Car guide method based on mobile phone sensor

Abstract

The application is a car guide method based on mobile phone sensor, which relates to a car guide method based on mobile phone sensor, which presets a program in the mobile phone, the program communicates with the server, and the server portraits the user. During the implementation, the program obtains the permission of the mobile phone sensor, obtains the sensor value; the mobile phone synchronizes the server locally, and realizes the front POI prediction under the high-speed mobile scene; when the vehicle approaches the POI, the program explains the POI through the mobile phone; when explaining a certain POI, if another POI also meets the broadcast condition, the system will execute the broadcast according to the user portrait; the program obtains the final destination and path of the mobile phone navigation, the server calls the related POI explanation of the destination and the POI along the way, and the server pushes the POI explanation to the program in real time through the adaptation of the user portrait, and pre-stores. Thus, the speed, direction and position of the vehicle can be captured and processed in real time, and the possible POI in front can be predicted accordingly.

Description

Technical Field

[0001] This invention relates to a vehicle-mounted tour guide method, and more particularly to a vehicle-mounted tour guide method based on mobile phone sensors. Background Technology

[0002] Self-driving tours have become increasingly popular with travelers in recent years. Traditional live tour guides are limited to fixed routes, with limited depth and breadth, and often cannot provide drivers with real-time, dynamic commentary along the way. While electronic tour guides offer convenient commentary services for walking tourists in scenic areas, there are currently no electronic tour guide products specifically designed for drivers, causing them to often miss many interesting stories and historical backgrounds during their trips.

[0003] Regarding existing technologies, CN201110382767 - a mobile navigation method that simultaneously considers intersection announcements and POI explanations, while providing some announcements and navigation, cannot achieve application in suitable scenarios and can only perform simple, step-by-step playback.

[0004] CN201810060126 - A location-based augmented reality interactive tour guide system and method for tourist attractions, mainly based on GPS voice narration within the scenic area. However, for vehicles, their travel speed is relatively high, and the single GPS positioning and corresponding data communication speed cannot achieve real-time updates. Furthermore, the vehicle's driving environment needs to consider more environmental factors.

[0005] CN202310706986 - A navigation and guidance system based on AR and LBS technologies. Its implementation is relatively simple; it cannot effectively predict Points of Interest (POIs) based on the user's path and destination, and can only update passively.

[0006] In view of the above-mentioned shortcomings, the designer has actively researched and innovated in order to create a car tour guide method based on mobile phone sensors, making it more valuable for industrial applications. Summary of the Invention

[0007] To address the aforementioned technical problems, the purpose of this invention is to provide a car navigation method based on mobile phone sensors.

[0008] The present invention provides a car navigation method based on mobile phone sensors, which includes the following steps: pre-setting a program in the mobile phone, wherein the program communicates with a server.

[0009] The server creates a user profile.

[0010] Step 1: The program obtains permission to access the phone's sensors, acquires sensor values, and dynamically adjusts the sampling frequency.

[0011] Step 2: The server predicts the points of interest (POIs) ahead based on the destination, path, and real-time sensors, and sends the N most matching items to the mobile phone.

[0012] Step 3: When the vehicle approaches the POI, the program uses real-time sensor data to filter and sort the content sent by the server, and plays the POI explanation via the mobile phone; when explaining a POI, if another POI explanation also meets the broadcast conditions, the system will determine and execute the playback based on the user profile.

[0013] Step four: The content already played during this trip, the playback duration, the remaining itinerary, and the attractions visited will all affect the prediction and content selection for subsequent POI explanations.

[0014] Furthermore, in the aforementioned car navigation method based on mobile phone sensors, the sensor values ​​include vehicle speed, position, and direction, and the program dynamically adjusts the data reading frequency based on the current vehicle speed and acceleration status.

[0015] The dynamic adjustment involves setting the sampling interval to T, where T ranges from 4 to 12 seconds. If the device is currently broadcasting an explanation, the predicted time T is adjusted to one-third of the remaining playback duration. If the cumulative change in the vehicle's direction of travel exceeds a threshold, the sampling interval is reassessed. If the next broadcast point is approaching, the real-time sampling interval reverts to 1.5 seconds. During implementation, a different radius is set for each POI to facilitate a more optimized playback interval. The threshold used can be the average vehicle speed, ranging from 30 to 90 kilometers per hour.

[0016] Furthermore, in the aforementioned car navigation method based on mobile phone sensors, step two involves using a sampling algorithm to predict in real time the points of interest (POIs) that the vehicle may pass ahead. A POI refers to any object containing cultural descriptions, historical information, stories, anecdotes, etc., and can include various observable landscapes, such as bridges, buildings, rivers, lakes, mountains, and archways. The sampling algorithm includes:

[0017] Step A, position and direction prediction: Based on the sensor data obtained by the mobile phone in the high-speed moving car, predict the car's position and driving direction within the next 5 minutes. If in navigation mode, the prediction is made directly through the navigation trajectory; if in non-navigation mode, the prediction is made based on the current position, direction, speed, and road information.

[0018] Step B involves scene recognition and keyword extraction. Based on the information obtained in Step A, an interactive query between the Geographic Information System (GIS) and the database is used to compare the predicted location with the geographic information in the database. Landscape features containing different locations are then filtered out, and pre-defined keywords for the corresponding POIs are extracted. Simultaneously, during implementation, this invention considers the vehicle's driving direction in driving scenarios. Only POIs within a 120-degree field of view in the vehicle's forward direction are considered observable POIs. This avoids excessive redundant POS data collection and optimizes for relevant POIs along the driving route. This differs from current conventional low-speed or walking scenarios.

[0019] Step C, User Interest Fusion: Read the user's historical interest data and scene keywords, and fuse them with the POI's preset keywords to obtain a keyword set;

[0020] Step D: Content matching. The keyword set is matched against the similarity of the speech content database to find the content with the highest matching degree.

[0021] Step E: Send the content to the program on the phone.

[0022] Furthermore, in the aforementioned car tour guide method based on mobile phone sensors, in step three, when the vehicle approaches the POI, the program uses a playback prediction algorithm to determine whether to play the narration content, based on the vehicle speed, distance from the POI, driving direction, and duration of the narration content.

[0023] The playback algorithm is as follows.

[0024] Step a, define the parameters. Let the current vehicle speed be V, the voice point be i, the distance from the current position to the voice point i be Li, the playback duration of the voice point i be T_len_i, the remaining duration of the currently playing voice be T_curr, the idle time between voice playbacks be T_idle, and the fatigue idle time be T_gap.

[0025] Step b: Set playback decision rules;

[0026] If a voice is currently playing and cannot be completed before reaching the next voice point (i.e., T_curr>Li / V), then ignore that voice point.

[0027] Calculate the driving time between two consecutive voice points T_drive=(Li+1-Li) / V. If the playback duration of a certain voice point exceeds the driving time between two consecutive voice points, that is, T_len_i+T_gap>T_drive, then one of the two voice points needs to be ignored.

[0028] If T_idle > 10 minutes, ignore the next speech point; otherwise, ignore the speech point with shorter duration. If T_curr > Li / V, ignore the current speech point.

[0029] If T_len_i > T_drive, then check if T_idle is greater than 10 minutes. If it is, ignore point i+1. If not, compare T_len_i and T_len_i+1, and ignore the shorter duration.

[0030] Furthermore, in the aforementioned car tour guide method based on mobile phone sensors, step three, the playback judgment includes: interrupting the current explanation and playing a new POI explanation; adding the new POI explanation to the queue and playing it in sequence after the previous POI explanation is completed; and abandoning the playback of the new POI explanation.

[0031] Furthermore, in the aforementioned car tour guide method based on mobile phone sensors, the server generates a user profile based on the user's historical behavior and interests, filters out the tour content that the user is interested in, and caches the tour content in advance in the local program of the mobile phone.

[0032] Furthermore, in the aforementioned car tour guide method based on mobile phone sensors, the historical behavior involves the program collecting the user's previously intended audio content, generating historical behavior tags, the server receiving the historical behavior tags, filtering audio content based on the historical behavior tags, and noting them as historical behavior content.

[0033] The interest is defined as follows: the program collects the explanation content that the user has previously actively requested, generates interest tags, the server receives the interest tags, filters the explanation content according to the interest tags, and notes the content as interest content.

[0034] The server compares historical behavior content and interest content, and prioritizes sending overlapping content to the program for pre-caching.

[0035] Furthermore, in the aforementioned car tour guide method based on mobile phone sensors, drivers are prone to fatigue from listening to knowledge-based audio for extended periods. Therefore, the program adjusts the playback interval based on the user's historical listening time during this trip. The playback interval is such that the length of a single audio clip does not exceed 8 minutes, the total listening time within 30 minutes does not exceed 17 minutes, and the total listening time within 60 minutes does not exceed 36 minutes.

[0036] By employing the above-described approach, the present invention's car navigation method based on mobile phone sensors has at least the following advantages:

[0037] 1. In high-speed driving scenarios, the system captures and processes vehicle speed, direction, and position information in real time, and uses algorithms to predict potential Points of Interest (POIs) ahead. By dynamically adjusting the data reading frequency, it ensures both data accuracy and effective power saving. It can comprehensively determine multiple parameters for playback conditions and manage playback decisions when multiple POI conditions are met.

[0038] 2. It can be effectively applied to the self-driving tourism sector, especially in in-vehicle applications, filling a gap in the industry with its real-time tour guide system designed for drivers. Furthermore, it optimizes the user experience, perfectly integrating tour guide content with the driving scenario.

[0039] 3. This invention expands the scope of tour guide explanations to cover all national and provincial highways across the country, breaking the previous limitations of explanations within scenic areas. It truly enables explanations to be given wherever the tour goes, broadening the boundaries of POIs (Points of Interest) to any landscape with a story, history, and cultural background. This not only presents users with richer and more diverse travel stories but also makes every trip a deep cultural and historical exploration.

[0040] The above description is merely an overview of the technical solution of the present invention. In order to better understand the technical means of the present invention and to implement it in accordance with the contents of the specification, the preferred embodiments of the present invention are described in detail below with reference to the accompanying drawings. Attached Figure Description

[0041] Figure 1 This is a schematic diagram illustrating the implementation process of a car navigation method based on mobile phone sensors. Detailed Implementation

[0042] The specific embodiments of the present invention will be described in further detail below with reference to the accompanying drawings and examples. The following examples are for illustrative purposes only and are not intended to limit the scope of the invention.

[0043] like Figure 1 The car navigation method based on mobile phone sensors is unique in that it includes the following steps: a program is pre-installed on the mobile phone, the program communicates with the server, and the server creates a user profile.

[0044] Step one: The program obtains permission to access the phone's sensors and acquires sensor data. Specifically, the sensor data includes vehicle speed, position, and direction. The program dynamically adjusts the data reading frequency based on the current vehicle speed and acceleration status.

[0045] For ease of implementation, the sampling interval is dynamically adjusted as follows: The initial sampling interval is set to T, with T ranging from 4 to 12 seconds (the specific value can be a function of the average vehicle speed over the past 10 seconds). If the device is currently broadcasting an explanation, the prediction time T is adjusted to 1 / 3 of the remaining playback time. If the cumulative change in the vehicle's direction of travel exceeds a threshold, the sampling interval is reassessed. For ease of implementation, this invention uses the average vehicle speed as the threshold, ranging from 30 to 90 kilometers per hour. If the next broadcast point is approaching, the sampling interval reverts to 1.5 seconds of real-time sampling. This ensures accurate data acquisition while conserving mobile phone power.

[0046] Step two involves predicting potential Points of Interest (POIs) ahead, either locally on the mobile phone or remotely on a server. During implementation, a sampling algorithm is used to predict POIs that the vehicle may pass by in real time. POIs include cultural descriptions, historical content, stories, and scenic anecdotes. Furthermore, any landscape with a story, history, and cultural background can be included in the POIs implemented in this invention.

[0047] The sampling algorithm used in this invention includes the following steps:

[0048] Step A, Position and Direction Prediction: Based on sensor data acquired by the mobile phone in a high-speed moving car, the position and direction of travel for the next 5 minutes are predicted. If in navigation mode, the prediction is made directly from the navigation trajectory; if in non-navigation mode, the prediction is made using the current position, direction, speed, and road information. Considering the vehicle's speed, only a 1-minute prediction is performed in non-navigation mode to avoid abnormal overlap of POIs.

[0049] Step B, scene recognition and keyword extraction, uses the information obtained in Step A to perform interactive queries between a Geographic Information System (GIS) and a database. The predicted location is compared with the geographic information in the database to filter out landscape features containing different locations, and the corresponding POI preset keywords are extracted. Landscape features used may include mountains, water bodies, buildings, tourist attractions, etc. POI preset keywords may include name, year, notable figures, history, type, etc.

[0050] Step C, User Interest Fusion: Read the user's historical interest data and scene keywords, and fuse them with the POI's preset keywords to obtain a keyword set.

[0051] Step D: Content matching. The keyword set is used to perform similarity matching in the voice content database to find the top-matching entries. Considering the convenience of having alternatives and options, six entries are typically selected.

[0052] Step E: Send the content to the program on the phone.

[0053] During implementation, the following parameters can be set: P_current, which is the current user's GPS location, represented as a two-dimensional vector (x, y); V_current, which is the current speed; V_avg_10, which is the average speed over the past 10 seconds; and θ, which is the angle between the current driving direction and true north, expressed in degrees or radians.

[0054] Then, a trajectory calculation method is provided, as follows:

[0055] Determine the driving direction vector. Assuming that due north is the positive y-axis direction, the driving direction vector D can be expressed as: D = (sin(θ), cos(θ)).

[0056] Calculate the displacement within a unit time T, denoted as displacement. This can be calculated based on velocity and time. Assuming the velocity has been relatively stable over the past 10 seconds, the displacement is estimated using V_avg_10. The displacement calculation formula is:

[0057] S = V_avg_10 × T × D.

[0058] Obtain the position after a unit time T, and get the new position P_new, which is the sum of the current position and the displacement vector: P_new = P_current + S.

[0059] The following methods can be used to estimate user interest integration.

[0060] There is a user feature vector U and a scene keyword feature vector S. The user feature vector U is constructed based on the user's historical interests and behaviors, where each dimension represents the degree of interest in a certain topic or category. The scene keyword feature vector S is constructed based on the keyword extraction results, where each dimension represents the importance of the keyword.

[0061] The score for user interest fusion is obtained by calculating the cosine similarity of the vectors of two users, U and S:

[0062]

[0063] This enables efficient database querying and accurate modeling of user data. Especially in the integration of user interests, understanding users' historical behavior and preferences facilitates the generation of targeted recommended content during application.

[0064] Step 3: When the vehicle approaches the POI, the program provides POI explanation via mobile phone; while explaining one POI, if another POI explanation also meets the broadcast conditions, the system will determine and execute the playback based on the user profile.

[0065] Assuming that after the previous steps, the app on the phone obtains a list of the best matching voice points from the server, namely a, b, c, d, e, and f, sorted from nearest to furthest. Assuming these contents are similar in terms of user interest, the app only needs to make playback selection decisions based on the completeness and timeliness of the content.

[0066] Specifically, during the implementation of this invention, when a vehicle approaches a POI, the program uses a playback prediction algorithm to determine whether to play the narration content. The determination is based on the vehicle speed, distance from the POI, driving direction, and duration of the narration content.

[0067] Based on actual implementation, the playback algorithm is as follows:

[0068] Step a: Define the parameters. Let the current vehicle speed be V, the voice point be i, the distance from the current position to voice point i be Li, the playback duration of voice point i be T_len_i, the remaining duration of the currently playing voice be T_curr, the idle time between voice playbacks be T_idle, and the fatigue idle time be T_gap. The fatigue idle time is set primarily to prevent user fatigue and boredom.

[0069] Step b: Set playback decision rules.

[0070] If a voice is currently playing and cannot be completed before reaching the next voice point (i.e., T_curr>Li / V), then ignore that voice point.

[0071] Calculate the travel time between two consecutive voice points, T_drive = (Li+1-Li) / V. If the playback duration of a certain voice point exceeds the travel time between two consecutive voice points, i.e., T_len_i + T_gap > T_drive, then one of the two voice points needs to be ignored.

[0072] If T_idle > 10 minutes, ignore the next speech point; otherwise, ignore the speech point with shorter duration. If T_curr > Li / V, ignore the current speech point.

[0073] If T_len_i > T_drive, then check if T_idle is greater than 10 minutes. If it is, ignore point i+1. If not, compare T_len_i and T_len_i+1, and ignore the shorter duration.

[0074] Furthermore, the playback decision-making process includes: interrupting the current explanation to play a new POI explanation; queuing the new POI explanation and playing it in sequence after the previous POI explanation has finished; and abandoning the playback of the new POI explanation. In this way, the most suitable playback strategy can be provided according to different application environments.

[0075] Step four: The program obtains the final destination from the phone's navigation. The server retrieves relevant Points of Interest (POIs) for the destination and, based on the user profile, pushes the destination POI explanations to the program for pre-storage. This ensures that the user receives the most relevant explanations before arriving at their destination.

[0076] Furthermore, this invention employs a server to generate user profiles based on users' historical behavior and interests, filtering out content that users are interested in and pre-caching this content locally on the user's phone. Regarding historical behavior, the program collects previously requested content from users, generating historical behavior tags. The server receives these tags and filters content based on them, noting them as "historical behavior content." Similarly, for interest tagging, the program collects previously requested content from users, generating interest tags. The server receives these tags and filters content based on them, noting them as "interest content." Thus, the server compares historical behavior content and interest content, prioritizing sending overlapping content to the program for pre-caching.

[0077] Furthermore, listening to educational audio for extended periods can be tiring and cause driver fatigue. Therefore, the program adjusts the playback interval based on the user's historical listening time during the current drive to prevent fatigue. Specifically, a single audio clip cannot exceed 8 minutes, the total listening time within a 30-minute period cannot exceed 17 minutes, and the total listening time within a 60-minute period cannot exceed 36 minutes. This prevents continuous playback from causing driver fatigue. Of course, in actual implementation, other pricing options can be used, such as total durations of 120 minutes or 180 minutes.

[0078] During implementation, the following parameters can be defined: L represents the playback duration of the new audio, T_current represents the current time, and Played represents a list containing the start time and duration of already played audio.

[0079] The function TotalPlayedLength(T_start, T_end, Played) is used to calculate the total playback duration within the time interval [T_start, T_end]. For each time window W (30 minutes, 60 minutes, 120 minutes), the playback duration P_W = TotalPlayedLength(T_current - W, T_current, Played) is calculated.

[0080] Simultaneously, the remaining playable time within the time window can be determined. For each time window W and the corresponding duration limit L_W (17 minutes, 36 minutes, and 67 minutes respectively), the remaining playable time R_W = L_W - P_W is calculated.

[0081] Next, the suggested playback delay time can be calculated, and the earliest possible start time T_earliest for new speech can be calculated to ensure that L and R_W are compatible for all W. The specific calculation formula is as follows: T_earliest = max_W(T_current + (L - R_W)), where W iterates through all time windows. The delay time T_gap is T_earliest minus T_current. If T_earliest is less than or equal to T_current, no delay is needed, and T_gap = 0.

[0082] In simple terms, the implementation process of this invention can be as follows: Figure 1 As shown.

[0083] The working principle of this invention is as follows:

[0084] The user sets navigation from their starting point to a scenic area in the mountains. The app on the phone first loads and generates a 10-minute POI (Points of Interest) playlist based on the destination scenic area and POIs along the route. Then, following the user's actual driving path, the app, through collaboration between the phone and the server, retrieves the most relevant POI explanations based on the current scenario and user profile, filtering and playing them according to the vehicle's current location and speed. When the vehicle is traveling at high speed, some content is automatically ignored; when the vehicle is traveling slowly (e.g., in traffic jams), more content is played based on the navigation time, ensuring a rich and varied travel experience. Because informative audio can cause fatigue, the app also adjusts fatigue parameters and controls the intervals and pace of the explanations based on different playback durations.

[0085] During this period, users can adjust the playback strategy of the POI explanation in the program at any time to suit their own interests.

[0086] When the distance to the destination is 2 to 5 kilometers, the program will prioritize playing the last POI explanation.

[0087] A new round of assessments will be initiated when the previous audio is about to finish playing or when the device is approaching another Point of Interest (POI).

[0088] As can be seen from the above description and the accompanying drawings, the present invention has the following advantages:

[0089] 1. In high-speed driving scenarios, the system captures and processes vehicle speed, direction, and position information in real time, and uses algorithms to predict potential Points of Interest (POIs) ahead. By dynamically adjusting the data reading frequency, it ensures both data accuracy and effective power saving. It can comprehensively determine multiple parameters for playback conditions and manage playback decisions when multiple POI conditions are met.

[0090] 2. It can be effectively applied to the self-driving tourism sector, especially in in-vehicle applications, filling a gap in the industry with its real-time tour guide system designed for drivers. Furthermore, it optimizes the user experience, perfectly integrating tour guide content with the driving scenario.

[0091] 3. This invention expands the scope of tour guide explanations to cover all national and provincial highways across the country, breaking the previous limitations of explanations within scenic areas and truly enabling explanations to be provided wherever the tour goes. Furthermore, the boundaries of the POIs covered are broadened to any landscape with a story, history, and cultural background. This not only presents users with richer and more diverse travel stories but also makes every trip a deep cultural and historical exploration.

[0092] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the technical principles of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A car navigation method based on mobile phone sensors, characterized in that... The process includes the following steps: Pre-installing a program on the phone, which communicates with the server. The server creates a user profile. Step 1: The program obtains permission to access the phone's sensors, acquires sensor values, and dynamically adjusts the sampling frequency. Step 2: The server predicts the points of interest (POIs) ahead based on the destination, path, and real-time sensors, and sends the N most matching items to the mobile phone. Step 3: When the vehicle approaches the POI, the program uses real-time sensor data to filter and sort the content sent by the server, and plays the POI explanation via the mobile phone; when explaining a POI, if another POI explanation also meets the broadcast conditions, the system will determine and execute the playback based on the user profile. Step 4: The content already played during this trip, the playback duration, the remaining itinerary, and the attractions visited will all affect the prediction and content selection for subsequent POI explanations. The sensor values ​​include vehicle speed, position, and direction. The program dynamically adjusts the data reading frequency based on the current vehicle speed and acceleration status. The dynamic adjustment is as follows: the sampling interval is set to T, which takes values ​​from 4 to 12 seconds; if the device is broadcasting an explanation, the prediction time T is adjusted to 1 / 3 of the remaining playback time; if the cumulative change in the angle of the vehicle's driving direction is greater than the threshold, the sampling interval is re-evaluated. If the next broadcast point is approached, the real-time sampling will resume at 1.5 seconds. The threshold is the average vehicle speed, and the value ranges from 30 to 90 kilometers per hour. In step three, when the vehicle approaches the POI, the program uses a playback prediction algorithm to determine whether to play the explanation content. The determination is based on the vehicle speed, distance from the POI, driving direction, and duration of the explanation content. The playback prediction algorithm is as follows: Step a, define the parameters. Let the current vehicle speed be V, the voice point be i, the distance from the current position to the voice point i be Li, the playback duration of the voice point i be T_len_i, the remaining duration of the currently playing voice be T_curr, the idle time between voice playbacks be T_idle, and the fatigue idle time be T_gap. Step b: Set playback decision rules; If a voice is currently playing and cannot be completed before reaching the next voice point (i.e., T_curr > Li / V), then ignore that voice point. Calculate the driving time between two consecutive voice points T_drive = (Li+1- Li) / V. If the playback duration of a certain voice point exceeds the driving time between two consecutive voice points, i.e., T_len_i+T_gap>T_drive, then one of the two voice points needs to be ignored. If T_idle > 10 minutes, ignore the next speech point; otherwise, ignore the speech point with shorter duration. If T_curr > Li / V, ignore the current speech point. If T_len_i > T_drive, then check if T_idle is greater than 10 minutes. If it is, ignore point i+1. If not, compare T_len_i and T_len_i+1, and ignore the shorter time.

2. The car navigation method based on mobile phone sensors according to claim 1, characterized in that: In step two, a sampling algorithm is used to predict in real time the points of interest (POIs) that the vehicle may pass in front of it. A POI refers to any location containing a cultural description, historical content, story content, anecdotes, or observable landscape features. The sampling algorithm includes: Step A, position and direction prediction: Based on the sensor data obtained by the mobile phone in the high-speed moving car, predict all positions and driving directions of the car in the next 5 minutes. If in navigation mode, the prediction is made directly through the navigation trajectory; if in non-navigation mode, the prediction is made through the current position, direction, speed and road information. Step B, scene recognition and keyword extraction: Based on the information obtained in Step A, the predicted location is compared with the geographic information in the database using an interactive query between the geographic information system and the database. The system filters out landscape features from different locations and extracts preset keywords for the corresponding POIs. Only POIs within a 120-degree field of view in the direction of vehicle travel can be considered observable POIs. Step C, User Interest Fusion: Read the user's historical interest data and scene keywords, and fuse them with the POI's preset keywords to obtain a keyword set; Step D: Content matching. The keyword set is matched against the similarity of the speech content database to find the content with the highest matching degree. Step E: Send the content to the program on the phone.

3. The car navigation method based on mobile phone sensors according to claim 1, characterized in that: In step three, the playback judgment includes: interrupting the current explanation and playing a new POI explanation; adding the new POI explanation to the queue and playing it in sequence after the previous POI explanation is finished; and abandoning the playback of the new POI explanation.

4. The car navigation method based on mobile phone sensors according to claim 1, characterized in that: The server generates user profiles based on users' historical behavior and interests, filters out content that users are interested in, and caches the content in advance on the phone's local application.

5. The car navigation method based on mobile phone sensors according to claim 4, characterized in that: The historical behavior refers to the process by which the program collects the content that the user previously intended to listen to, generates historical behavior tags, receives the historical behavior tags, filters the content based on the historical behavior tags, and notes it as historical behavior content. The interest is defined as follows: the program collects the explanation content that the user has previously actively requested, generates interest tags, the server receives the interest tags, filters the explanation content according to the interest tags, and notes the content as interest content. The server compares historical behavior content and interest content, and prioritizes sending overlapping content to the program for pre-caching.

6. The car navigation method based on mobile phone sensors according to claim 1, characterized in that: The program adjusts the fatigue idle time T_gap based on the user's historical listening time during this driving trip. The playback interval is such that the length of a single voice message does not exceed 8 minutes, the total listening time within 30 minutes does not exceed 17 minutes, and the total listening time within 60 minutes does not exceed 36 minutes.

Images