Parking method, device and equipment based on voice interaction and vehicle

Abstract

This application relates to the field of vehicle-related technology and discloses a parking method, apparatus, device, and vehicle based on voice interaction. The method includes: responding to a user's voice parking command and extracting parking constraints; the parking constraints include at least one of parking area, parking location attribute features, and parking method, each parking constraint including positive and / or negative requirements; while the vehicle is driving in the parking lot, sensing environmental information around the vehicle and obtaining multi-dimensional environmental features of available parking spaces; the multi-dimensional environmental features include at least the area and location attribute features; filtering available parking spaces based on the multi-dimensional environmental features according to the priority of the parking constraints, and controlling the vehicle to park in the available parking space when a parking space that meets the parking constraints is selected. Applying the solution of this application can solve the problems of existing automatic parking systems failing to understand users' personalized parking needs and resulting in parking results that do not meet user expectations.

Description

Technical Field

[0001] This application relates to the field of vehicle-related technology, specifically to a parking method, device, equipment, and vehicle based on voice interaction. Background Technology

[0002] With the continuous development of intelligent driving technology, some vehicles are now able to autonomously find available parking spaces and complete the basic function of parking in parking scenarios.

[0003] However, in complex parking environments, users may have more detailed and personalized parking needs, such as wanting to park in a space near a pillar, avoiding parking at the end of a passageway, or wanting to park in a specific area. Existing parking solutions cannot effectively understand these personalized needs, which may lead to parking results that do not meet user expectations, reducing parking efficiency and user experience. Summary of the Invention

[0004] In view of the above problems, this application provides a parking method, device, equipment and vehicle based on voice interaction to solve the problems that existing automatic parking systems cannot understand users' personalized parking needs, resulting in parking results that do not meet user expectations.

[0005] According to one aspect of the embodiments of this application, a parking method based on voice interaction is provided, the method comprising:

[0006] In response to a user's voice parking command, parking constraints are extracted; wherein, the parking constraints include at least one of parking area, parking location attribute features, and parking method, and each parking constraint includes positive requirements and / or negative requirements;

[0007] While the vehicle is driving in the parking lot, it senses the environmental information around the vehicle and obtains multi-dimensional environmental features of the available parking spaces; wherein, the multi-dimensional environmental features include at least the area and location attribute features.

[0008] According to the priority of the parking constraints, the available parking spaces are filtered based on the multi-dimensional environmental features. When an available parking space that meets the parking constraints is selected, the vehicle is controlled to park in the available parking space.

[0009] According to another aspect of the embodiments of this application, a parking device based on voice interaction is provided, comprising:

[0010] The first processing unit is used to extract parking constraints in response to the user's voice parking command; wherein the parking constraints include at least one of parking area, parking location attribute features and parking method, and each parking constraint includes positive requirements and / or negative requirements.

[0011] The second processing unit is used to sense the environmental information around the vehicle while the vehicle is driving in the parking lot and to obtain multi-dimensional environmental features of the available parking spaces; wherein, the multi-dimensional environmental features include at least the area and location attribute features.

[0012] The third processing unit is used to filter the available parking spaces according to the priority of the parking constraints and based on the multi-dimensional environmental features. When an available parking space that meets the parking constraints is selected, the unit controls the vehicle to park in the available parking space.

[0013] According to another aspect of the embodiments of this application, an electronic device is provided, including: a processor, a memory, a communication interface, and a communication bus, wherein the processor, the memory, and the communication interface communicate with each other through the communication bus;

[0014] The memory is used to store at least one executable instruction that causes the processor to perform the operation of the voice-interaction-based parking method as described in any of the above.

[0015] According to another aspect of the embodiments of this application, a vehicle is provided, the vehicle including the electronic equipment described above.

[0016] According to another aspect of the embodiments of this application, a computer-readable storage medium is provided, the storage medium storing at least one executable instruction that causes an electronic device / apparatus to perform the operation of the voice-interaction-based parking method as described in any of the above claims.

[0017] This application's embodiments utilize natural voice interaction to support users expressing parking needs with forward / reverse commands, enabling the vehicle to accurately understand complex intentions and achieving an intelligent upgrade from "automatically finding parking spaces" to "finding parking spaces on demand." By constructing multi-dimensional environmental features for available parking spaces and finely matching them with user constraints, it ensures that the parked space better meets user expectations, improving satisfaction. Simultaneously, a priority filtering mechanism is introduced to avoid full complex matching, quickly filtering parking spaces that do not meet core requirements, shortening search time, and improving parking efficiency. In summary, this application, through voice command parsing, multi-dimensional feature construction, and priority filtering, achieves accurate understanding and execution of users' complex parking intentions, significantly improving the intelligence level and user experience of automatic parking.

[0018] The above description is merely an overview of the technical solutions of the embodiments of this application. In order to better understand the technical means of the embodiments of this application and to implement them in accordance with the contents of the specification, and to make the above and other objects, features and advantages of the embodiments of this application more obvious and understandable, specific implementation methods of this application are described below. Attached Figure Description

[0019] The accompanying drawings are for illustrative purposes only and are not intended to limit the scope of this application. Furthermore, the same reference numerals denote the same parts throughout the drawings. In the drawings:

[0020] Figure 1 A flowchart illustrating a voice-interaction-based parking method provided in an embodiment of this application;

[0021] Figure 2 A flowchart illustrating another voice-interaction-based parking method provided in this application embodiment;

[0022] Figure 3 A schematic diagram of a parking device based on voice interaction provided in an embodiment of this application;

[0023] Figure 4 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Detailed Implementation

[0024] Exemplary embodiments of the present application will now be described in more detail with reference to the accompanying drawings. Although exemplary embodiments of the present application are shown in the drawings, it should be understood that the present application may be implemented in various forms and should not be limited to the embodiments set forth herein.

[0025] In modern urban transportation, the contradiction between the complexity of parking lots and users' personalized parking needs is becoming increasingly prominent. Especially in large public places such as shopping malls, office buildings, and hospitals, parking lots are usually divided into multiple levels and areas, with significant differences in convenience between each area. For example, the basement level B1 may be close to elevators or stairs, while level B2 may be far from the entrance or exit. Users may want to park in a location close to their destination, convenient for getting in and out of the car, or retrieving items (such as near elevators, stairs, charging stations, etc.).

[0026] However, existing technologies typically use onboard sensors (such as cameras and LiDAR) to scan the surrounding environment, identify the vacancy status and basic characteristics of parking spaces (such as space width and orientation), and select a parking space based on a vacancy-first strategy. This approach cannot effectively understand the user's personalized parking needs, making it impossible to determine whether the parking space meets the user's implicit needs for convenience, safety, etc. The parking result may not match the user's expectations, reducing parking efficiency and user experience.

[0027] To address the aforementioned technical issues, this application provides a voice-interaction-based parking method. Users can issue parking commands via voice, and the vehicle can then use natural language processing technology to analyze the multi-dimensional parking needs in the user's voice commands, combine environmental perception technology to extract the environmental features of available parking spaces, and use hierarchical matching logic to intelligently filter user needs and parking spaces. Ultimately, the vehicle is driven to complete automatic parking that meets personalized needs, thereby overcoming the shortcomings of existing technologies in terms of voice understanding, environmental perception, and matching flexibility.

[0028] Understandably, this application applies to the cockpit systems of intelligent vehicles, especially in scenarios where users in large parking lots (such as shopping malls, office buildings, hospitals, etc.) have personalized needs regarding parking areas, parking locations, and parking methods. The vehicle can receive user commands through an in-vehicle voice interaction module, and, in conjunction with an in-vehicle camera and environmental perception module, scan the parking space environment in real time. Finally, the vehicle's control system executes the parking operation.

[0029] The technical solution of this application and how the technical solution of this application solves the above-mentioned technical problems are described in detail below with specific embodiments. These specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described again in some embodiments. The embodiments of this application will be described below with reference to the accompanying drawings.

[0030] It should be noted that the entity executing the voice-interactive parking method provided in this application embodiment can be a voice-interactive parking device. This device can be deployed on the vehicle controller, intelligent driving domain controller or other electronic devices. This application embodiment does not impose any restrictions, and the method of this application can be implemented by software, hardware or a combination of software and hardware.

[0031] Figure 1 This is a flowchart illustrating a voice-interaction-based parking method provided in an embodiment of this application. This method can be executed by a voice-interaction-based parking device. Figure 1 As shown, the method may include the following steps:

[0032] Step 110: In response to the user's voice parking command, extract the parking constraints.

[0033] The parking constraints include at least one of the following: parking area, parking location attribute characteristics, and parking method. Each parking constraint includes positive requirements and / or negative requirements.

[0034] For example, when a user issues a voice parking command inside the vehicle (e.g., before getting out) or outside the vehicle (e.g., via remote control), the in-vehicle voice interaction system can receive and process the command. The voice parking command refers to a parking operation-related instruction given by the user via voice, such as "Find me a parking space on level B2 near the elevator" or "Park in a convenient spot to retrieve something, not near a pillar." Optionally, the voice parking command can be input by the user all at once, or it can be input sequentially according to the voice interaction prompts between the user and the vehicle after the user expresses a parking intention. It can include both spoken language and structured instructions that the vehicle system can understand; this embodiment does not impose any limitations.

[0035] Parking constraints refer to structured information extracted from voice parking commands that defines the conditions a target vacant parking space must meet. In this application, parking constraints include at least one or more of the following three dimensions: 1) Parking area: refers to the parking space range that the user expects or does not expect, such as floor (B1 / B2), zone (A / B zone), relative location (near entrance / exit, near elevator), etc. 2) Parking location attribute characteristics: refers to the user's preference for the physical attributes of the parking space itself, such as whether it is against a wall, against a pillar, next to an elevator, or has a charging pile, etc. 3) Parking method: refers to the user's requirements for the vehicle's posture after parking, such as left, right, center, forward, backward, etc.

[0036] Each parking constraint includes positive and / or negative requirements. Positive requirements refer to parking characteristics that users expect (such as "near the elevator"), while negative requirements refer to parking characteristics that users do not expect and need to exclude (such as "not near a pillar"). By introducing positive and negative requirements, this application can accurately capture the dual parking needs of users, namely, "what they want" and "what they don't want".

[0037] When the in-vehicle voice interaction system receives a user's voice parking command, the voice-based parking device can use a natural language understanding model to semantically parse the command and, through an intent generalization model, map the colloquial description of the command into structured parking constraints. For example, when a user says "easy to get something," the intent generalization model maps it to "parking method: park forward (leave trunk space)"; when the user says "easy to open the door," it maps it to "parking method: park left / right (leave door side space)," and so on.

[0038] By extracting multi-dimensional constraints and using positive and negative requirements mechanisms, users' vague and colloquial natural language expressions can be transformed into precise parking conditions that are understandable and executable by machines, laying a data foundation for subsequent accurate parking space selection.

[0039] Step 120: While the vehicle is driving in the parking lot, it senses the environmental information around the vehicle and obtains the multi-dimensional environmental features of the available parking spaces; among which, the multi-dimensional environmental features include at least the area and location attribute features.

[0040] For example, when a vehicle enters a parking lot, it is controlled to roam at low speed (e.g., cruise within the parking area) while simultaneously activating the onboard perception system to collect environmental information. The environmental information surrounding the vehicle refers to the physical environment data of the vehicle's surroundings collected by onboard sensors (such as surround-view cameras, ultrasonic radar, millimeter-wave radar, etc.). A vacant parking space refers to a parking space that is currently not occupied by other vehicles. Multidimensional environmental features refer to structured labels generated for each vacant parking space, containing multidimensional attributes.

[0041] In this application, the multidimensional environmental characteristics include at least the area and location attribute characteristics. The area can be determined by identifying area markers in the environment (such as the letter "B2" on a wall, "Area A" markings on a pillar, or signage information); or by using simultaneous positioning and mapping technology combined with an electronic map of the parking lot to confirm which fire compartment the parking space is located in or which landmark it is near. Location attribute characteristics can be determined by analyzing the spatial relative positional relationship between the parking space and surrounding objects, specifically including wall adjacency characteristics, pillar adjacency characteristics, stairwell adjacency characteristics, elevator adjacency characteristics, charging facility characteristics, and parking space adjacency relationship characteristics (e.g., whether the narrow side of the parking space is connected to another parking space).

[0042] In some examples, environmental images can be collected, and image recognition algorithms can be used to detect area indicators in the images. Combined with vehicle positioning information, the current location can be determined. At the same time, for each detected vacant parking space, the distribution of surrounding obstacles can be analyzed to determine whether the parking space is adjacent to walls, pillars, stairwells, elevator entrances, or whether it is equipped with charging piles, etc., and generate a multi-dimensional environmental feature label for the parking space.

[0043] This application upgrades traditional parking space vacancy detection to deep perception of parking space attributes, assigning rich semantic labels to each vacant parking space. This allows vehicles to not only know "where there are vacant spaces," but also "what each vacant space is like," providing data support for subsequent personalized selection.

[0044] Step 130: Based on the priority of parking constraints, select available parking spaces according to multi-dimensional environmental features. When an available parking space that meets the parking constraints is selected, control the vehicle to park in the available parking space.

[0045] For example, the voice-interaction-based parking device matches the user needs extracted in step 110 with the parking space features perceived in step 120, and selects the parking space that best suits the user's preferences according to a preset priority order, thereby enabling the parking operation to be performed.

[0046] Here, priority refers to the ranking of the importance of multiple parking constraints. For example, it can be set as "Parking Area > Parking Location Attributes > Parking Method," meaning that the correct area is ensured first, then parking spaces with matching location attributes are searched within that area, and finally, the parking method requirement is selected based on the best one. The priority can be a preset fixed order, or it can be dynamically adjusted according to user history preferences, emphasis in the command, etc., and this application embodiment does not impose any restrictions.

[0047] When filtering available parking spaces, if the user specifies a positive area requirement (e.g., "B2 level"), the vehicle can be controlled to drive into B2 level; if the user specifies a negative area requirement (e.g., "not B1 level"), B1 level is excluded, and the search continues in the remaining areas. If both positive and negative requirements exist, the negative area is excluded first, and then the positive area is filtered. Then, the scanned available parking spaces are traversed within the defined area to filter those that meet the parking location attribute characteristics. For each parking space, its location attribute label is checked to see if it matches the user's positive requirement (e.g., "near the elevator") and does not match any negative requirement (e.g., "near a pillar"). When evaluating location attribute requirements, the feasibility of the parking method is also evaluated. For example, the parking space size and the distribution of surrounding obstacles are obtained to calculate whether the user's requirement of "left-hand" or "front-hand" can be achieved, until available parking spaces that meet all parking constraints are found. Once the target parking space is determined, the parking route is planned, and the parking trajectory is dynamically adjusted according to the parking method requirements during the parking process to ensure that the vehicle is finally parked in the parking space in the manner expected by the user.

[0048] In some possible implementations, a fault-tolerant parking strategy may also be included. For example, if no suitable parking space is found, the system continues to roam within the area; if no suitable parking space is found after a period of time, such as 10 minutes, the system selects any available parking space in the area to park; if there are no available parking spaces in the area, the system searches for available parking spaces in adjacent areas.

[0049] Through a tiered filtering mechanism, users' multi-dimensional needs are systematically applied to the parking space search process, ensuring that search results strictly follow the user's priority intentions. Simultaneously, a fault-tolerance mechanism guarantees the engineering practicality of the technology, striving to meet users' personalized needs while providing reasonable compromises when these needs cannot be met, preventing vehicles from getting bogged down in an "endless search." Ultimately, the vehicle is parked in the target parking space in the manner expected by the user, achieving a leap from "passive parking" to "personalized parking service."

[0050] In this embodiment, parking constraints containing positive / negative requirements are extracted through natural voice interaction, enabling the vehicle to accurately understand the user's complex intentions. This achieves a leap from "automatically finding parking spaces" to "finding parking spaces on demand," significantly improving the intelligence level of the parking system. Simultaneously, by constructing multi-dimensional environmental features for available parking spaces and performing fine-grained matching with user constraints, it ensures that the parked space meets user expectations as much as possible, reducing user complaints or the need to reposition the car after parking, significantly improving user satisfaction. Furthermore, a priority filtering mechanism avoids performing a full, complex match on all parking spaces, quickly filtering out a large number of spaces that do not meet the core requirements, shortening the search time for the target parking space, and improving parking efficiency. This application, by introducing voice command parsing containing positive / negative requirements, constructing multi-dimensional environmental features, and employing a priority filtering mechanism, achieves accurate understanding and execution of the user's complex parking intentions, effectively improving the user experience and intelligence level of the automatic parking system.

[0051] Figure 2 This is a flowchart illustrating another voice-interaction-based parking method provided in an embodiment of this application. This method can be executed by a voice-interaction-based parking device. Compared to... Figure 1 The embodiment shown introduces mechanisms such as adaptive intent generalization and fault-tolerant feedback to further improve the intelligence level and user experience of the parking process. Figure 2 As shown, the method may include the following steps:

[0052] Step 210: In response to the user's voice parking command, the natural language understanding model is used to perform semantic parsing of the voice parking command, and the colloquial description in the voice parking command is mapped into structured parking constraints through an adaptive intent generalization model.

[0053] The parking constraints include at least one of the following: parking area, parking location attribute characteristics, and parking method. Each parking constraint includes positive requirements and / or negative requirements.

[0054] For example, a Natural Language Understanding (NLU) model refers to a machine learning model used to parse human natural language and extract semantic information, capable of recognizing entities (such as "B2 layer"), intents (such as "finding a parking space"), and modifiers (such as "don't," "near," etc.) in a sentence. For example, a natural language understanding model can be a BERT model, a neural network model, etc.

[0055] The adaptive intent generalization model is based on a pre-built knowledge graph and corpus of the parking domain. It can map users’ diverse colloquial abstract descriptions (such as “easy to get things”, “easy to open the door”) to specific parking parameters (for example, “easy to get things” means that there should be enough space behind the car, which is mapped to parking the car in front). It can also dynamically adjust the mapping rules according to users’ historical behavior data (for example, if the user is tall and is used to opening the door at a large angle, “easy to open the door” may be mapped to more lateral space, etc.).

[0056] Structured parking constraints refer to formatted data that the vehicle's infotainment system can directly understand and use after parsing and mapping. This step involves deep semantic understanding of voice commands. When the in-vehicle voice recognition system receives the user's original voice parking command, it converts it into text. Subsequently, a natural language understanding model performs syntactic analysis and semantic understanding on the text, identifying key entities and intents related to parking (such as "find parking space" and "park"). Next, an adaptive intent generalization model performs secondary processing on the recognition results. For explicit entity descriptions (such as "B2 layer"), they are directly mapped to structured conditions. For abstract descriptions (such as "easy to retrieve something"), a preset mapping rule base or user preference model is invoked to convert them into specific parking posture requirements. Finally, structured data containing multi-dimensional constraints (parking area, parking location attribute features, parking method) and supporting both positive and negative requirements is generated.

[0057] This step, by introducing an adaptive intent generalization model, not only overcomes the limitation of conventional semantic parsing in handling abstract descriptions, but also achieves personalized understanding. This significantly improves the naturalness and intelligence of human-computer interaction.

[0058] Step 220: While the vehicle is driving in the parking lot, collect environmental images of the vehicle's surroundings. By identifying available parking spaces and area indicators in the environmental images, determine the location of the available parking spaces.

[0059] For example, area signage refers to visual elements used to identify area information within a parking lot, including but not limited to: floor text on walls (such as "B2"), zone letters on pillars (such as "Area A"), hanging signs (such as "Entrance / Exit" and "Elevator Hall"), and area color blocks or codes sprayed on the ground.

[0060] When it is determined that a vehicle will be parked in the parking lot, the system controls the vehicle to cruise within the parking lot and activates the perception system (such as surround-view cameras, ultrasonic radar, millimeter-wave radar, etc.) to continuously collect environmental images of the vehicle's surroundings. Then, target detection is performed on each frame of the image to identify available parking spaces (achieved through techniques such as parking line detection and occupancy status determination). Simultaneously, text detection and recognition are performed on the images to extract text information from area indicators in the environmental images. Combined with real-time vehicle positioning information (such as wheel speed sensors, inertial measurement units, etc.), each detected available parking space is assigned an area label, for example, "Parking space A is located in area C on level B2".

[0061] This step decouples parking space detection from area recognition and allows them to work together, enabling vehicles to accurately identify the specific area of ​​the parking lot where each available parking space is located. This provides a reliable data foundation for subsequent area-priority selection. Compared to relying solely on GPS (which typically fails in indoor parking lots) or pre-built maps, this solution is more real-time and adaptable.

[0062] Step 230: Determine the location attribute characteristics of vacant parking spaces by analyzing their spatial relative positional relationship with surrounding objects.

[0063] The location attribute features include one or more of the following: wall adjacency features, column adjacency features, stairwell adjacency features, elevator adjacency features, charging facility features, and parking space adjacency features.

[0064] For example, for the vacant parking spaces detected in step 220, the surrounding area is further analyzed. By analyzing the spatial relative positional relationship between the vacant parking space and surrounding objects (such as walls, pillars, stairwell entrances, elevator entrances, charging piles, adjacent parking spaces, and other static environmental elements around the parking space), the location attribute characteristics of the vacant parking space can be determined. Specifically, semantic segmentation or object detection algorithms can be used to identify targets such as walls, pillars, stairwells, elevator entrances, and charging piles around the parking space; the distance and orientation relationship between the parking space boundary and each target can be calculated to determine whether the parking space possesses various location characteristics, thereby generating a set of location attribute feature labels for each vacant parking space.

[0065] In this embodiment of the application, the location attribute features may include one or more of the following: wall adjacency features, column adjacency features, stairwell adjacency features, elevator shaft adjacency features, charging facility features, and parking space adjacency relationship features. Specifically, the wall adjacency feature indicates whether one or more sides of a parking space are adjacent to a wall; the column adjacency feature indicates whether there is a column next to the parking space (usually meaning limited door opening space); the stairwell adjacency feature indicates whether the parking space is close to the stairwell entrance; the elevator shaft adjacency feature indicates whether the parking space is close to the elevator entrance; the charging facility feature indicates whether a charging pile is provided inside or next to the parking space; and the parking space adjacency relationship feature indicates whether an vacant parking space is connected to another parking space (for example, in a row of parking spaces, the middle parking space is flanked by parking spaces on both sides, while the edge parking space has a wall or column on one side).

[0066] This step elevates environmental perception from simply "whether there are available spaces" to a deeper understanding of "what kind of spaces are available." Through refined extraction of location attribute features, vehicles can accurately identify high-quality or low-quality parking spaces that users frequently focus on, such as "parking spaces against a wall," "parking spaces near a pillar," and "parking spaces near an elevator," providing refined data support for personalized selection.

[0067] Step 240: Use the parking area as the first-level screening condition and control the vehicle to drive to the area that meets the parking area criteria.

[0068] For example, when parking constraints include parking area, parking location attribute features, and parking method, parking spaces are selected in a hierarchical manner according to the priority order of parking area > parking location attribute features > parking method.

[0069] The voice-interactive parking system first extracts the positive and negative requirements related to the parking area from the parking constraints generated in step 210. If a positive area requirement exists (e.g., "Level B2"), a path is planned to guide the vehicle into Level B2. If a negative area requirement exists simultaneously (e.g., "Avoid Level B1"), Level B1 is actively avoided during path planning. During the journey, the system continuously confirms the current location using the area recognition capability from step 220 to ensure the vehicle does not deviate from the target area. Understandably, when planning the driving path, the vehicle can prioritize parking areas that simultaneously meet both positive and negative requirements based on its current location and the parking lot map.

[0070] This strategy of first performing a regional-level coarse screening can narrow the search scope from the entire parking lot to a specific area, avoiding wasting time for vehicles in areas that do not match the user's regional preferences and significantly improving search efficiency.

[0071] Optionally, in one possible embodiment, when there is no parking area that meets both the forward and reverse requirements, the voice-interactive parking device first excludes parking areas that meet the reverse requirements, and then selects vacant parking spaces from the parking areas that meet the forward requirements.

[0072] For example, when parking constraints include both positive and negative requirements for parking areas (e.g., "park on level B2"), and a parking area that meets both requirements cannot be found within a preset time, the system determines that there is no parking area that simultaneously meets both positive and negative requirements. In this case, the system can automatically exclude parking areas that meet the negative requirement; that is, it no longer uses the negative requirement as a filtering condition, but only retains parking areas that meet the positive requirement, and continues searching for available parking spaces within these areas. For instance, if a user requests "park on level B2, not level B1," but there are no available spaces on level B2 while there are available spaces on level B1, the system will abandon the negative requirement of "do not park on level B1" and allow further filtering on level B1 after entering the subsequent area expansion fault-tolerant process.

[0073] Step 250: Use parking location attribute features as the second-level filtering condition and parking method as the third-level filtering condition to filter available parking spaces within the area.

[0074] For example, once a vehicle enters an area that meets the parking zone requirements, a fine-grained screening process is initiated. For identified vacant parking spaces, their location attribute features are checked to see if they meet the parking location attribute feature requirements extracted in step 210 (including positive requirements and negative exclusion requirements). For example, if a user requests "near the elevator and not near a pillar," then only parking spaces that simultaneously have the "near elevator" label and do not have the "near pillar" label will pass the second-level screening.

[0075] Optionally, in one possible embodiment, when there are no available parking spaces that simultaneously meet both the positive and negative parking location attribute characteristics, the available parking spaces that meet the negative parking location attribute characteristics are first excluded, and then the available parking spaces that meet the positive parking location attribute characteristics are selected.

[0076] For example, when entering an area that meets the parking requirements, if the parking constraints include positive requirements (e.g., "near a pillar") and negative requirements (e.g., "not near a wall") for parking location attributes, and no available parking space that simultaneously meets both requirements is found within the area, the system determines that there is no available parking space that simultaneously meets both positive and negative requirements. In this case, the system can automatically exclude parking location attributes that meet the negative requirement, i.e., it will no longer consider the negative requirement of "not near a wall," but will only retain parking location attributes that meet the positive requirement (e.g., "near a pillar") to filter available parking spaces within the area. For example, if all parking spaces near walls in the area are not near pillars, and parking spaces near pillars are not near walls, then the system will abandon the "not near a wall" requirement and prioritize finding a parking space near a pillar.

[0077] For vacant parking spaces that pass the second-level screening, their parking method requirements are further evaluated. If the user has no parking method requirement, the parking space that passes the second-level screening is an available parking space. If the user has a parking method requirement (such as "park on the left"), the parking space size and the distribution of surrounding obstacles are obtained to calculate whether a "park on the left" posture can be achieved (e.g., is there enough space on the left side of the parking space), and whether a safe distance is maintained from other obstacles after posture adjustment. For achievable parking spaces, the safety margin after posture adjustment can be further calculated, and the parking space with the largest safety margin is selected first.

[0078] Optionally, in one possible embodiment, when there are no available parking spaces that meet both the forward and reverse parking requirements, the available parking spaces that meet the reverse parking requirements are first excluded, and then the available parking spaces that meet the forward parking requirements are selected.

[0079] For example, among the candidate parking spaces that meet the first two levels of screening conditions, if the parking constraints include forward requirements (e.g., "perpendicular parking") and reverse requirements (e.g., "do not park at an angle") for the parking method, and no available parking space that meets both requirements is found among these candidates, the system determines that there is no "available parking space that meets both forward and reverse requirements". In this case, the system can automatically exclude parking methods that meet the reverse requirements, that is, it no longer considers the reverse requirement of "do not park at an angle", but only retains parking methods that meet the forward requirements (such as "perpendicular parking"), and performs a final screening on the remaining parking spaces.

[0080] This application employs a two-tiered selection logic: prioritizing location attributes and selecting the optimal parking method. This ensures that the parking space itself aligns with the user's physical preferences (e.g., proximity to elevators) and that the final parking arrangement suits the user's operational habits or convenience needs (e.g., parking on the left for easy exit). These two levels of selection complement each other, achieving comprehensive personalized matching from "parking space selection" to "parking experience." Furthermore, when both positive and negative requirements cannot be simultaneously met, the system automatically downgrades the process by prioritizing positive requirements. This maximizes respect for the user's core desires (positive requirements) while reasonably foregoing secondary avoidance desires (negative requirements), reducing the number of human-computer interactions and improving the automation level and success rate of the parking process.

[0081] Step 260: Determine whether an available parking space that fully meets the parking constraints has been selected within the preset time.

[0082] If yes, proceed to step 270; otherwise, proceed to step 280.

[0083] For example, the preset time refers to the maximum time set for finding a parking space, such as 3 minutes, 5 minutes, or 10 minutes, which can be dynamically adjusted according to the size of the parking lot, the urgency of the user, etc. This time threshold is used to balance the contradiction between "finding the perfect parking space" and "avoiding endless waiting".

[0084] During the parking space filtering process, a timer can be set to record the time elapsed from the start of the search to the current moment. If at least one parking space meeting the criteria is successfully found within the preset time window through the above three-level filtering, the process jumps to step 270; if no space is found after the timeout, it means that all the constraints proposed by the user cannot be met simultaneously in the current parking lot environment, and the process enters the exception handling stage in step 280. This timeout mechanism avoids the system getting stuck in infinite waiting, ensuring the robustness of the method.

[0085] Step 270: Control the vehicle to park in an empty parking space.

[0086] For example, once a vacant parking space that fully meets the parking constraints is selected, the vehicle can automatically plan its parking path from its current location to the target space. During the parking process, the vehicle's steering, acceleration, and braking are dynamically controlled according to the user's requested parking method to ensure that the vehicle ultimately parks in the space in the manner required by the user. After parking is complete, feedback such as "The car has been parked as requested" can be provided to the user via the in-vehicle display screen or voice prompts.

[0087] This step transforms all the previous analysis results into the final parking execution, achieving a complete closed loop from "user intent" to "vehicle behavior." The user requires no additional action, and the vehicle can complete parking according to their personalized needs.

[0088] Step 280: Output a prompt message to the user and relax at least one parking constraint to filter according to the relaxed parking constraint; wherein the prompt message is used to indicate the parking constraint that cannot be met.

[0089] For example, when no perfectly matching parking space is found within the time limit, the voice-interactive parking system first outputs a prompt message to the user via voice synthesis technology. This prompt message informs the user of the current search status and can be output through voice broadcast, in-vehicle screen display, etc. For example: "Sorry, no vacant parking space that is both next to a pillar and perpendicular to the road was found on level B2. Would you like to consider a parking space that is not next to a pillar?" This prompt message clearly indicates the specific constraints that cannot be met, providing the user with a clear decision-making basis. After obtaining user authorization (e.g., the user replies "yes" or "okay") or after a certain period of time, the corresponding constraints can be automatically relaxed. The relaxation order can be preset, for example: first relax the parking method requirements, then relax the location attribute requirements, and finally relax the area requirements; it can also be intelligently selected based on the user's historical behavior or completed according to the user's instructions. The filtering process is then re-executed based on the relaxed conditions. The relaxed filtering is also subject to preset time management, and a secondary time threshold can be set.

[0090] For example, if a suitable parking space cannot be found within 10 minutes, then choose any available parking space in that parking area to park. If there are no available parking spaces in that area, then look for available parking spaces in adjacent areas.

[0091] Through a fault-tolerant mechanism combining transparent feedback and intelligent degradation, users can understand why the system failed to meet all their requirements, avoiding confusion or dissatisfaction. Furthermore, by gradually relaxing the conditions, the vehicle can always find the "optimal" parking space at any given moment, avoiding infinite loops or failed searches. This dynamic interactive adjustment mechanism allows users to participate in decision-making in unsolvable scenarios, preventing parking task failures, improving the flexibility of human-machine collaboration, and enhancing the system's robustness and user-friendliness.

[0092] In this embodiment, by introducing an adaptive intent generalization model, the user's vague, colloquial descriptions can be accurately mapped into structured parking constraints, solving the semantic gap between user intent and machine instructions. Through multi-dimensional environmental perception technology, not only are the locations of available parking spaces identified, but also refined location attribute features such as wall adjacency, column adjacency, and elevator adjacency are deeply extracted, providing a rich data foundation for personalized filtering. By establishing a hierarchical filtering mechanism based on parking area, parking location attribute features, and parking method, and supporting both positive and negative requirements for each constraint, accurate matching and priority ranking of complex user preferences are achieved. Simultaneously, through preset time thresholds and fault-tolerant feedback mechanisms, prompts are output to the user and constraints are intelligently relaxed when user needs cannot be fully met, ensuring search efficiency while improving user transparency and satisfaction. Based on this, this application achieves an intelligent upgrade from "passively searching for available parking spaces" to "actively matching personalized user needs," significantly improving the semantic understanding capability, environmental perception accuracy, and decision-making flexibility of the automatic parking system.

[0093] Optionally, in one possible embodiment, filtering available parking spaces based on multi-dimensional environmental features according to the priority of parking constraints may further include:

[0094] S1. Assign a weight to each parking constraint based on its priority.

[0095] S2. Calculate the overall confidence level of vacant parking spaces based on the degree of compliance and confidence weight of the environmental characteristics of the vacant parking spaces with each parking constraint condition.

[0096] S3. If the overall confidence level of the vacant parking space is greater than or equal to the preset threshold, then the vacant parking space is determined to meet the parking constraint conditions.

[0097] For example, in the aforementioned embodiments, steps 130 and 240-250 mainly employ a priority-based hard screening strategy, requiring vacant parking spaces to meet various parking constraints at each level, ultimately selecting parking spaces that fully meet the requirements. However, in actual parking scenarios, there may be situations where a parking space, while not perfectly meeting all constraints, matches the user's expectations very well (e.g., meeting 90% of the conditions). Therefore, a more flexible confidence-based screening method can be used when selecting vacant parking spaces.

[0098] Specifically, after obtaining the structured parking constraints, the voice-interaction-based parking device does not simply divide them into first, second, and third levels, but instead employs a quantitative weight allocation mechanism. Based on the priority of each constraint, a corresponding confidence weight is assigned to each constraint (including positive and negative requirements). The higher the priority, the larger the assigned weight. For example, the weight of the parking area can be set to 0.5, the weight of parking location attribute features to 0.3, and the weight of the parking method to 0.2, with the sum of the three weights being 1. The weights can be preset to fixed values ​​or dynamically adjusted based on user history preferences and the emphasis in the voice commands; this embodiment does not impose any limitations.

[0099] Then, for each detected vacant parking space, its multidimensional environmental features are compared with each parking constraint in step S1, and a single compliance score is calculated (typically between 0 and 1, where 1 indicates complete compliance and 0 indicates complete non-compliance). Subsequently, the overall confidence level of the parking space is calculated by weighted summation, taking into account the corresponding confidence weights. Next, the calculated overall confidence level of the vacant parking space is compared with a preset overall confidence threshold. When the overall confidence level of the vacant parking space is greater than or equal to the preset threshold, the vacant parking space is determined to "comply with the parking constraints," meaning it is considered a parking space that meets the user's requirements.

[0100] Based on confidence scoring and threshold evaluation mechanisms, vehicles can find a "good enough" parking space within a reasonable time when facing complex and ever-changing parking environments, rather than endlessly pursuing a "perfect match." Furthermore, the threshold can be dynamically adjusted according to the user's urgency and the congestion level of the parking lot. For example, when a user is in a hurry, the threshold can be lowered to find an acceptable parking space more quickly.

[0101] This optional solution transforms rigid hierarchical matching into flexible quantitative scoring decisions by introducing confidence weight allocation, comprehensive confidence calculation, and preset threshold judgment mechanisms. This achieves an intelligent upgrade in parking decision-making, significantly improving adaptability and user experience in complex scenarios. In practical applications, the vehicle system can choose to use hierarchical screening or confidence scoring methods for parking space matching based on specific scenarios (such as user preference complexity, parking lot congestion, etc.), and this application embodiment does not impose any limitations.

[0102] Figure 3 This is a schematic diagram of a voice-interactive-based parking device provided as an embodiment of this application. Figure 3 As shown, the device 30 includes a first processing unit 301, a second processing unit 302, and a third processing unit 303.

[0103] The first processing unit 301 is used to extract parking constraints in response to the user's voice parking command; wherein the parking constraints include at least one of parking area, parking location attribute features and parking method, and each parking constraint includes positive requirements and / or negative requirements.

[0104] The second processing unit 302 is used to sense the environmental information around the vehicle while the vehicle is driving in the parking lot and to obtain multi-dimensional environmental features of the available parking spaces; wherein, the multi-dimensional environmental features include at least the area and location attribute features.

[0105] The third processing unit 303 is used to filter available parking spaces based on multi-dimensional environmental features according to the priority of parking constraints. When an available parking space that meets the parking constraints is selected, the vehicle is controlled to park in the available parking space.

[0106] In one alternative embodiment, the first processing unit 301 is specifically used for:

[0107] In response to the user's voice parking command, the system uses a natural language understanding model to perform semantic parsing of the voice parking command and an adaptive intent generalization model to map the colloquial description in the voice parking command into structured parking constraints.

[0108] In one alternative embodiment, the second processing unit 302 is specifically used for:

[0109] Collect environmental images of the area around the vehicle in the parking lot, and determine the location of the available parking space by identifying the available parking space and area indicator signs in the environmental images;

[0110] By analyzing the spatial relative positional relationship between vacant parking spaces and surrounding objects, the location attribute characteristics of vacant parking spaces are determined; among them, the location attribute characteristics include one or more of the following: wall adjacency characteristics, column adjacency characteristics, stairwell adjacency characteristics, elevator adjacency characteristics, charging facility characteristics, and parking space adjacency relationship characteristics.

[0111] In one alternative embodiment, the third processing unit 303 is specifically used for:

[0112] Use parking areas as the first-level screening condition to control vehicles to drive within the area that meets the parking criteria.

[0113] Using parking location attributes as the second-level filtering condition and parking method as the third-level filtering condition, available parking spaces within the area are filtered.

[0114] In one alternative embodiment, the third processing unit 303 is further configured to:

[0115] If there is no parking area that meets both the forward and reverse requirements, then parking areas that meet the reverse requirements are excluded, and then available parking spaces are selected from the parking areas that meet the forward requirements.

[0116] If there are no available parking spaces that simultaneously meet both the positive and negative requirements, then the available parking spaces that meet the negative requirements are excluded, and then the selection is made from the available parking spaces that meet the positive requirements.

[0117] If there are no available parking spaces that meet both the forward and reverse parking requirements, then parking spaces that meet the reverse parking requirements are excluded, and then the remaining available parking spaces that meet the forward parking requirements are selected.

[0118] In one alternative embodiment, the third processing unit 303 is further configured to:

[0119] If no available parking space that fully meets the parking constraints is found within the preset time, a prompt message is displayed to the user, and at least one parking constraint is relaxed so that the selection can be performed according to the relaxed parking constraints; the prompt message is used to indicate the parking constraint that cannot be met.

[0120] In one alternative embodiment, the third processing unit 303 is specifically used for:

[0121] Assign a weight to each parking constraint based on its priority.

[0122] The overall confidence level of vacant parking spaces is calculated based on the degree of compliance and confidence weight of the environmental characteristics of the vacant parking spaces with each parking constraint.

[0123] If the overall confidence level of the vacant parking space is greater than or equal to the preset threshold, then the vacant parking space is determined to meet the parking constraint conditions.

[0124] As can be seen from the above, the voice-interaction-based parking device provided in this application, by introducing voice command parsing including positive / negative requirements, multi-dimensional environmental feature construction, and priority filtering mechanism, achieves accurate understanding and execution of the user's complex parking intentions, effectively improving the user experience and intelligence level of the automatic parking system.

[0125] Figure 4 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. The specific embodiments of this application do not limit the specific implementation of the electronic device.

[0126] like Figure 4As shown, the electronic device may include: a processor 402, a communications interface 404, a memory 406, and a communications bus 408.

[0127] The processor 402, communication interface 404, and memory 406 communicate with each other via communication bus 408. Communication interface 404 is used to communicate with other network elements such as clients or other servers. The processor 402 executes program 410, specifically performing the relevant steps described in the above embodiment of the parking method for voice interaction.

[0128] Specifically, program 410 may include program code, which includes computer-executable instructions.

[0129] Processor 402 may be a central processing unit (CPU), an application-specific integrated circuit (ASIC), or one or more integrated circuits configured to implement the embodiments of this application. The electronic device includes one or more processors, which may be processors of the same type, such as one or more CPUs; or they may be processors of different types, such as one or more CPUs and one or more ASICs.

[0130] Memory 406 is used to store program 410. Memory 406 may include high-speed RAM memory, and may also include non-volatile memory, such as at least one disk storage device.

[0131] Specifically, program 410 can be called by processor 402 to cause the electronic device to execute the relevant steps in the above-described embodiments of the parking method for voice interaction provided in this application.

[0132] This application also provides a vehicle that includes the electronic devices described above.

[0133] This application also provides a computer-readable storage medium storing at least one executable instruction that, when executed on an electronic device / app, causes the electronic device / app to perform the voice-interaction-based parking method in any of the above method embodiments.

[0134] The algorithms or displays provided herein are not inherently related to any particular computer, virtual system, or other device. Furthermore, the embodiments in this application are not directed to any particular programming language.

[0135] Numerous specific details are set forth in the specification provided herein. However, it will be understood that embodiments of this application may be practiced without these specific details. Similarly, for the purpose of simplification and aiding understanding of one or more aspects of the invention, in the above description of exemplary embodiments of this application, various features of the embodiments are sometimes grouped together in a single embodiment, figure, or description thereof. The claims, which follow the detailed description, are hereby expressly incorporated into that detailed description, wherein each claim itself is a separate embodiment of this application.

[0136] Those skilled in the art will understand that the modules in the device of the embodiment can be adaptively changed and placed in one or more devices different from that embodiment. Modules, units, or components in the embodiment can be combined into a single module, unit, or component, and further, they can be divided into multiple sub-modules, sub-units, or sub-components, except that at least some of such features and / or processes or units are mutually exclusive.

[0137] It should be noted that the above embodiments are illustrative of this application and not restrictive, and those skilled in the art can devise alternative embodiments without departing from the scope of the appended claims. In the claims, any reference signs placed between parentheses should not be construed as limiting the claims. The word "comprising" does not exclude the presence of elements or steps not listed in the claims. The word "a" or "an" preceding an element does not exclude the presence of a plurality of such elements. This application can be implemented by means of hardware comprising several different elements and by means of a suitably programmed computer. In the unit claims enumerating several means, several of these means may be embodied by the same item of hardware. The use of the words first, second, and third, etc., does not indicate any order. These words can be interpreted as names. The steps in the above embodiments, unless otherwise specified, should not be construed as limiting the order of execution.

Claims

1. A parking method based on voice interaction, characterized in that, The method includes: In response to a user's voice parking command, parking constraints are extracted; wherein, the parking constraints include at least one of parking area, parking location attribute features, and parking method, and each parking constraint includes positive requirements and / or negative requirements; While the vehicle is driving in the parking lot, it senses the environmental information around the vehicle and obtains multi-dimensional environmental features of the available parking spaces; wherein, the multi-dimensional environmental features include at least the area and location attribute features. According to the priority of the parking constraints, the available parking spaces are filtered based on the multi-dimensional environmental features. When an available parking space that meets the parking constraints is selected, the vehicle is controlled to park in the available parking space.

2. The method according to claim 1, characterized in that, The process of extracting parking constraints in response to a user's voice parking command includes: In response to the user's voice parking command, the system performs semantic parsing of the voice parking command using a natural language understanding model, and maps the colloquial description in the voice parking command into structured parking constraints using an adaptive intent generalization model.

3. The method according to claim 1, characterized in that, The sensing of the surrounding environment of the vehicle, obtaining multi-dimensional environmental features of the available parking spaces, includes: Collect environmental images of the area around the vehicle in the parking lot, and determine the location of the vacant parking space by identifying the vacant parking space and area indicator in the environmental images; By analyzing the spatial relative positional relationship between the vacant parking space and surrounding objects, the location attribute characteristics of the vacant parking space are determined; wherein, the location attribute characteristics include one or more of the following: wall adjacency characteristics, column adjacency characteristics, stairwell adjacency characteristics, elevator adjacency characteristics, charging facility characteristics, and parking space adjacency relationship characteristics.

4. The method according to claim 1, characterized in that, The step of filtering available parking spaces based on the priority of the parking constraints and the multi-dimensional environmental features includes: The parking area is used as the first-level screening condition to control the vehicle to drive to the area that meets the parking area criteria. The parking location attribute features are used as the second-level filtering condition, and the parking method is used as the third-level filtering condition to filter available parking spaces within the area.

5. The method according to claim 4, characterized in that, The method further includes: If there is no parking area that meets both the forward and reverse requirements, then parking areas that meet the reverse requirements are excluded, and then available parking spaces are selected from the parking areas that meet the forward requirements. If there are no available parking spaces that simultaneously meet both the positive and negative requirements, then the available parking spaces that meet the negative requirements are excluded, and then the selection is made from the available parking spaces that meet the positive requirements. If there are no available parking spaces that meet both the forward and reverse parking requirements, then parking spaces that meet the reverse parking requirements are excluded, and then the remaining available parking spaces that meet the forward parking requirements are selected.

6. The method according to any one of claims 1-5, characterized in that, The method further includes: If no available parking space that fully meets the parking constraints is found within a preset time, a prompt message is output to the user, and at least one parking constraint is relaxed so that the selection can be performed according to the relaxed parking constraints; wherein, the prompt message is used to indicate the parking constraints that cannot be met.

7. The method according to claim 1, characterized in that, The step of filtering available parking spaces based on the priority of the parking constraints and the multi-dimensional environmental features includes: Based on the priority of the parking constraints, assign a weight to each of the parking constraints. The overall confidence level of the vacant parking space is calculated based on the degree of compliance of the environmental characteristics of the vacant parking space with each parking constraint and the confidence weight. If the overall confidence level of the vacant parking space is greater than or equal to a preset threshold, then the vacant parking space is determined to meet the parking constraint conditions.

8. A parking device based on voice interaction, characterized in that, The device includes: The first processing unit is used to extract parking constraints in response to the user's voice parking command; wherein the parking constraints include at least one of parking area, parking location attribute features and parking method, and each parking constraint includes positive requirements and / or negative requirements. The second processing unit is used to sense the environmental information around the vehicle while the vehicle is driving in the parking lot and to obtain multi-dimensional environmental features of the available parking spaces; wherein, the multi-dimensional environmental features include at least the area and location attribute features. The third processing unit is used to filter the available parking spaces according to the priority of the parking constraints and based on the multi-dimensional environmental features. When an available parking space that meets the parking constraints is selected, the unit controls the vehicle to park in the available parking space.

9. An electronic device, characterized in that, include: The processor, memory, communication interface, and communication bus are provided, wherein the processor, memory, and communication interface communicate with each other via the communication bus. The memory is used to store at least one executable instruction that causes the processor to perform the operation of the voice-interaction-based parking method as described in any one of claims 1-7.

10. A vehicle, characterized in that, The vehicle includes the electronic equipment as described in claim 9.

Images