An intelligent automotive interaction system for realizing AI interaction

Abstract

The present invention discloses an automobile intelligent interaction system for realizing AI interaction, including an intelligent control box, a Beidou chip, a communication chip, a camera, a massager, a sleepiness reminder module, a screen projection module, an emergency rescue module, an AI voice invitation module and a line recommendation module. The emergency rescue module is connected to the intelligent control box, and the intelligent control box analyzes whether it is necessary to actively send rescue information based on external sensors. If emergency rescue is required, the specific coordinates of the vehicle, the owner's contact information and the image are automatically uploaded to the emergency rescue module, and the rescue personnel go to rescue according to the information; the AI ​​voice invitation module is connected to the intelligent control box, and the address book personnel are invited by AI voice; the line recommendation module is connected to the intelligent control box, and the line recommendation module is based on user behavior and resource characteristics. The present invention can realize multiple functions such as emergency rescue, AI voice invitation, intelligent line recommendation, sleepiness reminder, and intelligent screen projection.

Description

Technical Field

[0001] The present invention relates to an automobile intelligent interaction system for realizing AI interaction, and belongs to the technical field of intelligent interaction. Background Art

[0002] With the rapid development of the Internet, Internet technology has entered the traditional automobile industry, providing intelligent services on the new generation of cars and integrating the experience of technology into car life. However, there is still a large stock of cars in the domestic market that have not been connected to the Internet of Vehicles. How to help some car owners to quickly and easily access the Internet of Vehicles and enjoy the benefits brought by technology is a relatively large market challenge.

[0003] Due to the long service life of cars, it is not realistic to replace the car in time. The car intelligent interaction system is designed to meet the needs of such car owners. The intelligent upgrade of the car mainly involves several aspects: car owner service, safe driving, emergency rescue, and intelligent AI. Summary of the invention

[0004] In view of the shortcomings of the prior art, the purpose of the present invention is to provide an intelligent interactive system for automobiles that realizes AI interaction, which can realize multiple functions such as emergency rescue, AI voice invitation, intelligent route recommendation, sleepiness reminder, and intelligent screen projection.

[0005] In order to achieve the above object, the present invention is implemented through the following technical solutions:

[0006] The present invention provides an intelligent automobile interaction system for realizing AI interaction, including an intelligent control box and

[0007] Beidou chip, used to provide precise location for vehicle trajectory and rescue;

[0008] Communication chip, used to transmit driving trajectory and rescue information to the intelligent control box;

[0009] The camera is connected to the smart control box to collect the owner's driving status information and send it to the drowsiness reminder module;

[0010] The massager is installed on the owner's neck and communicates with the sleepiness reminder module via Bluetooth wireless connection;

[0011] A drowsiness reminder module is connected to the smart control box and is used to detect the driving state information. If the driver is in a drowsy state, the smart control box sends a reminder instruction to the massager;

[0012] The screen projection module is used to wirelessly connect the mobile phone to the smart control box, and the smart control box then projects the content of the app to the car screen through a data cable to achieve car-machine interconnection;

[0013] The emergency rescue module is connected to the smart control box. The smart control box analyzes whether it is necessary to actively send rescue information based on external sensors. If emergency rescue is needed, the specific coordinates of the vehicle, the owner's contact information and images are automatically uploaded to the emergency rescue module, and rescue personnel go to the rescue based on the information. The owner can also submit a rescue application through the app or voice interaction;

[0014] AI voice invitation module, connected to the smart control box, to invite people in the address book through AI voice;

[0015] And the route recommendation module is connected to the smart control box to recommend routes based on user behavior and resource characteristics.

[0016] The method by which the smart control box analyzes whether it is necessary to actively send rescue information based on external sensors is as follows: when the vehicle encounters abnormal deceleration or stop within a set time range, based on the vehicle's last motion state, if the speed is reduced within the trigger condition within the preset time, the system will actively send a rescue request.

[0017] The method for inviting people in the address book through AI voice is as follows: the client collects voice data and transmits it to the voice recognition module through a network connection to identify the text content; the identified text content is transmitted to the intention analysis module through a network connection to analyze the intention behavior and content; the intention analysis module returns the analysis result to the client, and the client searches for matching contact information in the address book based on the intention behavior and content; after finding the contact in the address book, it is sent to the back-end server through network transmission, and the back-end server pushes the invitation information to the invited client device based on the unique identification ID of the client device.

[0018] Before or while the client collects voice data, the client accesses the names of all contacts in the address book to form a vocabulary library for narrowing the search scope in subsequent searches.

[0019] The specific processing method of the above-mentioned speech recognition module is as follows: preprocessing; extracting feature parameters; matching mode.

[0020] The method for recommending routes based on user behavior and resource characteristics is as follows: receiving a route recommendation request submitted by a client; recommending a first itinerary according to the destination carried in the route recommendation request; centering on the first itinerary, based on the screening conditions carried in the route recommendation request and a user portrait generated according to a POI resource library in the user's visited geographic information, and obtaining a user tag weight for reflecting the user's preferences and behavioral intentions according to the user portrait; finally planning a travel route that meets the user's needs; and recommending the planned travel route to the client.

[0021] When the above-mentioned destination is a scenic spot, the scenic spot is taken as the first itinerary; when the destination is a city, one of the scenic spots that match the user tag is randomly selected as the destination.

[0022] The above method for generating a user portrait based on a user's visit to a POI resource library in geographic information is as follows: based on the user's behavior of visiting the POI resource library, weight classification is performed through the TF-IDF algorithm to extract the user's preferences and features and generate a user portrait.

[0023] The present invention also includes a glass-breaking warning module, which is connected to the intelligent control box via a data cable and transmits vehicle status information to the intelligent control box in real time. When the vehicle falls into water, the glass is broken and the intelligent control box is notified to send rescue information. The present invention also includes a tire pressure detection module: connected to the intelligent control box via a USB data cable, the tire pressure data is transmitted to the intelligent control box and displayed on the vehicle screen. The tire pressure has a normal range value, and a prompt will be given on the screen when it exceeds this range.

[0024] The present invention also includes a driving recorder module, which is used to record the screen recording of the vehicle within 7 days and store it locally. The data will be cleaned up regularly. A screen recording can be selected from the app for uploading as the basis for the vehicle to be in danger;

[0025] The present invention also includes a reversing image module, which transmits the camera data around the vehicle body to the screen through a data cable connection, displays it in the form of an image, gives a warning reminder according to the distance, and reduces collisions when reversing.

[0026] The present invention has the functions of screen projection, driving record, reversing image, tire pressure detection, drowsiness reminder, glass breakage warning, etc.; in addition, the AI ​​voice invitation module can eliminate the need to operate the screen while driving, thereby improving safety, adding fun and a sense of technology; based on the route recommendation module, multi-day routes can be generated according to user preferences.

[0027] The present invention displays functions by connecting to the vehicle screen, and the mobile phone is connected to the system through WIFE and Bluetooth, thereby eliminating the dependence of the car owner on the mobile phone and avoiding the embarrassing situation of using the mobile phone holder in the driving seat. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] Figure 1 A working principle diagram of an automobile intelligent interaction system for realizing AI interaction according to the present invention;

[0029] Figure 2 This is a specific workflow diagram of the AI ​​voice invitation of the present invention;

[0030] Figure 3 A specific workflow diagram is recommended for the intelligent circuit of the present invention. DETAILED DESCRIPTION

[0031] In order to make the technical means, creative features, objectives and effects achieved by the present invention easy to understand, the present invention is further explained below in conjunction with specific implementation methods.

[0032] See also Figure 1 , an automobile intelligent interaction system for realizing AI interaction of the present invention comprises the following modules:

[0033] Beidou chip is used to provide precise location for vehicle driving trajectory and rescue.

[0034] The communication chip communicates with the server and transmits the driving trajectory and rescue information to the server.

[0035] The camera is used to collect the owner's driving status information and send it to the drowsiness reminder module.

[0036] The massager is installed on the owner's neck and communicates with the sleepiness reminder module via Bluetooth wireless connection;

[0037] The drowsiness reminder module is used to analyze the driving status information. If the driver is in a drowsy state, a reminder instruction is sent to the massager, and the massager will remind the owner through massage. The camera on the smart control box detects the driver's driving status. If the driver is drowsy or fatigued, the smart control box will send a command to the massager on the driver's neck, which will instantly reach a high-intensity vibration to remind the driver. The smart control box also gives a voice reminder: Please don't drive when you are tired. The detection method is: by scanning the pupil condition and judging the driver's fatigue state through data analysis, once the driver is in a fatigued or drowsy state, the system will give the driver an early warning.

[0038] Screen projection module: The mobile phone and the car's intelligent interactive system are wirelessly connected based on WIFI and Bluetooth, and the smart control box then projects the app content to the car screen through a data cable to achieve car-machine interconnection. The smart control box will be equipped with a corresponding app, which can project some functions to the smart control box screen (or car screen), for example: when using the formation navigation function, it can be displayed horizontally on the smart control box screen.

[0039] In the emergency rescue module, the intelligent control box analyzes whether it is necessary to actively send rescue information based on external sensors. In the event of an emergency rescue, the specific coordinates of the vehicle, the owner's contact information and images will be automatically uploaded to the rescue system, and rescue personnel will go to the rescue based on the information; the owner can also submit a rescue application through the app or voice interaction. The specific analysis method is: when the vehicle encounters abnormal deceleration or stops within a certain time range, according to the vehicle's last motion state, if the speed is reduced within the triggering conditions within a certain period of time, the system will actively send a rescue request.

[0040] The AI ​​voice invitation module is connected to the smart control box to invite people in the address book through AI voice.

[0041] The route recommendation module is connected to the smart control box and recommends routes based on user behavior and resource characteristics.

[0042] 360 reversing image module: It is an external device that transmits the camera data around the vehicle body to the screen via a data cable and displays it in the form of images. It gives warnings based on the distance to reduce collisions when reversing.

[0043] Tire pressure detection module: Use a USB data cable to connect to the smart control box, transmit the tire pressure data to the smart control box, and display it on the car screen. The tire pressure has a normal range value, and a prompt will be given on the screen when it exceeds this range. Set a tire pressure warning value. In the case of abnormal tire pressure, the smart control box can give a voice reminder, and you can also receive a tire pressure alarm push on the mobile phone app.

[0044] Driving recorder module: The car's intelligent interactive system will be equipped with a driving record camera, which will record the vehicle's driving for 7 days and store it locally. The data will be cleaned up regularly. You can select a recording from the app and upload it as a basis for some business scenarios, such as when the vehicle is in danger.

[0045] The glass breakage warning module is connected to the smart control box via a data cable, and transmits the vehicle status information to the smart control box in real time. When the vehicle falls into water, while breaking the glass, it notifies the smart control box to send rescue information.

[0046] The car's intelligent interactive system displays its functions by connecting to the car's screen, and the mobile phone is connected to the car's intelligent interactive system via WIFE and Bluetooth, eliminating the owner's dependence on the mobile phone and avoiding the embarrassing situation of using a mobile phone holder in the driving seat.

[0047] AI is used as the voice assistant of the smart control box app, eliminating the need to operate the screen while driving, improving safety, and adding fun and a sense of technology;

[0048] Usage scenarios: 1) Various voice reminders and alarms; 2) Use voice to replace manual operations; 3) Ask about weather conditions, ask about the destination when getting on the bus, etc.; 4) Invite friends to join the team when navigating in a formation; 5) Make some content recommendations: attractions, routes, activities... 6) Call for rescue.

[0049] In this embodiment, see Figure 2 , the method to invite people in the address book through AI voice is as follows:

[0050] The client collects voice data and passes it to the voice recognition module through a network connection to identify the text content; the identified text content is passed to the intention analysis module through a network connection to analyze the intention behavior and content; the intention analysis module returns the analysis results to the client, and the client searches for matching contact information in the address book based on the intention behavior and content; after finding the contact in the address book, it is sent to the back-end server through network transmission, and the back-end server pushes the invitation information to the invited client device based on the unique identification ID of the client device. Before or while the client collects voice data, it also includes that the client accesses the names of all contacts in the address book to form a vocabulary library, which is used to narrow the search scope in subsequent searches.

[0051] The specific processing method of the speech recognition module is as follows:

[0052] Preprocessing; extracting feature parameters; matching patterns.

[0053] Preprocessing specifically includes pre-filtering, sampling, quantization, windowing, endpoint detection, and pre-emphasis steps. The extracted feature parameters must meet the following requirements:

[0054] (1) The extracted feature parameters can effectively represent the speech features and are discriminative;

[0055] (2) There is independence between the parameters of each order;

[0056] (3) The feature parameters should be easy to calculate to ensure real-time implementation of speech recognition.

[0057] The pattern matching method is as follows: first, for each word in the system, a codebook is made as the reference template of the word. There are M words in total, so there are M codebooks in total, forming a template library; during recognition, for any input speech feature vector sequence X = {X1, X2, ..., X N}, calculate the total average distortion error of each feature vector in the sequence to each codebook in the template library, find the codebook corresponding to the minimum distortion error, and output the corresponding word as the recognition result.

[0058] The intention analysis module analyzes the intention behavior and content. The AI ​​module can be trained and the keyword invitation can be regarded as a behavior, and the content after the invitation can be regarded as the invited object. The specific method is as follows: the recognized text content is classified into strong intent and weak intent, and the intention is analyzed by rule module analysis and intention vocabulary; if there is a strong invitation intention, proceed to the next step of name matching; if the invitation intention is not analyzed, it is necessary to re-enter the voice collection.

[0059] After the client receives the result of intent analysis, it matches the address book contact information according to the result. The specific method is as follows: find the most matching contact in the address book, send it to the back-end server through network transmission, and the back-end server then sends the invitation message to the invitee's mobile device through the message push service; the invitation will produce three results, matching multiple contacts, one contact, and no contact; when there are multiple contacts, choose which contact to invite; when there is only one contact, it is invited by default; when there is no invitee in the address book, give text and voice prompts in time.

[0060] See also Figure 3 In this embodiment, the route recommendation method based on user behavior and resource characteristics includes: receiving a route recommendation request submitted by a client; recommending a first itinerary according to the destination carried in the route recommendation request; centering on the first itinerary, based on the screening conditions carried in the route recommendation request and the user portrait generated according to the POI resource library in the user's accessed geographic information, and obtaining a user tag weight for reflecting the user's preferences and behavioral intentions according to the user portrait; finally planning a travel route that meets the user's needs; and recommending the planned travel route to the client.

[0061] When the destination is a scenic spot, the scenic spot is taken as the first itinerary; when the destination is a city, one of the scenic spots that match the user's tag is randomly selected as the destination.

[0062] The method for generating a user portrait based on the POI resource library in the user's access to geographic information is as follows:

[0063] According to the user's behavior in visiting the POI resource library, weight classification is performed through the TF-IDF algorithm to extract the user's preferences and characteristics and generate a user portrait.

[0064] The specific method of weight classification by the TF-IDF algorithm is as follows:

[0065] Among them, T represents a label, T i represents all tags, P represents a user, W(P,T) represents the number of times a tag T is used to mark user P, W(P,T i) represents the number of all labels of user P, and TF(P,T) represents the proportion of the number of times label T is marked in all labels of user P;

[0066] Where Pi represents all users, W(Pi,,Ti) represents the sum of all tags of all users, W(Pi,T) represents the sum of all users with tag T, and IDF(P,T) represents the scarcity of tag T among all tags.

[0067] TF-IDF=TF*IDF, where TF-IDF indicates the user's attention to a tag or the importance of a tag to the user.

[0068] The specific method for obtaining the user tag weight according to the user portrait is as follows:

[0069] User tag weight = behavior type weight * time decay * TF-IDF * number of behaviors;

[0070] The behavior type weight indicates the different importance of browsing, searching, collecting, ordering, and purchasing to the user, which is expressed by the high and low scores;

[0071] The time decay means that the importance of the behavior will decay with the passage of time. Different decay amplitudes are set according to time periods, and the value range is 0 to 1. If the visit is continuous, the time decay coefficient is between 1 and 2, indicating frequent attention.

[0072] Each POI resource in the POI resource library has its own independent tags and features, including type, theme, play time and play intensity.

[0073] The filtering conditions carried in the route recommendation request include hotel level, travel intensity, route theme, travel days, price range and number of people.

[0074] When the number of days of travel in the filter condition is one day, the hotel will be recommended based on the last itinerary point;

[0075] If the tour duration is two days or more, the best accommodation area will be calculated by combining the last itinerary of the day and the first itinerary of the next day, and hotels in the accommodation area that meet the level requirements will be recommended.

[0076] The above shows and describes the basic principles and main features of the present invention and the advantages of the present invention. It should be understood by those skilled in the art that the present invention is not limited to the above embodiments. The above embodiments and descriptions are only for explaining the principles of the present invention. Without departing from the spirit and scope of the present invention, the present invention may have various changes and improvements, which fall within the scope of the present invention to be protected. The scope of protection of the present invention is defined by the attached claims and their equivalents.

Claims

1. An intelligent automobile interaction system for realizing AI interaction, characterized in that: Including smart control box, also includes Beidou chip, used to provide precise location for vehicle trajectory and rescue; Communication chip, used to transmit driving trajectory and rescue information to the intelligent control box; The camera is connected to the smart control box to collect the owner's driving status information and send it to the drowsiness reminder module; The massager is installed on the owner's neck and communicates with the sleepiness reminder module via Bluetooth wireless connection; A drowsiness reminder module is connected to the smart control box and is used to detect the driving state information. If the driver is in a drowsy state, the smart control box sends a reminder instruction to the massager; The screen projection module is used to wirelessly connect the mobile phone to the smart control box, and the smart control box then projects the content of the app to the car screen through a data cable to achieve car-machine interconnection; The emergency rescue module is connected to the smart control box. The smart control box analyzes whether it is necessary to actively send rescue information based on external sensors. If emergency rescue is needed, the specific coordinates of the vehicle, the owner's contact information and images are automatically uploaded to the emergency rescue module, and rescue personnel go to the rescue based on the information. The owner can also submit a rescue application through the app or voice interaction; AI voice invitation module, connected to the smart control box, to invite people in the address book through AI voice; and route recommendation module, connected to the smart control box, to recommend routes based on user behavior and resource features; the method for the smart control box to analyze whether it is necessary to actively send rescue information based on external sensors is as follows: when the vehicle encounters abnormal deceleration or stops within a set time range, based on the vehicle's last motion state, if the speed within the trigger condition is reduced within the preset time, the system will actively send a rescue request; the method for inviting people in the address book through AI voice is as follows: the client collects voice data and transmits it to the voice recognition module through a network connection to identify the text content; the identified text content is transmitted to the intent analysis module through a network connection to analyze the intent behavior and content; the intent analysis module returns the analysis result to the client, and the client acts according to the intent Search for matching contact information in the address book for the content; after finding the contact in the address book, send it to the back-end server through network transmission, and the back-end server pushes the invitation information to the invited client device according to the unique identification ID of the client device; the method for recommending routes based on user behavior and resource characteristics is as follows: receive a route recommendation request submitted by the client; recommend a first itinerary according to the destination carried in the route recommendation request; centering on the first itinerary, based on the screening conditions carried in the route recommendation request and the user portrait generated according to the POI resource library in the user's access geographic information, and obtaining the user tag weight for reflecting the user's preferences and behavioral intentions according to the user portrait; finally plan a travel route that meets the user's needs; The planned travel route is recommended to the client; it also includes a glass breakage warning module, which is connected to the smart control box through a data cable and transmits the vehicle status information to the smart control box in real time. When the vehicle falls into water and breaks the glass, it notifies the smart control box to send rescue information; It also includes a tire pressure detection module: use a USB data cable to connect to the smart control box, transmit the tire pressure data to the smart control box, and display it on the car screen. The tire pressure has a normal range of values, and a prompt will be given on the screen when it exceeds this range.

2. The automobile intelligent interaction system for realizing AI interaction according to claim 1, characterized in that: Before or while the client collects voice data, the client accesses the names of all contacts in the address book to form a vocabulary library for narrowing the search scope in subsequent searches.

3. The automobile intelligent interaction system for realizing AI interaction according to claim 2, characterized in that: The specific processing method of the speech recognition module is as follows: preprocessing; extracting feature parameters; matching patterns.

4. The automobile intelligent interaction system for realizing AI interaction according to claim 1, characterized in that: When the destination is a scenic spot, the scenic spot is taken as the first itinerary; when the destination is a city, one of the scenic spots that match the user tag is randomly selected as the destination.

5. The automobile intelligent interaction system for realizing AI interaction according to claim 1, characterized in that: The method for generating a user portrait based on a user's visit to a POI resource library in geographic information is as follows: based on the user's behavior of visiting the POI resource library, weight classification is performed through the TF-IDF algorithm to extract the user's preferences and features and generate a user portrait.

6. The automobile intelligent interaction system for realizing AI interaction according to claim 1, characterized in that: It also includes a driving recorder module, which is used to record the vehicle's driving screen within 7 days and store it locally. The data will be cleaned up regularly. You can select a screen recording from the app and upload it as the basis for the vehicle's accident. It also includes a reversing image module, which transmits camera data around the vehicle body to the screen via a data cable and displays it in the form of an image. It gives warnings based on the distance to reduce collisions when reversing.

Images