodiment of the present invention, conflict detection is performed on the global scheduling plan to generate a scheduling instruction set, including: Perform time and space-dimensional conflict detection on the global scheduling plan; If there is no conflict, a scheduling instruction set is generated based on the global scheduling plan; If a conflict exists, the global scheduling plan will be coordinated using a pre-defined conflict resolution mechanism, and a scheduling instruction set will be generated based on the coordinated global scheduling plan.

[0058] Specifically, the scheduling plans for multiple vehicles generated by the multi-objective collaborative optimizer are subjected to spatiotemporal conflict detection to check whether there are situations where two vehicles are expected to occupy the same spatial resources (such as intersections or narrow roads) at the same time.

[0059] If no conflict is detected, it is determined that there is no conflict. At this time, a set of conflict-free scheduling instructions containing the time dimension can be directly output according to the global scheduling plan.

[0060] If a conflict is detected, it is determined that a conflict exists. In this embodiment of the invention, when it is determined that two vehicles are expected to simultaneously occupy the same spatial resource (such as an intersection or narrow road), instead of re-optimizing the global system, a rapid arbitration mechanism (preset adjudication mechanism) based on preset rules is initiated. For example, a "spatiotemporal reservation adjustment" rule is used to make a slight delay adjustment to the path time window of the later-arriving vehicle. Or, during resource contention, a ruling is made based on preset adjudication mechanisms such as "task priority" (e.g., priority for emergency vehicles) and "maximizing the overall system benefit." Finally, a conflict-free scheduling instruction set containing a time dimension is output.

[0061] In one embodiment of the present invention, generating a resource reservation table based on a scheduling instruction set may include: Based on the dispatch instruction set, the resources in the parking lot are reserved on a timeline basis, and a resource reservation table is generated. At the same time, the resource reservation table is updated based on the actual driving information and predicted driving information of the vehicles to be dispatched, and / or the resource reservation table is updated based on the reservation update request.

[0062] This invention abstracts key parking resources into "spatiotemporal units" that can be exclusively reserved. For example, each parking space is modeled as a reservable time block; each potentially conflict-prone path point (such as the center area of ​​an intersection or a merging point) is modeled as a "space-time" cube, meaning that within a specific time window, this spatial area can only be occupied by one vehicle. Upon receiving a scheduling instruction set, reservations are immediately made for the vehicles to be scheduled on the resource timeline. For example, "parking space B203, from time T1 to time T2" ​​is reserved for vehicle A; "the spatiotemporal unit through intersection C, from time T1+30 to T1+35" is also reserved for vehicle A. All reservations are recorded in a resource reservation table.

[0063] It should be noted that the resource reservation in this embodiment of the invention is not rigid. The resource reservation table is dynamically updated based on the actual driving information and predicted driving information of the vehicles to be dispatched or user needs (receiving a reservation update request sent by the user).

[0064] For example, when there is a discrepancy between the actual driving information and the predicted driving information of a vehicle to be dispatched, the vehicle to be dispatched or the roadside unit can initiate a "reservation update request" (e.g., requesting a 5-second delay in the time window for passing through a certain point). The impact of this update on subsequent vehicles and the overall system is assessed. In this process, a large model fine-tuned for the parking scenario can be used as a "negotiation assistant." For complex conflict coordination (such as a chain conflict of reservations involving multiple vehicles), the impact of different adjustment schemes can be quickly simulated, and coordination suggestions can be generated using natural language, for example: "Approve vehicle A's delay request, but suggest that vehicle B pass through 3 seconds earlier. This will have the least impact on B's energy consumption, and the overall delay will only increase by 2 seconds." A "flexible reservation" mechanism is adopted to balance the rigor of the plan with the fault tolerance of its execution, thereby improving the intelligence level of the negotiation.

[0065] In one embodiment of the present invention, the automatic parking vehicle scheduling method in a parking lot may further include: periodically publishing key information of the resource reservation table (especially the space area to be occupied and its time window) to all networked vehicles via V2X or roadside broadcast, so that each vehicle is aware of the future occupation plan of public resources and provides "common knowledge" for its autonomous decision-making.

[0066] In one embodiment of the present invention, such as Figure 4As shown, structured guidance information is generated based on the resource reservation table, dynamic right-of-way rules, and the macro-path corresponding to the vehicles to be dispatched. This information may include: S401 generates dynamic right-of-way rules based on real-time traffic data in the parking lot and the overall dispatch plan; S402 allocates passage corridors to vehicles to be dispatched based on dynamic right-of-way rules and the macro-paths corresponding to the vehicles to be dispatched. S403 packages the passage corridor, dynamic right-of-way rules, and resource reservation forms to obtain structured guidance information.

[0067] Specifically, a large language model, fine-tuned for parking scenarios, is introduced to automatically generate and publish dynamic right-of-way rules based on real-time traffic data within the parking lot (e.g., excessive traffic flow in a certain lane) and global scheduling plans, thereby enhancing the rationality and interpretability of the generated dynamic right-of-way rules. Implementably, real-time traffic data is converted into semantic traffic situation data. This semantic traffic situation data, along with global scheduling plans (e.g., "Eastbound traffic is three times that of westbound traffic, and this is expected to last for 5 minutes"), is input into the fine-tuned large language model. The large language model can then generate rule suggestions that better align with human management intuition, such as: "It is recommended that lane 3 be temporarily converted to one-way eastbound traffic within the next 5 minutes. Reason: Significant tidal flow, which can increase overall throughput by 15%." Administrators can review and confirm these suggestions, or the system can automatically adopt highly credible suggestions as dynamic right-of-way rules.

[0068] For each vehicle awaiting dispatch, its macroscopic path is combined with the current dynamic right-of-way rules to generate a clear traffic corridor. It's important to note that a traffic corridor is not a precise trajectory, but rather a description of a traffic area that is conflict-free in both space and time. For example, the description of a traffic corridor could be: "Please travel east along the main road (currently one-way east), and at the intersection marked X, turn left into lane B according to the right-of-way rules (you have priority). Your target parking space is B203, and the reservation period is 10:30-12:00."

[0069] The passageway, dynamic right-of-way rules, and resource reservation forms are packaged into structured guidance information and sent to the target vehicles awaiting dispatch via a low-latency communication link. For non-connected vehicles, guidance is provided through enhanced parking guidance screens (displaying dynamic directional arrows and area status).

[0070] In one embodiment of the present invention, such as Figure 5 As shown, a pre-trained global optimization model generates a global scheduling plan based on the scheduling requests and scheduling strategy suggestions of the vehicles to be scheduled. This plan may include: S501, perform intent recognition and intent extraction on the scheduling request to obtain structured demand features, and at the same time query the corresponding target vehicle capability model from the pre-established vehicle capability model library based on the vehicle type information of the vehicle to be scheduled. S502 inputs the structured demand characteristics, target vehicle capability model and scheduling strategy suggestions into the pre-trained global optimization model to obtain the global scheduling plan.

[0071] This invention utilizes a large language model fine-tuned for parking lot scenarios to perform deep demand understanding and instruction compilation on user-sent scheduling requests, generating structured demand features. Specifically, users express their scheduling needs through natural language, voice, or mixed input (e.g., "Find me a charging station near the elevator; I'll be parking for about 2 hours and need to leave after shopping"). The user's natural language input is fed into the large language model fine-tuned for parking lot scenarios. The large language model fine-tuned for parking lot scenarios identifies the user's core intent, determining whether it is "parking," "retrieving the car," "charging," or "temporary parking," and extracts structured preferences and constraints. For example, multi-dimensional, quantifiable parameters are extracted. It should be noted that in this invention, the large language model fine-tuned for parking lot scenarios infers implicit needs: it can infer that "needing to leave after shopping" may mean "wanting to park near the mall exit to reduce walking." For example, from the above scheduling request (find me a charging station near the elevator; I'll be parking for about 2 hours and need to leave after shopping), we can extract: {Core: charging parking; Hard constraints: charging parking space, estimated duration ~2 hours; Soft preferences: near elevator (high priority), near mall exit (medium priority)}. The large language model, fine-tuned for the parking lot scenario, can understand vague concepts such as "near" and "convenient" and quantify them into parameters that the system can understand (e.g., "walking distance to elevator < 50 meters").

[0072] This invention maintains a mapping library (vehicle capability model library) between vehicle models and capability parameters in the background of the automatic parking vehicle dispatching system in parking lots. When a vehicle model corresponding to a vehicle to be dispatched enters the system for the first time, it registers its default capability model through its VIN code or vehicle model information, including: minimum turning radius, overall dimensions, wheelbase, whether it has automatic parking function and parking type (such as perpendicular, parallel, diagonal), maximum range (affecting the urgency of charging needs), etc. Vehicle owners or operators can also customize certain parameters (such as driving style preference: conservative / agile). This invention retrieves the corresponding target vehicle capability model from the pre-established vehicle capability model library based on the vehicle model information of the vehicle to be dispatched.

[0073] The structured demand characteristics, target vehicle capability model, and scheduling strategy suggestions are input into a pre-trained global optimization model. The pre-trained global optimization model combines the list of qualified parking spaces retrieved from the parking lot's resource database to generate a highly customized, goal-oriented global scheduling plan.

[0074] The global scheduling plan generates scheduling instructions through conflict detection. For example, a generated scheduling instruction could be: "Assign charging space C-05 to a Tesla Model Y (equipped with Autopark). This space is 45 meters from elevator number 3 and 60 meters from the mall's east exit. Execute the 'Efficient Autopark and Charge' task, expected to be occupied from 14:00 to 16:00. Please note that there is a pillar on the left side of this space; it is recommended to park slightly to the right to facilitate door opening." This instruction is semantically rich and directly guides the vehicle's behavior selection.

[0075] In one embodiment of the present invention, vehicles to be scheduled based on structured guidance information can be parsed to generate fine paths. Based on the fine paths, vehicles to be scheduled can call corresponding parking behavior strategies from a pre-established behavior parking strategy library to park in the target parking space.

[0076] In this embodiment of the invention, the structured guidance information can be parsed and processed by the vehicle to be scheduled to generate a fine path. The generated fine path can be used by the vehicle to be scheduled to call the corresponding parking behavior strategy from a pre-established behavior parking strategy library to achieve parking in the target parking space.

[0077] Specifically, after receiving structured guidance information, the vehicle to be dispatched performs task parsing and local path replanning to generate a refined path. In a feasible manner, after receiving the structured guidance information, the vehicle's onboard system first parses the traffic corridor and the final parking task. Based on the real-time perception of the surrounding environment by the onboard sensors (integrating remote information from a global dynamic map and near information from its own sensors), it performs refined local path planning and trajectory generation within the traffic corridor constraints. For example, within the corridor, specific lane-changing and following trajectories are planned based on the actual obstacle locations.

[0078] In this embodiment of the invention, the vehicle to be dispatched locally stores a verified "parking behavior strategy library". Upon receiving a high-level task instruction (structured guidance information), the task planner of the vehicle to be dispatched parses the semantics of the structured guidance information and calls and combines corresponding strategies from the parking behavior strategy library to execute. For example, it calls the "automatic parking strategy for pillared parking spaces" and fine-tunes the parameters according to the "slightly veer to the right" instruction, so that vehicles with different capabilities can complete complex personalized tasks in the most suitable way for themselves.

[0079] It should be noted that during the parking process, the vehicle-side collaborative control and local collision avoidance module of the vehicle to be dispatched can be used to assist in the control of the vehicle, ensuring that the vehicle can safely and smoothly complete the final driving and parking actions after receiving structured guidance information, and achieve local coordination in case of emergencies.

[0080] In the embodiments of the present invention, during the parking process, rule-based distributed negotiation and a lightweight vehicle / roadside large language model are used for auxiliary decision-making.

[0081] Specifically, when vehicles awaiting dispatch meet in a shared area without a clearly defined reservation (such as a wide two-way lane) and a conflict may occur, the vehicles or the vehicles and roadside units make autonomous decisions based on a pre-defined distributed negotiation protocol. The protocol uses explicit and computable rules, such as: reservation-based yielding: comparing the reservation times of each vehicle's next critical point, with the more urgent one taking priority; imitation of intersection traffic rules: at intersections without a center signal, vehicles approaching from the right are given priority by default, or vehicles proceed according to temporary "stop / yield" signs issued by the roadside units; and negotiation based on safety clearances: exchanging intentions through V2V communication to negotiate who should slow down and yield first to create a safe passage gap.

[0082] In complex or deadlocked situations, participating units (vehicles or RSUs) can invoke a lightweight onboard / roadside big language model to quickly reason about the intentions, priorities, and overall traffic efficiency of multiple parties, and propose a better yielding suggestion (such as "Although vehicle A's reservation is more urgent, vehicle B has children and is in manual mode, so it is recommended that vehicle A yield, and the system will compensate vehicle A for the time"), to help all parties break the deadlock and make more efficient decisions.

[0083] In this embodiment of the invention, during the parking scheduling process, safety monitoring and anomaly reporting are performed on the vehicles to be scheduled. Specifically, the vehicles to be scheduled continuously monitor their own status and the degree of matching with the environment during their journey. If a vehicle is found to have deviated significantly from the predetermined plan (such as being unable to enter the reserved corridor for an extended period of time), detects static obstacles not marked on the map, or encounters abnormal behavior from other vehicles, it immediately reports the abnormal event and its own status to the collaborative decision-making and resource scheduling center to request new instructions or trigger a system-level rescheduling.

[0084] To enable the self-evaluation and continuous improvement of the automatic parking vehicle dispatching system in the parking lot and form a management closed loop, this embodiment of the invention sets up a system performance monitoring and adaptive optimization module.

[0085] Implementably, establish a multi-dimensional performance indicator (KPI) system. Specifically, define and calculate a series of KPIs in real time, including: efficiency indicators: average parking time (from request to completion), average parking search time, parking space turnover rate, system throughput (number of vehicles served per unit time); safety and reliability indicators: number of conflicts, success rate of automatic conflict resolution, task failure rate (e.g., failure to park in a designated parking space), system anomaly response time; resource and energy consumption indicators: channel utilization distribution, average detour distance, average vehicle energy consumption; service quality indicators: user preference satisfaction rate, timeliness of charging demand fulfillment.

[0086] This invention utilizes a large-scale model for in-depth analysis and root cause diagnosis. Specifically, it continuously collects all KPI data, event logs, and vehicle trajectories. For example, when it detects a 20% increase in average travel time in Zone D during evening rush hour, the large-scale model can correlate and analyze perception logs and scheduling decision records, outputting the analysis conclusion: "The primary reason is that traffic flow at the entrance of Zone D and the exit of Zone E intersects between 18:00 and 18:30, causing a delay in the response of existing right-of-way rules. A secondary reason is that three large SUVs entered the parking area during this period, exacerbating congestion." Based on the diagnostic conclusion of the large-scale model, an optimization process is initiated. The large-scale model can further propose specific and understandable optimization suggestions, such as: "Suggestion 1: During 17:45-18:45, set the DE connecting channel as a tidal flow lane, prioritizing entry in the first 30 minutes and exiting in the last 30 minutes. This is expected to increase throughput by 12%. Suggestion 2: Optimize the large vehicle allocation algorithm, guiding them to the more spacious Zone F during peak hours." Simulation verification is then performed based on the intelligent strategy optimization.

[0087] This invention provides an automatic parking vehicle dispatching system in a parking lot.

[0088] like Figure 6 As shown, the automatic parking vehicle dispatching system in the parking lot may include a full-element perception and collaborative positioning layer, a collaborative decision-making and resource dispatching center, a spatiotemporal resource reservation and allocation management module, a dynamic right-of-way management and route guidance module, a vehicle-side collaborative control and local collision module, a personalized task adaptation and flexible execution module, and a system performance monitoring and adaptive optimization module.

[0089] In one specific embodiment of the present invention, the all-element perception and collaborative positioning layer can acquire field-end perception data streams and vehicle-end perception data streams, and fuse multi-source data (field-end perception data streams and vehicle-end perception data streams). Implementably, a field-end perception network can be deployed within the parking lot to acquire the parking lot's field-end perception data streams. Communication connections can be established between roadside communication units (RSUs) deployed in the parking lot and vehicles entering the parking lot (possessing V2X (Vehicle to Everything) communication capabilities) to acquire the vehicle-end perception data streams of each vehicle within the parking lot. A central fusion processor is used to fuse multi-source data (field-end perception data streams and vehicle-end perception data streams) and construct a global dynamic map.

[0090] In practice, a collaborative decision-making and resource scheduling center can be run on a cloud or parking lot edge server. In this embodiment of the invention, the collaborative decision-making and resource scheduling center is the intelligent hub of the automated parking vehicle scheduling system, responsible for predicting and optimizing scheduling plans based on the global situation and generating scheduling instruction sets.

[0091] In one specific embodiment of the present invention, the collaborative decision-making and resource scheduling center may include a traffic situation analysis and prediction unit, an enhanced large model unit, a multi-objective system optimizer, and a conflict detection and arbitration unit. The traffic situation analysis and prediction unit calculates real-time traffic data for each lane within the parking lot based on a global dynamic map, and predicts the resources within the parking lot based on the real-time traffic data and historical traffic data for each lane. The enhanced large model unit receives a semantic scene description, resource prediction results, and the current resource reservation status, and performs decision analysis and strategy suggestion generation based on the enhanced large language model to obtain scheduling strategy suggestions. The multi-objective collaborative optimizer allocates target parking spaces and macro-paths to the vehicles to be scheduled based on their scheduling requests and scheduling strategy suggestions. The conflict detection and arbitration unit performs spatiotemporal conflict detection on the scheduling plans for multiple vehicles generated by the multi-objective collaborative optimizer, checking for situations where two vehicles are expected to simultaneously occupy the same spatial resource (such as an intersection or narrow passage), and outputs a conflict-free scheduling instruction set containing a time dimension based on the conflict detection results.

[0092] In practice, the time and space resource reservation and allocation management module is responsible for transforming abstract scheduling instructions into specific, exclusive resource occupancy commitments, thereby achieving refined management of parking lot time and space resources.

[0093] In one specific embodiment of the present invention, the spatiotemporal resource reservation and allocation management module may include a resource modeling unit, a reservation execution unit, and a dynamic reservation management unit. The resource modeling unit abstracts the key resources of the parking lot into "spatiotemporal units" that can be exclusively reserved. Upon receiving a set of scheduling instructions, the reservation execution unit immediately makes reservations for the vehicles to be scheduled on the resource timeline. The dynamic reservation management unit updates the resource reservation table based on the actual and predicted driving information of the vehicles to be scheduled, and / or updates the resource reservation table based on reservation update requests.

[0094] In practice, the dynamic right-of-way management and route guidance module is responsible for transforming macro-level route plans and resource reservations into dynamic traffic rules and guidance suggestions that can be understood and executed by vehicles.

[0095] In one specific embodiment of the present invention, the dynamic right-of-way management and route guidance module may include a dynamic right-of-way rule generation unit, a conflict-free passage corridor unit, and a guidance information unit. The dynamic right-of-way rule generation unit automatically generates and publishes dynamic right-of-way rules based on real-time traffic data (such as excessive traffic flow in a certain lane) and the global scheduling plan. The conflict-free passage corridor unit allocates passage corridors to vehicles to be scheduled based on the dynamic right-of-way rules and the macro-paths corresponding to the vehicles to be scheduled. The guidance information unit packages the passage corridors, dynamic right-of-way rules, and resource reservation table, and outputs structured guidance information.

[0096] In practice, the vehicle-side collaborative control and local collision module perform safety monitoring and anomaly reporting, as well as local path replanning, to ensure that the vehicle can safely and smoothly complete the final driving and parking actions after receiving guidance from the upper layer, and achieve local coordination in case of emergencies.

[0097] In practice, the personalized task adaptation and flexible execution module performs intent recognition and extraction on scheduling requests and queries the target vehicle capability model. Based on structured demand characteristics, target vehicle capability models, and scheduling strategy suggestions, it generates a global scheduling plan, enabling the automatic parking vehicle scheduling system in the parking lot to handle diverse vehicle capabilities and user needs, ensuring the feasibility of scheduling instructions and user satisfaction.

[0098] In practice, the system performance monitoring and adaptive optimization module is used to enable the system to self-evaluate and continuously improve, forming a management closed loop.

[0099] This invention employs a hybrid control architecture that integrates centralized path management and distributed negotiation. A collaborative decision-making and resource scheduling center (located in the cloud or at the site) dynamically divides and publishes time-sensitive "right-of-way windows" or one-way guidance paths on a digital map based on the overall traffic situation. This manages traffic flow on main roads and key nodes, preventing systemic deadlocks. In non-critical areas, vehicles autonomously negotiate and avoid obstacles based on V2X communication and preset rules (such as spatiotemporal resource reservation protocols). This architecture ensures overall traffic order while distributing real-time computing load and reducing communication latency dependencies.

[0100] This invention employs a global allocation algorithm based on predictive parking space status and collaborative optimization. Instead of relying on the instantaneous state of parking spaces, this invention combines real-time and historical traffic data from parking lot sensing devices to predict parking space occupancy and release in future time periods. During allocation, discrete resources such as parking spaces and charging stations, along with differentiated vehicle needs (e.g., vehicle type, charging duration), are input into a global optimization model for calculation. When multiple vehicles conflict, a lightweight game theory method based on priority or utility is used for coordination, optimizing traffic flow distribution from the source and improving parking space utilization and system throughput.

[0101] This invention supports a task description and adaptive execution framework that addresses vehicle heterogeneity, enhancing the system's adaptability to diverse vehicles and tasks. The system establishes digital profiles for each vehicle, describing its motion performance, dimensions, and other key capability parameters. The scheduling instructions generated at the parking lot are not specific action sequences, but rather high-level task descriptions independent of the vehicle's specific implementation. The vehicle-side system selects and executes appropriate strategies from a pre-defined behavior strategy library based on its own capabilities and real-time environmental perception results to complete the task, achieving precise "task-capability" matching. This enables the system to uniformly schedule vehicles with different performance levels to complete diverse tasks, improving overall reliability and resilience.

[0102] Traditional scheduling methods primarily allocate parking spaces based on the current availability, a passive and short-sighted approach that easily leads to traffic congestion and area bottlenecks. This invention, by predicting future parking space availability and performing global collaborative planning, proactively optimizes parking resource utilization from both temporal and spatial perspectives, thereby improving overall parking lot throughput and parking space turnover rate from the source. In terms of resource allocation, it elevates the process from static, instantaneous allocation to dynamic, global optimization.

[0103] Existing scheduling methods have significant shortcomings in path coordination: distributed avoidance relying entirely on simple traffic rules is prone to deadlock; while centralized detailed path planning incurs excessive computational load. This invention addresses this by dynamically defining lightweight "right-of-way corridors" in a central system to ensure traffic order on main roads and eliminate system-level deadlocks, while allowing vehicles to autonomously and efficiently negotiate in non-critical areas. This architecture significantly reduces the real-time computational burden and communication latency requirements of the central system while ensuring overall traffic flow. In path coordination, it upgrades from rule-based avoidance or centralized planning to a deterministic hybrid control approach of "centralized corridors + distributed negotiation."

[0104] Existing scheduling methods typically treat vehicles as homogeneous units, making it difficult to effectively handle differences in vehicle performance and diverse, nuanced user needs (such as charging and preferred parking space types). This invention establishes vehicle "capability profiles" and a "task-capability" matching model, enabling the system to issue advanced tasks and allow vehicles to autonomously select the optimal strategy for execution based on their own circumstances. This not only achieves unified and efficient scheduling of heterogeneous vehicles but also makes it possible to meet personalized needs such as "near elevators" and "independent parking spaces," significantly improving the system's usability and user experience. In terms of task adaptability, it evolves from homogeneous scheduling to precise matching and flexible execution based on capabilities.

[0105] The automatic parking vehicle scheduling method in this invention addresses the core problem of low overall operational efficiency, safety conflicts, and deadlock risks caused by the lack of systematic coordination among multiple automatic parking vehicles in high-density, multi-vehicle, and highly dynamic smart parking environments. Specifically, the current model of "simple superposition of multiple independent intelligent agents" creates three disconnects between the system and between vehicles and resources: First, the disconnect between static, short-sighted parking space allocation and dynamic, continuous traffic flow demands leads to uneven resource utilization and localized congestion; second, the disconnect between the decision-making rules for individual vehicle autonomous obstacle avoidance and the overall traffic order required for multiple vehicles to coexist leads to deadlocks at complex intersections and a lack of efficient arbitration; third, the disconnect between unified scheduling instructions and the diversity of vehicle performance and parking tasks (such as charging and large parking spaces) results in insufficient system flexibility, making it difficult to meet personalized needs and leverage the heterogeneous potential of vehicles.

[0106] The automatic parking vehicle scheduling method in the parking lot of this invention can realize an intelligent collaborative scheduling system that deeply integrates vehicles, roads, parking lots, and the cloud. It optimizes parking spaces, routes, and tasks from a global perspective, thereby significantly improving the overall throughput, task response speed, and parking space resource utilization rate of the parking lot while ensuring safe scheduling and avoiding vehicle traffic deadlock.

[0107] This invention provides a computer-readable storage medium.

[0108] In this embodiment, a computer program is stored on a computer-readable storage medium. When the computer program is executed by a processor, it implements the automatic parking vehicle scheduling method in a parking lot as described above.

[0109] This invention provides an electronic device.

[0110] In this embodiment, the electronic device may include a memory and a processor. The memory stores a computer program. The computer program, when executed by the processor, implements the automatic parking vehicle scheduling method described above.

[0111] Figure 7 This is a structural block diagram of an electronic device according to an embodiment of the present invention.

[0112] like Figure 7 As shown, the controller 500 includes a processor 501 and a memory 503. The processor 501 and the memory 503 are connected, for example, via a bus 502. Optionally, the controller 500 may also include a transceiver 504. It should be noted that in practical applications, the transceiver 504 is not limited to one, and the structure of the controller 500 does not constitute a limitation on the embodiments of the present invention.

[0113] Processor 501 may be a CPU (Central Processing Unit), a general-purpose processor, a DSP (Digital Signal Processor), an ASIC (Application Specific Integrated Circuit), an FPGA (Field Programmable Gate Array), or other programmable logic devices, transistor logic devices, hardware components, or any combination thereof. It can implement or execute the various exemplary logic blocks, modules, and circuits described in conjunction with the disclosure of this invention. Processor 501 may also be a combination that implements computational functions, such as including one or more microprocessor combinations, a combination of a DSP and a microprocessor, etc.

[0114] Bus 502 may include a pathway for transmitting information between the aforementioned components. Bus 502 may be a PCI (Peripheral Component Interconnect) bus or an EISA (Extended Industry Standard Architecture) bus, etc. Bus 502 can be divided into address bus, data bus, control bus, etc. For ease of representation, Figure 7 The bus is represented by a single thick line, but this does not mean that there is only one bus or one type of bus.

[0115] The memory 503 stores a computer program corresponding to the automatic parking vehicle scheduling method in the parking lot according to the above embodiments of the present invention. This computer program is controlled and executed by the processor 501. The processor 501 executes the computer program stored in the memory 503 to implement the content shown in the aforementioned method embodiments.

[0116] The controller 500 includes, but is not limited to, mobile terminals such as mobile phones, laptops, digital radio receivers, PDAs (personal digital assistants), PADs (tablet computers), PMPs (portable multimedia players), and in-vehicle terminals (such as in-vehicle navigation terminals), as well as fixed terminals such as digital TVs and desktop computers. Figure 7 The controller 500 shown is merely an example and should not be construed as limiting the functionality and scope of use of the embodiments of the present invention.

[0117] This invention provides a method for parking a vehicle.

[0118] Figure 8 This is a flowchart of a vehicle parking method according to an embodiment of the present invention. Figure 8 As shown, a vehicle parking method may include: S601 generates a scheduling request and sends the scheduling request to the automatic parking vehicle scheduling system in the parking lot; S602, in response to the structured guidance information fed back by the automatic parking vehicle scheduling system in the parking lot based on the scheduling request, the vehicle is parked in the target parking space, wherein the structured guidance information is generated by the automatic parking vehicle scheduling method in the parking lot as described above.

[0119] Specifically, when a vehicle enters the parking lot, a scheduling request can be generated based on parking demand. This request is then sent to the parking lot's automated parking vehicle scheduling system. The system then executes the aforementioned automated parking vehicle scheduling method to generate structured guidance information based on the request. Finally, the system receives the structured guidance information from the automated parking vehicle scheduling system and, based on this information, parks the vehicle in the target parking space.

[0120] The vehicle parking method in this embodiment of the invention utilizes an automatic parking vehicle scheduling system in a parking lot. Based on the scheduling request generated by the vehicle, it feeds back structured guidance information, and the vehicle can automatically park into the target parking space based on the structured guidance information.

[0121] It should be noted that the logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a sequenced list of executable instructions for implementing logical functions, and can be specifically implemented in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (such as a computer-based system, a processor-included system, or other system that can fetch and execute instructions from, an instruction execution system, apparatus, or device). For the purposes of this specification, "computer-readable medium" can be any means that can contain, store, communicate, propagate, or transmit programs for use by, or in conjunction with, an instruction execution system, apparatus, or device. More specific examples (a non-exhaustive list) of computer-readable media include: an electrical connection having one or more wires (electronic device), a portable computer disk drive (magnetic device), random access memory (RAM), read-only memory (ROM), erasable and editable read-only memory (EPROM or flash memory), fiber optic devices, and portable optical disc read-only memory (CDROM). Alternatively, the computer-readable medium may be paper or other suitable media on which the program can be printed, since the program can be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, interpreting, or otherwise processing as necessary, and then stored in a computer memory.

[0122] It should be understood that various parts of the present invention can be implemented in hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented in software or firmware stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, it can be implemented using any one or a combination of the following techniques known in the art: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.

[0123] In the description of this specification, references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0124] In the description of this invention, it should be understood that the terms "center," "longitudinal," "lateral," "length," "width," "thickness," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," "outer," "clockwise," "counterclockwise," "axial," "radial," and "circumferential" indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are used only for the convenience of describing this invention and simplifying the description, and are not intended to indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on this invention.

[0125] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of that feature. In the description of this invention, "a plurality of" means at least two, such as two, three, etc., unless otherwise explicitly specified.

[0126] In this invention, unless otherwise explicitly specified and limited, the terms "installation," "connection," "linking," and "fixing," etc., should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral part; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; they can refer to the internal communication of two components or the interaction between two components, unless otherwise explicitly limited. Those skilled in the art can understand the specific meaning of the above terms in this invention according to the specific circumstances.

[0127] In this invention, unless otherwise explicitly specified and limited, "above" or "below" the second feature can mean that the first feature is in direct contact with the second feature, or that the first feature is in indirect contact with the second feature through an intermediate medium. Furthermore, "above," "over," and "on top" of the second feature can mean that the first feature is directly above or diagonally above the second feature, or simply that the first feature is at a higher horizontal level than the second feature. "Below," "below," and "under" the second feature can mean that the first feature is directly below or diagonally below the second feature, or simply that the first feature is at a lower horizontal level than the second feature.

[0128] Although embodiments of the present invention have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting the present invention. Those skilled in the art can make changes, modifications, substitutions and variations to the above embodiments within the scope of the present invention.

Claims

1. A method for automatic parking vehicle scheduling in a parking lot, characterized in that, The method includes: A global scheduling plan is generated based on the scheduling requests and scheduling strategy suggestions of the vehicles to be scheduled using a pre-trained global optimization model. Conflict detection is then performed on the global scheduling plan to generate a scheduling instruction set. The scheduling strategy suggestions are determined by the current resource reservation status of the parking lot, semantic scene description, and resource prediction results. The resource prediction results are determined by the current global dynamic map and historical traffic data of the parking lot. The global scheduling plan includes the target parking space and macro-path corresponding to the vehicles to be scheduled. A resource reservation table is generated based on the scheduling instruction set. Structured guidance information is generated based on the resource reservation table, dynamic right-of-way rules, and the macro-path corresponding to the vehicle to be scheduled. The structured guidance information is then sent to the vehicle to be scheduled to guide it to park in the target parking space. The dynamic right-of-way rules are determined by real-time traffic data in the parking lot and the global scheduling plan.

2. The automatic parking vehicle dispatching method in a parking lot according to claim 1, characterized in that, Before generating a global scheduling plan based on the scheduling requests and scheduling strategy suggestions of the vehicles to be scheduled using a pre-trained global optimization model, the method includes: After performing time and space alignment processing on the field-end sensing data stream and vehicle-end sensing data stream in the parking lot, they are converted to the parking lot digital coordinate system to obtain field-end conversion data and vehicle-end conversion data. Based on the field-end conversion data and the vehicle-end conversion data, the same physical entity is associated and fused to obtain fused sensing data; The global dynamic map is built based on the fused data, and the semantic scene description is generated based on the fused perception data using a pre-trained large language model.

3. The automatic parking vehicle dispatching method in a parking lot according to claim 2, characterized in that, The method includes: Real-time traffic data for each lane in the parking lot is calculated based on the global dynamic map. Based on the real-time traffic data for each lane and the historical traffic data, the resources in the parking lot are predicted to obtain resource prediction results. The scheduling strategy recommendation is generated based on the current resource reservation status, the semantic scenario description, and the resource prediction results. The scheduling strategy recommendation includes at least one scheduling strategy and its corresponding global key performance indicator.

4. The automatic parking vehicle dispatching method in a parking lot according to claim 1, characterized in that, The step of performing conflict detection and generating a scheduling instruction set for the global scheduling plan includes: Conflict detection is performed on the global scheduling plan in both time and space dimensions; If there is no conflict, the scheduling instruction set is generated according to the global scheduling plan; If a conflict exists, the global scheduling plan is coordinated using a preset conflict resolution mechanism, and the scheduling instruction set is generated based on the coordinated global scheduling plan.

5. The automatic parking vehicle dispatching method in a parking lot according to claim 1, characterized in that, The step of generating a resource reservation table based on the scheduling instruction set includes: The system makes time-line reservations for resources in the parking lot according to the scheduling instruction set, generates the resource reservation table, and updates the resource reservation table according to the actual and predicted driving information of the vehicles to be scheduled, and / or updates the resource reservation table according to reservation update requests.

6. The automatic parking vehicle scheduling method in a parking lot according to claim 1, characterized in that, The step of generating structured guidance information based on the resource reservation table, dynamic right-of-way rules, and the macro-path corresponding to the vehicle to be dispatched includes: Based on the real-time traffic data in the parking lot and the global scheduling plan, dynamic right-of-way rules are generated. Based on the dynamic right-of-way rules and the macro-path corresponding to the vehicle to be dispatched, a passage corridor is allocated to the vehicle to be dispatched. The passageway, the dynamic right-of-way rules, and the resource reservation table are packaged to obtain the structured guidance information.

7. The automatic parking vehicle dispatching method in a parking lot according to claim 1, characterized in that, The process of generating a global scheduling plan using a pre-trained global optimization model based on the scheduling requests and scheduling strategy suggestions of the vehicles to be scheduled includes: The scheduling request is subjected to intent recognition and intent extraction to obtain structured demand features. At the same time, the corresponding target vehicle capability model is queried from the pre-established vehicle capability model library based on the vehicle model information of the vehicle to be scheduled. The structured demand features, the target vehicle capability model, and the scheduling strategy suggestions are input into the pre-trained global optimization model to obtain the global scheduling plan.

8. The automatic parking vehicle dispatching method in a parking lot according to claim 1, characterized in that, Based on the structured guidance information, the vehicle to be scheduled is parsed to generate a fine path. Based on the fine path, the vehicle to be scheduled calls the corresponding parking behavior strategy from the pre-established behavior parking strategy library to park in the target parking space.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the automatic parking vehicle scheduling method in a parking lot as described in any one of claims 1-8.

10. An electronic device, comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the computer program is executed by the processor, it implements the automatic parking vehicle scheduling method in a parking lot as described in any one of claims 1-8.

11. A method for parking a vehicle, characterized in that, The method includes: Generate a scheduling request and send the scheduling request to the automatic parking vehicle scheduling system in the parking lot; In response to the structured guidance information fed back by the automatic parking vehicle scheduling system in the parking lot based on the scheduling request, the vehicle is parked in the target parking space, wherein the structured guidance information is generated by the automatic parking vehicle scheduling method in the parking lot as described in any one of claims 1-8.

Images