Passenger behavior driven navigation method and related equipment

Abstract

The invention discloses a passenger behavior-driven navigation method and related equipment, and relates to the technical field of vehicle navigation, and the method comprises the following steps: obtaining behavior data of a passenger; on the basis of the behavior data, attention parameters used for representing continuous attention behaviors of the passengers on the interest points outside the vehicle are determined, and the attention parameters comprise the accumulated attention duration and the space matching degree; when the attention parameter meets a preset triggering condition, generating interest point information based on the behavior data and the position data; generating interaction prompt information containing the target landmark based on the interest point information; and dynamically adjusting a navigation path according to a feedback result of the passenger to the interaction prompt information. According to the invention, the active perception capability and the interactive intelligence level of the navigation system are improved, while the operation intervention of a driver is reduced, personalized and scenarized navigation services are provided for passengers, and the driving safety and the user experience are effectively enhanced.

Description

Technical Field

[0001] This application relates to the technical field of vehicle navigation, and particularly to a navigation method and related devices driven by passenger behavior. Background Art

[0002] With the development of in-vehicle intelligent systems, existing navigation technologies mainly focus on the path planning and driving safety of drivers, but there are deficiencies in passenger interaction and personalized services. Traditional navigation systems rely on manual input or preset paths by drivers and cannot actively sense the attention behaviors of passengers towards points of interest outside the vehicle, resulting in one-way interaction and passive recommendation. In addition, passengers need to add waypoints through the operation of the driver, which not only increases the communication cost but also may distract the driver's attention and affect driving safety. Therefore, there is an urgent need for a navigation method driven by passenger behavior to solve the above-mentioned technical problems. Summary of the Invention

[0003] A series of simplified concepts are introduced in the Summary of the Invention section, which will be further elaborated in the Detailed Description section. The Summary of the Invention section of this application does not mean to attempt to define the key features and essential technical features of the claimed technical solution, nor does it mean to attempt to determine the protection scope of the claimed technical solution.

[0004] In a first aspect, this application provides a navigation method driven by passenger behavior, the method comprising:

[0005] Obtaining the behavior data of the passenger;

[0006] Based on the behavior data, determining a degree of attention parameter for characterizing the continuous attention behavior of the passenger towards points of interest outside the vehicle, wherein the degree of attention parameter includes the cumulative attention duration and the spatial matching degree;

[0007] When the degree of attention parameter meets a preset trigger condition, generating point of interest information based on the behavior data and the location data, the point of interest information including the target type, the target geographical location and the attention intensity of the target landmark;

[0008] Generating an interaction prompt message including the target landmark based on the point of interest information;

[0009] Dynamically adjusting the navigation path according to the feedback result of the passenger on the interaction prompt message.

[0010] In some embodiments, obtaining the behavior data of the passenger includes:

[0011] Collecting the facial and head image data of the passenger, and extracting the line of sight direction and the head pose from the image data by using computer vision algorithms;

[0012] Capture the spatial position and movement trajectory of the passenger's hand to generate gesture data including gesture actions and gesture trajectories;

[0013] Perform timestamp synchronization processing on the line-of-sight direction, head pose, and gesture data, and perform denoising fusion on the synchronized data through a preset filtering algorithm to obtain behavior data.

[0014] In some embodiments, based on the behavior data, determine the attention parameter used to characterize the continuous attention behavior of the passenger to the points of interest outside the vehicle, including:

[0015] Extract features from the line-of-sight direction and head pose in the behavior data through a convolutional neural network to determine the passenger's line-of-sight fixation area and the cumulative attention duration of the line-of-sight fixation area;

[0016] Analyze the gesture actions and gesture trajectories in the behavior data through a trajectory tracking algorithm to generate spatial trajectory data of the hand pointing direction;

[0017] Calculate the matching degree between the line-of-sight fixation area and the spatial trajectory data to obtain the spatial matching degree between the line-of-sight fixation area and the hand pointing direction.

[0018] In some embodiments, when the attention parameter meets the preset trigger condition, generate point-of-interest information based on the behavior data and location data, including:

[0019] When the cumulative attention duration exceeds the first preset threshold and the spatial matching degree exceeds the second preset threshold, it is determined that the attention parameter meets the preset trigger condition;

[0020] When the attention parameter meets the preset trigger condition, based on the geographical location of the line-of-sight fixation area, match the preset landmark database to obtain the target type and target geographical location of the target landmark, where the preset landmark database includes multiple geographical locations and the landmark types corresponding to each geographical location;

[0021] Determine the attention intensity of the target landmark according to the weighted calculation result of the cumulative attention duration and the spatial matching degree;

[0022] Integrate the target type, target geographical location, and attention intensity into point-of-interest information.

[0023] In some embodiments, generate interaction prompt information including the target landmark based on the point-of-interest information, including:

[0024] Based on the current vehicle driving state, select an interaction method, where the interaction method includes voice broadcast prompt or navigation interface pop-up window;

[0025] Generate interactive content including the name of the target landmark, the target type, the target geographical location, and the estimated arrival time, where the estimated arrival time is calculated based on the real-time navigation path and traffic information;

[0026] According to the interaction method, dynamically render the interactive content into executable interactive prompt information.

[0027] In some embodiments, before generating the interactive prompt information including the target landmark based on the point-of-interest information, it further includes:

[0028] Obtain the current vehicle speed and the current road condition complexity;

[0029] If the current vehicle speed is less than or equal to the preset vehicle speed threshold and the current road condition complexity is less than or equal to the preset risk level, determine that the current vehicle driving state is in the safe interaction mode, and execute the step of generating the interactive prompt information including the target landmark based on the point-of-interest information.

[0030] In some embodiments, according to the feedback result of the passenger on the interactive prompt information, dynamically adjust the navigation path, including:

[0031] Detect the feedback result of the passenger on the interactive prompt information, where the feedback result includes confirming to add, rejecting to add, or ignoring the prompt;

[0032] When the feedback result is confirming to add, add the target landmark as a dynamic waypoint to the navigation path, and re-plan the navigation path based on the real-time traffic information;

[0033] When the feedback result is rejecting to add or ignoring the prompt, based on the target type of the target landmark, reduce the recommendation priority of at least one landmark corresponding to the target type, and maintain the current navigation path.

[0034] In a second aspect, the present application proposes a navigation device driven by passenger behavior, and the device includes:

[0035] A behavior data acquisition unit for acquiring the behavior data of the passenger;

[0036] A focus parameter extraction unit for determining, based on the behavior data, a focus parameter used to characterize the continuous focus behavior of the passenger on the points of interest outside the vehicle, where the focus parameter includes the cumulative focus duration and the spatial matching degree;

[0037] An interest information generation unit for generating point-of-interest information based on the behavior data and the location data when the focus parameter meets the preset trigger condition, where the point-of-interest information includes the target type, the target geographical location, and the focus intensity of the target landmark;

[0038] An interactive information prompt unit for generating interactive prompt information including the target landmark based on the point-of-interest information;

[0039] A navigation path optimization unit, configured to dynamically adjust the navigation path according to the feedback result of the passenger on the interactive prompt information.

[0040] In a third aspect, an electronic device includes: a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program stored in the memory, the steps of the navigation method driven by passenger behavior according to any one of the first aspects are implemented.

[0041] In a fourth aspect, the present application provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the navigation method driven by passenger behavior according to any one of the first aspects is implemented.

[0042] In summary, the present application obtains multi-modal behavior data such as the passenger's line of sight direction, head posture, and gesture actions in real time, accurately identifies the continuous attention behavior of the passenger to the points of interest outside the vehicle, and generates interest point information including the target landmark type, location, and attention intensity based on a preset trigger condition. By dynamically generating interactive prompt information and adjusting the navigation path according to the passenger's feedback, the initiative and intelligence level of the navigation system are improved. The present application reduces the driver's operation intervention, enhances driving safety, and at the same time provides a personalized and intuitive interaction experience for passengers, effectively improving user satisfaction. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] By reading the detailed description of the preferred embodiments below, various other advantages and benefits will become clear to those of ordinary skill in the art. The drawings are only for the purpose of illustrating the preferred embodiments and are not considered to limit this specification. And throughout the drawings, the same reference numerals are used to represent the same components. In the drawings:

[0044] Figure 1 It is a schematic flow chart of a navigation method driven by passenger behavior provided by an embodiment of the present application;

[0045] Figure 2 It is a schematic structural diagram of a navigation device driven by passenger behavior provided by an embodiment of the present application;

[0046] Figure 3 It is a structural diagram of a navigation electronic device driven by passenger behavior provided by an embodiment of the present application. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0047] In the description, claims and the above-mentioned drawings of this application, the terms "first", "second", "third", "fourth", etc. (if any) are used to distinguish similar objects and do not necessarily describe a specific order or sequence. It should be understood that the data used in this way can be interchanged under appropriate circumstances so that the embodiments described here can be implemented in an order other than those illustrated or described here. In addition, the terms "include" and "have" and any variations thereof are intended to cover non-exclusive inclusion. For example, a process, method, system, product or device that includes a series of steps or units does not necessarily have to be limited to those steps or units clearly listed, but may include other steps or units not clearly listed or inherent to these processes, methods, products or devices. Next, the technical solutions in the embodiments of this application will be clearly and completely described in conjunction with the drawings in the embodiments of this application. Obviously, the described embodiments are only a part of the embodiments of this application, rather than all of the embodiments.

[0048] Please refer to Figure 1 , which is a schematic flow chart of a passenger behavior-driven navigation method provided by an embodiment of this application, and specifically may include:

[0049] S110. Obtain the behavior data of the passenger;

[0050] Exemplarily, obtaining the passenger behavior data is realized through the cooperation of in-vehicle multimodal sensors, and its core lies in real-time capturing the physiological and motion characteristics of the passenger. The on-vehicle camera is used to collect the facial and head image data of the passenger, and combined with the depth sensor to detect the spatial position information, the line-of-sight direction and head posture can be analyzed; at the same time, the hand gesture recognition device is used to track the hand movement trajectory and capture the pointing intention of the hand gesture action. After the multi-source data is synchronized and filtered by the time stamp, a fused behavior data stream is formed, providing a basic input for subsequent analysis.

[0051] The technical essence of this step is to convert the visual attention and limb movements of the passenger into quantifiable digital signals. The error interference of a single data source is eliminated through sensor fusion technology to ensure the spatio-temporal consistency of the behavior data, so as to accurately reflect the passenger's attention behavior to the points of interest outside the vehicle and establish data support for subsequent continuous attention recognition and point-of-interest recommendation.

[0052] S120. Based on the behavior data, determine the attention parameter used to characterize the continuous attention behavior of the passenger to the points of interest outside the vehicle, where the attention parameter includes the cumulative attention duration and the spatial matching degree;

[0053] Exemplarily, determining the attention parameter is the core link for identifying passengers' continuous attention behavior. It constructs key indicators representing the intensity of passengers' interest by quantitatively analyzing the time and space dimension features in the behavior data. The cumulative attention duration reflects the duration of passengers' attention to specific out-of-vehicle interest points and is achieved by counting the effective fixation duration within a preset time window; the spatial matching degree measures the consistency between the passenger's gesture direction and the line-of-sight focus area, and calculates the collaborative correlation between the two based on the geometric coincidence degree. These two parameters together constitute the objective quantitative basis for passengers' attention behavior.

[0054] The collaborative design of the attention parameter aims to solve the limitations of a single behavior signal. The cumulative attention duration captures the persistence of visual attention and avoids misjudgment caused by short-term distractions; the spatial matching degree enhances the authenticity of the behavior intention through the cross-verification of gestures and the line of sight. The combination of the two not only improves the reliability of interest point recognition but also provides multi-dimensional data support for subsequent trigger condition determination, ensuring that the system can accurately capture the passengers' active interaction needs and lay a foundation for intelligent navigation decision-making.

[0055] S130. When the attention parameter meets the preset trigger condition, generate interest point information based on the behavior data and location data. The interest point information includes the target type of the target landmark, the target geographical location, and the attention intensity.

[0056] Exemplarily, when the cumulative attention duration and the spatial matching degree reach the preset threshold, the system generates structured interest point information based on the fused behavior data (line-of-sight focus, gesture direction) and the vehicle's real-time location data. This process matches the spatial area of passengers' attention with the pre-set landmark information by associating with the map database, determines the actual geographical location and attribute type of the target landmark, and completes the accurate mapping from the behavior signal to the physical landmark.

[0057] The generated interest point information contains three core elements: the target type reflects the functional attributes of the landmark, the target geographical location provides the navigation coordinate reference, and the attention intensity quantifies the degree of passengers' interest. These information provide a decision-making basis for subsequent interaction prompts and route optimization, enabling the system to convert passengers' unconscious behaviors into executable navigation instructions and realizing the closed-loop link from behavior perception to service response.

[0058] S140. Generate interaction prompt information containing the target landmark based on the interest point information.

[0059] Exemplarily, generating interactive prompt information is a key step for the system to convert passengers' points of interest into actionable services. When the attention parameter meets the triggering condition, the system dynamically generates interactive content including the name, type, and estimated arrival time of the target landmark based on the point of interest information, combined with the vehicle's real-time position and navigation data. According to the current driving state, the system adaptively selects interactive methods such as voice broadcast or interface pop-up to ensure the timeliness and safety of information transmission.

[0060] The technical core of this step lies in converting structured point of interest information into interactive operation instructions. By associating the semantic information of the landmark with real-time environmental data, dynamically generating prompt content including the key attributes of the target landmark and navigation suggestions, and achieving seamless connection from passengers' behavior perception to navigation decision-making, improving the accuracy and scene adaptability of the interaction.

[0061] S150. Dynamically adjust the navigation path according to the feedback result of the passenger on the interactive prompt information.

[0062] Exemplarily, the core logic of dynamically adjusting the navigation path lies in directly associating the passenger's real-time feedback with the system decision-making. When the passenger confirms adding a point of interest, the system includes the target landmark as a dynamic waypoint in the path planning and recalculates the optimal route in combination with real-time traffic information. If the passenger refuses or ignores the prompt, the system reduces the recommended priority of the landmark corresponding to the target type, reduces the triggering frequency of subsequent similar interactions, and at the same time maintains the stability of the current navigation path.

[0063] This step realizes the adaptive optimization of the navigation system through a closed-loop feedback mechanism. The passenger's feedback result directly affects the immediate adjustment of the path. This dynamic response ability enables the system to take into account personalized needs and driving efficiency, reduce ineffective interference, and ensure that navigation decisions always revolve around the passenger's true intentions, ultimately improving driving safety and user satisfaction.

[0064] In summary, in the embodiments of the present application, by obtaining passengers' behavioral data such as line of sight direction, head posture, and gesture actions in real time, combining deep learning and trajectory matching algorithms, accurately identifying passengers' continuous attention behaviors to points of interest outside the vehicle, and dynamically generating point of interest information including the type, geographical location, and attention intensity of the target landmark. When the attention intensity meets the preset triggering condition, the system automatically generates multimodal interactive prompts and dynamically adjusts the navigation path according to the passenger's feedback. This method significantly improves the active perception and intelligent level of the navigation system, reduces the dependence on the driver's operations, and reduces the distraction risk during driving. At the same time, by integrating real-time traffic information and users' historical preferences, optimizing the path planning efficiency, providing personalized and high-precision navigation services for passengers, and effectively enhancing the user experience and driving safety.

[0065] In some instances, obtaining the passengers' behavioral data includes:

[0066] Collect the facial and head image data of passengers, and use computer vision algorithms to extract the line-of-sight direction and head pose from the image data;

[0067] Capture the spatial position and movement trajectory of the passenger's hand, and generate gesture data including gesture actions and gesture trajectories;

[0068] Perform timestamp synchronization processing on the line-of-sight direction, head pose, and gesture data, and perform denoising fusion on the synchronized data through a preset filtering algorithm to obtain behavior data.

[0069] Exemplarily, the facial and head image data of passengers are collected in real time through high-definition cameras (including infrared cameras) configured in the vehicle, and the computer vision algorithms are used to process the image data. Specifically, a face key point detection algorithm is used to locate the facial feature points of passengers (such as pupils, the tip of the nose, and the corners of the mouth), and combined with a head pose estimation algorithm (such as Euler angle calculation based on three-dimensional space geometric transformation), the line-of-sight direction vector and the head rotation angle are determined. The line-of-sight direction is calculated through the relative relationship between the pupil position and the head pose, and the head pose is resolved into pitch angle, yaw angle, and roll angle through the spatial distribution of facial feature points, so as to quantify the visual attention area of passengers.

[0070] The three-dimensional spatial position and continuous movement trajectory of the passenger's hand are captured in real time through gesture sensors (such as ToF sensors, capacitive arrays). The displacement paths of hand key points (such as fingertips, palm centers) are analyzed using a trajectory tracking algorithm (such as continuous frame-by-frame motion prediction based on Kalman filtering) to generate gesture action types (such as pointing, waving) and gesture trajectory data (such as the pointing path from the seat to the window). Gesture actions are recognized through a preset classification model (such as SVM or CNN), and the trajectory data describes the dynamic characteristics of hand movement through a spatial coordinate sequence.

[0071] Perform timestamp synchronization processing on the line-of-sight direction, head pose, and gesture data from different sensors to ensure that multi-modal data is aligned under a unified time reference. Specifically, an accurate time tag based on the system clock is assigned to each frame of image data and sensor signal, and the interpolation algorithm is used to compensate for the differences in the sampling rates of different sensors. The synchronized data is segmented according to time windows to ensure that the line-of-sight, pose, and gesture data within the same window reflect the real-time collaborative behavior of passengers.

[0072] The synchronized raw data is filtered and denoised. For image data, Gaussian filtering is used to eliminate light noise, and motion compensation algorithm is used to reduce image blur caused by rapid head rotation. For gesture data, median filtering is used to smooth trajectory jitter noise. Subsequently, the gaze direction, head posture and gesture trajectory data are integrated into structured behavior data through feature-level fusion algorithms using weighted splicing or attention mechanisms. This data is used as the basic input for subsequent behavior recognition and point of interest analysis, and is transmitted to the on-board computing unit through a unified data interface to ensure real-time performance and processing efficiency.

[0073] In some examples, based on the behavior data, determining an attention parameter for characterizing the passenger's sustained attention behavior to a point of interest outside the vehicle includes:

[0074] The convolutional neural network is used to extract features of the gaze direction and head posture in the behavior data to determine the passenger's gaze area and the cumulative attention duration of the gaze area.

[0075] The gesture movements and gesture trajectories in the behavior data are analyzed through the trajectory tracking algorithm to generate spatial trajectory data of the hand pointing direction;

[0076] The matching degree between the gaze area and the spatial trajectory data is calculated to obtain the spatial matching degree between the gaze area and the hand pointing direction.

[0077] Exemplarily, the feature extraction of the sight direction and head posture in the behavior data is performed by a convolutional neural network to determine the passenger's sight fixation area and the cumulative attention duration of the sight fixation area, including:

[0078] The extracted sight direction vector (indicating the three-dimensional spatial direction of the passenger's sight) and head posture parameters (including pitch angle, yaw angle, and roll angle) are used as input data. These data are standardized through the preprocessing module of the on-board computing unit to ensure the uniformity of the input format. For example, the head posture angle is normalized to a range of 0 to 360 degrees, and the sight direction vector is converted to unit length to eliminate the dimensional effects caused by vehicle bumps or differences in passenger sitting postures.

[0079] The pre-trained convolutional neural network (CNN) receives the above preprocessed data and analyzes the spatiotemporal correlation between the gaze direction and the head posture through multi-layer convolution operations. Specifically, the first layer uses a small 3×3 convolution kernel to extract local features (such as short-term fluctuations in the gaze direction), and the second layer uses a larger 5×5 convolution kernel to capture global patterns (such as the long-term correlation between changes in head posture and gaze direction). The model reduces the dimension of the features through the pooling layer and retains key information (such as the stability characteristics of the gaze focus). The final output high-order features include the fluctuation amplitude of the gaze direction within the time window, the focus offset frequency, etc., which are used to quantify the passenger's concentration.

[0080] Based on GPS to obtain the real-time position of the vehicle and the vehicle body coordinate system, the system maps the line-of-sight direction vector to the external geographic coordinate system. Through the ray tracing algorithm, calculate the intersection points of the line-of-sight vector and the external environment model (such as pre-set 3D map data or landmark database), and determine the physical area that the passenger actually gazes at. If the intersection point is within the geographic fence boundary of the predefined point of interest, then determine that area as the line-of-sight fixation area, and record its boundary coordinates and the position of the center point.

[0081] Real-time track whether the line-of-sight focus continuously falls on the same area of interest. Through frame-by-frame analysis, if the deviation of the line-of-sight direction between adjacent frames is less than the tolerance threshold (such as 5 degrees), then determine that frame as a valid fixation frame. The cumulative attention duration is calculated by accumulating the durations of all valid fixation frames (for example, if 25 out of 30 frames per second are valid, then the cumulative duration increases by 0.83 seconds). If the line-of-sight focus deviates from the target area by more than the threshold, then reset the cumulative timing to ensure that only continuous and stable attention behaviors are counted. The finally generated cumulative attention duration parameter provides a quantitative basis in the time dimension for subsequent trigger condition determination.

[0082] Analyze the gesture actions and gesture trajectories in the behavior data through the trajectory tracking algorithm to generate spatial trajectory data of the hand pointing direction, including:

[0083] Real-time capture the three-dimensional coordinate data of the key points of the passenger's hand through a depth sensor, including the precise spatial positions of parts such as fingertips and palm centers. The sensor generates a continuous frame sequence at a sampling rate of 60Hz, and each frame contains the coordinate information of the key points of the hand. Due to sensor noise and environmental interference (such as light changes, rapid hand movement), there are jitters and outliers in the original trajectory data. For this reason, the Kalman filter algorithm is used to predict and correct the hand coordinates of each frame: first predict the coordinates of the current frame based on the position and speed of the previous frame, and then adjust the prediction result through the actual measurement value, so as to smooth the trajectory and suppress noise, and generate a stable and continuous hand movement path.

[0084] Analyze the filtered hand trajectory data to identify the movement trend of pointing gestures. Detect the consistency of the hand movement direction through a sliding window mechanism (such as a continuous 5 frames as a window). If the average angular change of the hand movement direction within the window is less than the preset threshold (such as 10 degrees), then determine it as a valid pointing action. Subsequently, use the least squares method to perform linear fitting on the coordinate points within the window to obtain the best fitting straight line path of the hand pointing. The starting point of this path is the hand position of the initial frame, the end point is the predicted position of the last frame, and the direction angle is the angle between the fitting straight line and the reference axis of the vehicle coordinate system, fully characterizing the spatial trajectory characteristics of the hand pointing.

[0085] The fitted trajectory data is input into a pre-trained gesture classification model to identify the type of gesture action. The classification model makes decisions based on the length of the trajectory, the rate of change of the direction angle, and the speed characteristics: Pointing gestures usually show a stable trajectory direction and a gentle speed; Waving gestures have larger direction fluctuations and speed changes. If the classification result is a pointing action, the trajectory is marked as a valid interaction signal, and its spatial parameters (starting point, ending point, direction angle) are extracted for subsequent matching calculations; Non-pointing gestures are filtered to avoid invalid interference.

[0086] Calculate the matching degree between the line-of-sight fixation area and the spatial trajectory data to obtain the spatial matching degree between the line-of-sight fixation area and the hand pointing direction, including:

[0087] Extract the boundary information of the passenger's line-of-sight fixation area, which is composed of a polygon surrounded by multiple vertex coordinates, representing the spatial range of the passenger's visual focus. At the same time, obtain the end coordinate and pointing direction of the hand pointing trajectory, which is determined by the continuous movement path of the hand from the starting position to the ending position, and the direction is characterized by the straight-line extension from the starting point to the ending point.

[0088] Through the geometric projection algorithm, extend the hand pointing trajectory infinitely along its natural direction into the vehicle exterior environment. The system calculates the intersection position of the extended trajectory and the line-of-sight fixation area polygon. Specifically, use the ray and polygon intersection detection algorithm to determine whether the extended trajectory passes through the boundary of the line-of-sight fixation area. If there are intersections, record the coordinates of all intersections and determine whether they are inside the polygon.

[0089] If the extended trajectory and the line-of-sight fixation area have intersections and are inside the polygon, calculate the projection coverage area (i.e., the overlapping area) of the hand pointing trajectory within the line-of-sight area. The overlapping area is obtained by triangulating the polygon sub-region within the projection range and accumulating the areas of all triangles. The matching degree is defined as the ratio of the overlapping area to the total area of the line-of-sight fixation area, expressed as a percentage. For example, if the total area of the line-of-sight area is 100 square meters and the overlapping area is 70 square meters, the matching degree is 70%.

[0090] The system presets a matching degree threshold (such as 70%). If the actual matching degree exceeds this threshold, it is determined that the gesture and the line of sight co-point to the same point of interest, triggering the subsequent interaction process. If the matching degree does not reach the threshold, it is considered that the gesture and the line of sight do not form an effective collaboration, and the system ignores this behavior data to avoid mis-triggering. The matching degree threshold can be dynamically adjusted according to the actual scenario (such as increasing the threshold during high-speed driving to reduce interference) to balance sensitivity and reliability.

[0091] The system independently encodes the cumulative attention duration and spatial matching degree to ensure the unity and comparability of data formats. The cumulative attention duration is directly encoded as a numerical scalar in seconds (5 seconds of continuous gaze is encoded as 5.0), reflecting the duration of the passenger's attention to the point of interest. The spatial matching degree is encoded in percentage form (70% matching degree is encoded as 0.7), quantifying the coordination degree between the gesture pointing and the line-of-sight area.

[0092] In some instances, it also includes:

[0093] Based on the cumulative attention duration and spatial matching degree, determine whether the passenger has continuous attention behavior towards the points of interest outside the vehicle;

[0094] If there is continuous attention behavior, extract the cumulative attention duration and spatial matching degree as attention parameters.

[0095] Exemplarily, the judgment of continuous attention behavior towards the points of interest outside the vehicle is achieved through the collaborative analysis of the cumulative attention duration and spatial matching degree. Specifically, according to the preset continuous attention determination logic, when the cumulative attention duration exceeds the minimum continuous threshold and the spatial matching degree reaches the minimum consistency requirement, it is determined that the passenger has continuous attention behavior towards the points of interest outside the vehicle. The determination logic excludes the interference of instantaneous gaze or random gesture actions by jointly analyzing the behavioral characteristics in the time dimension and spatial dimension, ensuring the accuracy of behavior recognition.

[0096] After determining that there is continuous attention behavior, the system extracts the cumulative attention duration and spatial matching degree as behavior parameters. The behavior parameters are recorded in the form of structured data, including the specific value of the cumulative attention duration and the normalized score of the spatial matching degree, providing quantitative input for subsequent trigger condition judgment and point-of-interest information generation. The technical essence of this step is to abstract the original behavior data into operable behavior characteristics and establish a bridge from behavior perception to decision-making analysis.

[0097] In some instances, when the attention parameter meets the preset trigger condition, generate point-of-interest information based on the behavior data and location data, including:

[0098] When the cumulative attention duration exceeds the first preset threshold and the spatial matching degree exceeds the second preset threshold, it is determined that the attention parameter meets the preset trigger condition;

[0099] When the attention parameter meets the preset trigger condition, based on the geographical location of the line-of-sight fixation area, match the preset landmark database to obtain the target type and target geographical location of the target landmark, where the preset landmark database includes multiple geographical locations and the corresponding landmark types for each geographical location;

[0100] Determine the attention intensity of the target landmark according to the weighted result of the cumulative attention duration and spatial matching degree;

[0101] Integrate the target type, target geographical location, and attention intensity into point of interest information.

[0102] Exemplarily, when the cumulative attention duration exceeds the first preset threshold (e.g., 5 seconds) and the spatial matching degree exceeds the second preset threshold (e.g., 0.7), it is determined that the attention parameter meets the preset trigger condition. The determination logic is implemented through a preset threshold comparison module, which compares the cumulative attention duration with the first preset threshold and simultaneously compares the spatial matching degree with the second preset threshold. When both meet the threshold requirements, the point of interest information generation process is triggered to ensure the reliability and accuracy of the recommendation behavior.

[0103] After the trigger condition is met, the system matches the geographical location of the line-of-sight fixation area with a preset landmark database. The preset landmark database takes geofencing technology as the core and stores multiple geographical location points (latitude and longitude coordinates) and their corresponding landmark types (such as shopping malls, gas stations, scenic spots). By performing radius matching on the geographical coordinates of the line-of-sight fixation area (provided by the vehicle-mounted positioning system) with the geographical locations in the database, the target type and precise geographical location of the target landmark are determined, realizing the semantic mapping from visual attention to geographical entities.

[0104] Furthermore, the attention intensity of the target landmark is calculated by weighted fusion of the cumulative attention duration and the spatial matching degree. Specifically, a cumulative attention duration weight coefficient (e.g., 0.6) and a spatial matching degree weight coefficient (e.g., 0.4) are set, and normalization processing is performed on the two (e.g., mapping the duration to the 0-1 range and keeping the matching degree at its original value). The calculation formula is:

[0105] Attention intensity = duration weight × normalized duration + matching degree weight × spatial matching degree

[0106] Integrate the matched target type, target geographical location, and the calculated attention intensity into structured point of interest information. The information is stored in a standardized data format and contains the following fields:

[0107] Target type field: String type, recording the functional attributes of the landmark (such as "shopping mall", "parking lot"); Target geographical location field: Latitude and longitude coordinate pair, accurate to four decimal places; Attention intensity field: Floating-point value, the higher the value, the stronger the interest. The information is stored in JSON or XML format, containing the fields: {"type": "catering", "location": [X, Y], "intensity": 0.85}, providing standardized data input for the generation of subsequent interaction prompt information and path adjustment. This step realizes the technical closed-loop from behavior analysis to navigation decision-making through multi-dimensional data fusion and structured output.

[0108] In some embodiments, generating interactive prompt information including a target landmark based on point-of-interest information includes:

[0109] Selecting an interaction method based on the current vehicle driving state, where the interaction method includes voice broadcast prompt or navigation interface pop-up window;

[0110] Generating interactive content including the target landmark name, target type, target geographical location, and estimated arrival time, where the estimated arrival time is calculated through the real-time navigation path and traffic information;

[0111] According to the interaction method, dynamically rendering the interactive content into executable interactive prompt information.

[0112] Exemplarily, the selection of the interaction method is dynamically determined based on the current vehicle driving state. Vehicle driving parameters (such as vehicle speed, acceleration, steering angle) and road condition complexity data (such as number of lanes, traffic flow) are obtained in real time through in-vehicle sensors, and a preset safe interaction determination model is used for analysis: when the vehicle speed is lower than a preset threshold (such as 40 km / h) and the road condition complexity level is lower than a preset risk level, it is determined as a safe interaction mode, and at this time, a combined interaction method of voice broadcast and interface pop-up window can be selected; if the vehicle speed or road condition complexity exceeds the safe threshold, only the interface pop-up window prompt is enabled to avoid voice interference with the driver's attention and ensure driving safety.

[0113] The generation of the interactive content is achieved by integrating point-of-interest information and real-time navigation data. After the target landmark name, type, and geographical location in the point-of-interest information are extracted from the preset landmark database, they are input into the path planning engine together with the vehicle's current position, real-time navigation path, and traffic information (such as congestion index, traffic light status) to calculate the estimated arrival time (ETA) of the target landmark. Specifically, the ETA calculation formula is:

[0114] ETA = unobstructed path distance / average vehicle speed + congested delay distance / congested average vehicle speed + traffic signal delay

[0115] Where the congested delay distance is the difference between the remaining distance and the unobstructed path distance, and the congested delay distance and traffic signal delay are dynamically corrected through vehicle network real-time data to ensure the accuracy of ETA.

[0116] Based on the selected interaction method, the system calls the corresponding rendering engine to generate executable interaction prompt information. For the navigation interface pop-up window, a graphic rendering engine is used to draw a floating window containing landmark icons, text descriptions, and countdown components; for voice announcements, the TTS (Text-to-Speech) engine is used to convert the interaction content into a voice stream, and the landmark name, type, and ETA are synthesized into a natural language instruction (such as "There is a XX restaurant 500 meters ahead, estimated to arrive in 2 minutes"); the rendering process is implemented based on the UI framework of the in-vehicle operating system to ensure that the interaction content is compatible with the resolution and interaction protocol of the in-vehicle display device.

[0117] In some instances, it further includes:

[0118] Determine the priority of the interaction prompt information according to the attention intensity in the point of interest information;

[0119] Exemplarily, when the same point of interest is recognized multiple times within a continuous time window, the system updates its attention intensity value in real time and dynamically adjusts the priority of the interaction prompt according to the latest calculation result. For example, if the attention intensity of a certain landmark increases from the initial 0.6 to 0.8, its priority increases accordingly, triggering a more prominent prompt method (such as voice announcement instead of pop-up window). This mechanism realizes the smooth update of the intensity value through the sliding time window algorithm to ensure that the priority adjustment is synchronized with the changes in the passenger's behavior.

[0120] When multiple points of interest simultaneously meet the trigger conditions, the system establishes a global priority queue and arranges all candidate landmarks in descending order of attention intensity values. Interaction prompts are generated only for the top N (N≥1) points of interest, and the specific quantity is dynamically determined according to the current vehicle state. For example, when driving at a low speed, the first 3 landmarks are allowed to be prompted simultaneously, and only the landmark with the highest priority is displayed at high speed. This mechanism realizes resource allocation through the priority arbitration module, balancing the information transmission efficiency and driving safety.

[0121] In some embodiments, before generating the interaction prompt information containing the destination landmark based on the point of interest information, it further includes:

[0122] Obtain the current vehicle speed and the current road condition complexity;

[0123] If the current vehicle speed is less than or equal to the preset vehicle speed threshold and the current road condition complexity is less than or equal to the preset risk level, it is determined that the current vehicle driving state is in the safe interaction mode, and the step of generating the interaction prompt information containing the destination landmark based on the point of interest information is executed.

[0124] Exemplarily, vehicle driving state data is obtained in real time through an in-vehicle sensor network, specifically including the current vehicle speed (read through a wheel speed sensor or CAN bus) and the road condition complexity (calculated based on the obstacle density and lane line change frequency output by a camera and radar fusion perception module). The vehicle speed is recorded in kilometers per hour, and the road condition complexity is quantified into a risk level from 0 to 10 (for example: 0 - 3 is low risk, 4 - 7 is medium risk, 8 - 10 is high risk). The data is updated at a fixed frequency (such as 10Hz) and transmitted to the safety determination module through a data bus.

[0125] The safety determination module receives the current vehicle speed and road condition complexity data, and compares them with a preset vehicle speed threshold (such as 40 kilometers per hour) and a preset risk level (such as level 5). If both the current vehicle speed ≤ the preset vehicle speed threshold and the road condition complexity ≤ the preset risk level are satisfied simultaneously; then it is determined that the vehicle is in a safe interaction mode, and a mode enabling signal is generated. If either condition is not satisfied, it is determined as a high-risk driving state, and the interaction prompt process is prohibited from being triggered. The determination result is transmitted in real time through a status flag (such as a boolean value True / False).

[0126] When the safety interaction mode enabling signal is valid, the system activates the interaction prompt generation module to dynamically render interaction content based on the point of interest information. At the same time, the system continuously monitors changes in the vehicle driving state. If it detects that the vehicle speed exceeds the threshold or the road condition complexity is upgraded during the prompt generation process, the current interaction process is immediately aborted, and the pending prompt information is cleared to ensure that the driver's attention is not disturbed.

[0127] The system periodically analyzes historical determination data and user feedback (such as the number of emergency cancellation operations), and dynamically optimizes the preset vehicle speed threshold and risk level. The optimized parameters are downloaded and updated from the cloud through the vehicle networking module and written into the non-volatile memory of the in-vehicle system to achieve the adjustment of the safety strategy.

[0128] In some embodiments, according to the feedback result of the passenger on the interaction prompt information, the navigation path is dynamically adjusted, including:

[0129] Detect the feedback result of the passenger on the interaction prompt information, where the feedback result includes confirming addition, rejecting addition, or ignoring the prompt;

[0130] When the feedback result is confirming addition, the target landmark is added as a dynamic waypoint to the navigation path, and the navigation path is re-planned based on real-time traffic information;

[0131] When the feedback result is rejecting addition or ignoring the prompt, based on the target type of the target landmark, the recommended priority of at least one landmark corresponding to the target type is reduced, and the current navigation path is maintained.

[0132] Exemplarily, the system detects the feedback result of the passenger on the interaction prompt information through the in-vehicle interaction interface (such as a touch screen or a voice recognition module). The feedback results include confirmation to add, rejection to add, and ignoring the prompt.

[0133] Confirmation to add means that the passenger clearly accepts the recommendation by clicking the confirmation button or giving a voice command (such as "add to the route");

[0134] Rejection to add means that the passenger clearly rejects it by pressing the cancel button or giving a negative voice command (such as "don't need");

[0135] Ignoring the prompt means that the passenger does not perform any operation within the preset timeout period (such as 10 seconds).

[0136] The feedback results are stored in the form of event codes (such as 0 - confirmation, 1 - rejection, 2 - ignoring), and are transmitted to the route planning module and the recommendation strategy module through the in-vehicle data bus.

[0137] When the feedback result is confirmation to add, the system inserts the geographical location of the target landmark as a dynamic waypoint into the current navigation route. The route planning module recalculates the optimal route based on real-time traffic information (such as congestion index, accident alert), using the Dijkstra or A* algorithm. For example, if the original route is A → B → C, a new route A → X → B → C is generated after inserting the waypoint X, and the estimated time of arrival (ETA) is updated in real time. The newly planned route highlights the newly added waypoint through the navigation interface, and at the same time, a voice broadcast prompts the route change (such as "XX landmark has been added, the route has been updated").

[0138] When the feedback result is rejection or ignoring, the system reduces the weight value of the landmark of this type in the recommendation priority table according to the type of the target landmark. For example, if the passenger rejects the recommendation of a certain gas station, the priority weight of the "gas station" type is reduced from 0.8 to 0.5. The downgrading strategy is implemented through the following rules:

[0139] Direct rejection: The weight of the target type is immediately reduced by a preset attenuation value;

[0140] Continuous ignoring: If the same type of landmark is continuously ignored (such as 3 times), the weight is further attenuated.

[0141] The updated weight value is stored in the user preference database, which affects the determination of the triggering conditions for subsequent similar points of interest (such as increasing the triggering threshold).

[0142] After the path adjustment or priority downgrading operation is completed, the following mechanism is used to ensure that the changes take effect in real time:

[0143] The updated navigation route is synchronized to the cloud through the vehicle networking module and remains consistent on multiple terminals (such as mobile phones, in-vehicle navigation);

[0144] The downgraded recommended weight immediately takes effect on the behavior recognition module, and the triggering conditions for subsequent similar points of interest are automatically increased (for example, the cumulative duration threshold increases from 5 seconds to 8 seconds).

[0145] In some instances, it also includes:

[0146] The system pre-sets a general recommended priority prediction model based on group behavior data. This model is trained through a large amount of historical data (including feedback results, attention intensity, and landmark types of different passengers) and serves as the benchmark model for all passengers when they first use it. When a new passenger first triggers a point of interest recommendation, the system calls the general model to set the initial thresholds: the first preset threshold (cumulative attention duration) and the second preset threshold (spatial matching degree) are fine-tuned according to the basic attributes of the passenger (such as age, gender) and vehicle type respectively to ensure the universality of the initial recommendation strategy.

[0147] During the passenger's use, the system continuously collects their behavior data, point of interest information, and feedback results, and performs local feature extraction and data desensitization processing through the in-vehicle computing unit. After the data is packaged and encrypted by time window, it is uploaded to the cloud server through the vehicle networking module. The cloud performs incremental training on the passenger's individual data through a distributed machine learning framework (such as TensorFlow Serving), updates the parameters of the recommended priority prediction model, and generates a personalized model. The model is stored in the cloud database indexed by the passenger's unique identifier (such as vehicle VIN code + user account) to enable cross-terminal invocation.

[0148] When the passenger uses the system again, the in-vehicle terminal obtains the personalized model identifier through identity recognition (such as face authentication or account login), and downloads the corresponding model parameters and historical preference data from the cloud. When identifying continuous attention behavior, the system inputs the real-time behavior data (current cumulative duration, spatial matching degree) into the personalized model, and outputs the dynamically adjusted first preset threshold and second preset threshold. For example: if the model determines that the passenger's confirmation rate for points of interest of the "gas station" type is as high as 80%, the cumulative duration threshold for this type of landmark is reduced from 5 seconds to 3 seconds, and the spatial matching degree threshold is reduced from 0.7 to 0.6, thereby improving the recommendation sensitivity; conversely, for the "billboard" type of landmark that is frequently rejected, the thresholds are increased to 8 seconds and 0.8 to reduce the false trigger rate.

[0149] The system realizes continuous optimization of the model through federated learning technology. The in-vehicle terminal calculates the model gradient using local new data, encrypts it and uploads it to the cloud for aggregation and update to avoid leakage of raw data. At the same time, the differential privacy algorithm is used to add noise to the training data to ensure that the behavior patterns of individual passengers cannot be inferred. When a significant change in passenger preferences is detected, a forced model update instruction is triggered, and the cloud retrains and distributes new parameters. This mechanism realizes real-time synchronization of the threshold strategy and user preferences on the premise of ensuring data security, forming a closed-loop optimization link of "collection - training - adaptation".

[0150] The mechanism of the embodiment of this application realizes the optimization of the recommendation strategy through the combination of in-depth mining of historical feedback data and machine learning algorithms. A complete intelligent closed-loop including data collection, feature engineering, model training, and threshold adjustment is constructed. By dynamically balancing the generalization ability of the global model and the accuracy of the personalized model, while ensuring the robustness of the service, the universality of the navigation recommendation strategy is realized, and user stickiness and service satisfaction are improved.

[0151] Please refer to Figure 2 , which is a schematic structural diagram of a navigation device driven by passenger behavior provided by the embodiment of this application, including:

[0152] A behavior data acquisition unit 21 for acquiring the behavior data of passengers;

[0153] A focus parameter extraction unit 22, based on the behavior data, determines a focus parameter for characterizing the continuous focus behavior of passengers on points of interest outside the vehicle, where the focus parameter includes the cumulative focus duration and the spatial matching degree;

[0154] An interest information generation unit 23 for generating point-of-interest information based on the behavior data and location data when the focus parameter meets a preset trigger condition, where the point-of-interest information includes the target type, target geographical location, and focus intensity of the target landmark;

[0155] An interaction information prompt unit 24 generates an interaction prompt information including the target landmark based on the point-of-interest information;

[0156] A navigation path optimization unit 25 for dynamically adjusting the navigation path according to the feedback result of the passenger on the interaction prompt information.

[0157] Please refer to Figure 3 , the embodiment of this application also provides an electronic device 300, including a memory 310, a processor 320, and a computer program 311 stored on the memory 310 and operable on the processor. When the processor 320 executes the computer program 311, it implements the steps of any method of navigation driven by passenger behavior.

[0158] Since the electronic device introduced in this embodiment is the device adopted by a passenger behavior-driven navigation device in an embodiment of the present application, based on the method introduced in the embodiment of the present application, those skilled in the art can understand the specific implementation manners and various variations of the electronic device in this embodiment. Therefore, the specific implementation of how this electronic device implements the method in the embodiment of the present application will not be described in detail here. As long as the device adopted by those skilled in the art to implement the method in the embodiment of the present application belongs to the scope protected by the present application.

[0159] In the specific implementation process, when the computer program 311 is executed by the processor, it can implement any implementation manner in the corresponding embodiment of the first aspect.

[0160] It should be noted that in the above embodiments, the descriptions of each embodiment have their own focuses. For the parts not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0161] Those skilled in the art should understand that the embodiments of the present application can provide methods, systems, or computer program products. Therefore, the present application can take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present application can take the form of a computer program product implemented on one or more computer-readable storage media containing computer-readable program code.

[0162] The present application is described with reference to the flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each flow and / or block in the flowcharts and / or block diagrams can be implemented by computer program instructions, and the combination of the flows and / or blocks in the flowcharts and / or block diagrams can also be implemented by computer program instructions. These computer program instructions can be provided to the processors of general-purpose computers, special-purpose computers, embedded computers, or other programmable data processing devices to generate a machine, so that the instructions executed by the processors of the computer or other programmable data processing devices generate means for implementing the specified functions in Figure 1 one or more flows and / or Figure 1 blocks or multiple blocks.

[0163] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer-readable memory generate a manufactured article including instruction means, and the instruction means implements the specified functions in Figure 1 one or more flows and / or Figure 1 blocks or multiple blocks.

[0164] These computer program instructions can also be loaded onto a computer or other programmable data processing device, so that a series of operation steps are executed on the computer or other programmable device to generate a computer-implemented process. Thus, the instructions executed on the computer or other programmable device provide steps for implementing the functions specified in one process Figure 1 one process or multiple processes and / or blocks Figure 1 steps for implementing the functions specified in one block or multiple blocks.

[0165] An embodiment of the present application also provides a computer program product, which includes computer software instructions. When the computer software instructions run on a processing device, the processing device is caused to execute Figure 1 the process of a navigation method driven by passenger behavior in the corresponding embodiment.

[0166] The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, the process or function according to the embodiment of the present application is generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center in a wired or wireless manner. The computer-readable storage medium can be any available medium that a computer can store or a data storage device such as a server or data center that includes one or more integrated available media. The available medium can be a magnetic medium, an optical medium, or a semiconductor medium, etc.

[0167] Those skilled in the art can clearly understand that for the convenience and brevity of description, the specific working processes of the systems, devices, and units described above can refer to the corresponding processes in the foregoing method embodiments and will not be elaborated here.

[0168] In several embodiments provided by the present application, it should be understood that the disclosed devices, apparatuses, and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the division of units is only a logical function division, and there can be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the displayed or discussed mutual coupling, direct coupling, or communication connection can be through some interfaces, and the indirect coupling or communication connection of devices or units can be in an electrical, mechanical, or other form.

[0169] The unit described as a separation component may or may not be physically separated. The component displayed as a unit may or may not be a physical unit, that is, it may be located in one place or distributed across multiple network units. Some or all of these units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0170] In addition, in each embodiment of the present application, each functional unit can be integrated in a processing unit, or each unit can exist physically alone, or two or more units can be integrated in one unit. The above-mentioned integrated unit can be implemented in the form of a hardware and / or software functional unit.

[0171] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application, in essence, or the part that contributes to the prior art, or all or part of this technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for causing a computer device to execute all or part of the steps of the methods in each embodiment of the present application.

[0172] The above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them; although the present application has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions described in the foregoing embodiments, or perform equivalent replacements for some of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of each embodiment of the present application.

[0173] Although the preferred embodiments of this specification have been described, those skilled in the art can make additional changes and modifications once they learn the basic creative concept. Therefore, the appended claims are intended to be construed as including the preferred embodiments as well as all changes and modifications falling within the scope of this specification.

[0174] Obviously, those skilled in the art can make various changes and deformations to this specification without departing from the spirit and scope of this specification. Thus, if these modifications and deformations of this specification fall within the scope of the claims of this specification and their equivalent technologies, this specification is also intended to include these changes and deformations.

Claims

1. A passenger behavior driven navigation method, characterized in that: The method comprises: Obtain passenger behavior data; Based on the behavior data, determining an attention parameter for characterizing the passenger's continuous attention behavior to the point of interest outside the vehicle, wherein the attention parameter includes a cumulative attention duration and a spatial matching degree; When the attention parameter meets the preset trigger condition, generating point of interest information based on the behavior data and the location data, wherein the point of interest information includes the target type, target geographical location and attention intensity of the target landmark; generating interactive prompt information including a target landmark based on the point of interest information; The navigation path is dynamically adjusted according to the feedback result of the passenger to the interactive prompt information.

2. The method according to claim 1, characterized in that The obtaining of passenger behavior data includes: Collecting facial and head image data of the passenger, and extracting the gaze direction and head posture from the image data using a computer vision algorithm; Capture the spatial position and motion trajectory of the passenger's hand, and generate gesture data including gesture movements and gesture trajectories; The sight direction, the head posture and the gesture data are timestamped and synchronized, and the synchronized data are denoised and fused using a preset filtering algorithm to obtain behavior data.

3. The method according to claim 1, characterized in that The determining, based on the behavior data, an attention parameter for characterizing the passenger's continuous attention behavior to the point of interest outside the vehicle includes: Extracting features of the sight direction and head posture in the behavior data by using a convolutional neural network to determine the passenger's sight fixation area and the cumulative attention duration of the sight fixation area; Analyzing the gesture actions and gesture trajectories in the behavior data by means of a trajectory tracking algorithm to generate spatial trajectory data of the hand pointing direction; The matching degree between the gaze area and the spatial trajectory data is calculated to obtain the spatial matching degree between the gaze area and the hand pointing direction.

4. The method according to claim 3, characterized in that When the attention parameter satisfies a preset trigger condition, generating the point of interest information based on the behavior data and the location data includes: When the accumulated attention duration exceeds a first preset threshold and the spatial matching degree exceeds a second preset threshold, it is determined that the attention degree parameter meets a preset trigger condition; When the attention parameter meets the preset trigger condition, based on the geographical location of the gaze area, a preset landmark database is matched to obtain a target type and a target geographical location of the target landmark, wherein the preset landmark database includes multiple geographical locations and a landmark type corresponding to each geographical location; Determining the attention intensity of the target landmark according to a weighted calculation result of the accumulated attention duration and the spatial matching degree; The target type, the target geographic location and the attention intensity are integrated into point of interest information.

5. The method according to claim 1, characterized in that The generating of interactive prompt information including a target landmark based on the point of interest information includes: Based on the current driving state of the vehicle, select an interaction mode, wherein the interaction mode includes a voice broadcast prompt or a navigation interface pop-up window; Generate interactive content including a target landmark name, a target type, a target geographic location, and an estimated time of arrival, wherein the estimated time of arrival is calculated using real-time navigation path and traffic information; According to the interaction mode, the interaction content is dynamically rendered into executable interaction prompt information.

6. The method according to claim 1, characterized in that Before generating the interactive prompt information including the target landmark based on the interest point information, the method further includes: Get the current speed of the vehicle and the complexity of the current road conditions; If the current vehicle speed is less than or equal to the preset vehicle speed threshold and the current road condition complexity is less than or equal to the preset risk level, it is determined that the current vehicle driving state is in a safe interaction mode, and the step of generating interactive prompt information containing the target landmark based on the point of interest information is executed.

7. The method according to claim 1, characterized in that The dynamically adjusting the navigation path according to the feedback result of the passenger on the interactive prompt information includes: Detecting the passenger's feedback result on the interactive prompt information, wherein the feedback result includes confirming to add, rejecting to add, or ignoring the prompt; When the feedback result is confirmation to add, the target landmark is added to the navigation path as a dynamic waypoint, and the navigation path is replanned based on the real-time traffic information; When the feedback result is to refuse to add or ignore the prompt, based on the target type of the target landmark, the recommendation priority of at least one landmark corresponding to the target type is reduced, and the current navigation path is maintained.

8. A navigation device driven by passenger behavior, characterized in that: The device comprises: A behavior data acquisition unit, used to acquire the behavior data of passengers; An attention parameter extraction unit, based on the behavior data, determines an attention parameter for characterizing the passenger's continuous attention behavior to the point of interest outside the vehicle, wherein the attention parameter includes a cumulative attention duration and a spatial matching degree; An interest information generating unit, configured to generate interest point information based on the behavior data and the location data when the attention parameter satisfies a preset trigger condition, wherein the interest point information includes a target type, a target geographical location, and an attention intensity of a target landmark; An interactive information prompting unit, which generates interactive prompting information including a target landmark based on the point of interest information; The navigation path optimization unit is used to dynamically adjust the navigation path according to the feedback result of the passenger to the interactive prompt information.

9. An electronic device, comprising: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor is used to implement the steps of the passenger behavior-driven navigation method as described in any one of claims 1 to 7 when executing the computer program stored in the memory.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the passenger behavior-driven navigation method according to any one of claims 1 to 7 is implemented.

Images