Vehicle-mounted virtual assistant generation method

Abstract

The invention provides a method for generating a vehicle-mounted virtual assistant, which comprises the following steps of: classifying all scenes according to the triggering forms of the scenes, constructing a plurality of virtual assistant characters, and enabling each virtual assistant character to have a corresponding expression form in each scene. And collecting user behavior habits according to various data burying points preset in the in-vehicle infotainment system. And on the basis of the collected user behavior habits, through a preset formula and weight distribution, calculating the matching degree between each virtual assistant character and the user behavior habits, and obtaining the virtual assistant character which is most matched with the user. And according to the matched character of the virtual assistant, the vehicle-mounted virtual assistant displays a personalized representation form in a corresponding scene. According to the vehicle-mounted virtual assistant generation method provided by the invention, the representation form matched with the behavior habits of the user can be displayed when voice broadcasting is carried out according to the behavior habits of the user, so that the virtual assistant is endowed with emotional and personalized attributes.

Description

Technical Field

[0001] The present invention relates to the field of computer information, and in particular to a method for generating a vehicle-mounted virtual assistant. Background Art

[0002] With the rapid development of artificial intelligence technology, cars have gradually surpassed the scope of traditional means of transportation and gradually evolved into users' third living space. In this transformation process, the role of in-vehicle smart assistants, as an important embodiment of automobile intelligence, has become increasingly prominent. They are not only responsible for providing basic information services such as navigation and entertainment, but also bear the heavy responsibility of deeply interacting with users and establishing emotional connections. However, most of the in-vehicle smart assistants on the current market have the problems of single functions and lack of personalized and emotional expression. In particular, when interacting with users, smart assistants often use unified animations and voice broadcasts, and cannot provide customized service experiences based on individual differences of users.

[0003] In the known field, there are some solutions that attempt to solve this problem. For example, by presetting multiple virtual assistant images of different styles for users to choose from, each virtual assistant image is designed with a specific form of companionship, in order to increase the user's personalized choices. However, the disadvantage of this solution is that the number of preset virtual assistant images is limited, and the companionship form of each image is fixed and cannot be dynamically adjusted according to the user's behavioral habits, thereby limiting the depth and breadth of the user experience.

[0004] In order to overcome the shortcomings of existing solutions, the industry has also made some attempts to improve them. On the one hand, by creating multiple corresponding virtual assistant companion animations for each user interaction scenario, when the user triggers a certain scenario, the corresponding animation of the scenario is randomly played to increase the diversity and fun of the interaction. However, this method still fails to achieve true personalization, because the playback of the animation is random and cannot accurately reflect the user's personalized needs and behavioral habits. On the other hand, allowing users to customize the accessories of the virtual assistant, although it can increase the user's sense of freshness to a certain extent, the changes in accessories are limited to the appearance level and cannot touch the personalized customization of the smart assistant's personality and behavior. Therefore, the existing in-vehicle smart assistant solutions still have obvious deficiencies in personalization and emotionalization, and cannot meet users' expectations for deep interaction and emotional resonance. Summary of the invention

[0005] In order to solve the above problems, the present invention proposes a method for generating an in-vehicle virtual assistant, which aims to build an intelligent assistant personality model suitable for the user by deeply learning the user's daily car use behavior and combining the user's expected intelligent assistant personality definition.

[0006] The present invention proposes a method for generating a vehicle-mounted virtual assistant, comprising:

[0007] All scenarios are classified according to their triggering forms, and multiple virtual assistant personalities are constructed. Each virtual assistant personality has a corresponding manifestation in each scenario;

[0008] Collect user behavior habits based on various data points preset in the vehicle system;

[0009] Based on the collected user behavior habits, the matching degree between each virtual assistant personality and the user behavior habits is calculated through a preset formula and weight distribution, and the virtual assistant personality that best matches the user is obtained;

[0010] Based on the matched virtual assistant personality, the in-vehicle virtual assistant displays personalized expressions in corresponding scenarios.

[0011] In one embodiment, the scenarios include: basic state scenario, awakening state scenario, broadcast state scenario, vehicle control scenario, multimedia control scenario, AI recommendation scenario, nearby food recommendation scenario, companionship scenario, and self-entertainment scenario, and the companionship scenario includes music and navigation.

[0012] In one embodiment, the method of classifying all scenes according to the triggering form of the scenes includes:

[0013] The basic state scene, the awakening state scene, and the broadcast state scene are classified into the first category;

[0014] The vehicle control scenarios are classified into the second category;

[0015] Multimedia control scenarios are divided into the third category;

[0016] AI recommendation scenarios and nearby food recommendation scenarios are divided into the fourth category;

[0017] The companionship scenes and self-entertainment scenes are divided into the fifth category.

[0018] In one embodiment, in constructing multiple virtual assistant personalities, each virtual assistant personality has multiple personality attributes, and each virtual assistant personality has two opposite expression directions under the same personality attribute, and the two opposite expression directions are an upward-biased personality and a downward-biased personality.

[0019] In one embodiment, each virtual assistant personality has a corresponding expression form in each scene, and the expression form includes: frequency of interaction with the car owner, action performance, broadcast content, broadcast voice, and accessories worn.

[0020] In one embodiment, the user behavior habits include: vehicle purpose, voice usage frequency, voice package type selected by TTS, virtual assistant image matching style, vehicle speed high and low frequency, navigation destination type, and car washing or maintenance appointment frequency.

[0021] In one embodiment, after collecting user behavior habits, the method further includes: classifying the collected user behavior habits and assigning a value to each user behavior habit to indicate the influence of the user behavior habit on the personality of the virtual assistant.

[0022] In one embodiment, the weight allocation is: based on the correlation between the user's behavior habits and personality, different weights are allocated to each user's behavior habits.

[0023] In one embodiment, the preset formula is:

[0024] y=ax1+bx2+…+nx p (1);

[0025] Among them, p is the number of user behavior habits that affect the personality of the virtual assistant;

[0026] x1,x2,…,x p is the user's behavior habit, with a value of 1 or -1, indicating the user's behavior habit's bias towards the virtual assistant's personality under this personality attribute, including an upward bias and a downward bias;

[0027] a, b, …, n are the preset weights of the user's daily behavior habits, indicating the degree of influence of the user's behavior habits on the character of this virtual assistant;

[0028] When y is less than 0, the lower-biased personality under this personality attribute is selected, and when y is greater than 0, the upper-biased personality under this personality attribute is selected.

[0029] In one embodiment, the number of user behavior habits that affect the personality of a virtual assistant does not exceed two.

[0030] The in-vehicle virtual assistant generation method of the present invention has the following beneficial effects:

[0031] 1. The core of the present invention is to be user-centric, and to build the most suitable virtual assistant personality for the user through in-depth study and analysis of the user's daily car-use behavior and the desired virtual assistant personality. This personalized virtual assistant can not only provide a more personalized expression that is more in line with the user's needs when the user interacts with the smart cockpit, such as the frequency of interaction with the car owner, action performance, broadcast content, broadcast voice, and accessories, but also dynamically adjust its service methods and content according to the user's preferences and habits. In this way, users can not only get a more convenient and efficient operating experience, but also feel more intimate and personalized services, thereby greatly improving user satisfaction and loyalty.

[0032] 2. The present invention further enhances the adaptability and interactivity of the intelligent assistant by designing multiple sets of scene-related virtual assistant action performances, broadcast content and broadcast voice for different scenarios. In different scenarios, the intelligent assistant can automatically switch to the personality performance that best suits the atmosphere and needs of the current scene, and interact with the user more naturally and smoothly. This scenario-based design not only improves the intelligence level of the virtual assistant, but also enables users to obtain a more comfortable and pleasant driving experience in different scenarios. At the same time, the multi-personality performance of the virtual assistant also increases the interactivity between the user and the smart cockpit, allowing the user to enjoy more fun and surprises during use.

[0033] 3. The present invention also collects and analyzes user behavior habits through the data points preset by the vehicle system, providing data support for the character generation of the virtual assistant. This data-driven method not only makes the character generation of the virtual assistant more scientific and accurate, but also can be continuously optimized and upgraded according to the actual usage of the user. As user behavior habits change and needs upgrade, the intelligent assistant can adjust its personality performance and service methods in a timely manner to adapt to the new needs of users. This intelligent upgrade not only enhances the competitiveness of the intelligent assistant, but also provides strong support for the intelligent upgrade of the entire smart cockpit system. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] Figure 1 A schematic diagram of a flow chart of a method for generating a vehicle-mounted virtual assistant according to an embodiment of the present invention;

[0035] Figure 2 This is a schematic diagram of how user behavior habits affect the personality of a virtual assistant according to an embodiment of the present invention. DETAILED DESCRIPTION

[0036] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the invention and are not used to limit the invention.

[0037] All virtual assistant companionship styles in existing solutions are fixed. After the user selects a certain companionship style group, the companionship form in some scenarios may not be what the user wants. The present invention proposes a method for generating an in-vehicle virtual assistant, which can display personalized expressions according to the user's behavioral habits when the user interacts with the smart cockpit, such as the frequency of interaction with the car owner, action performance, broadcast content, broadcast voice, and accessories, etc., thereby giving the virtual assistant emotional and personalized attributes, and creating a virtual assistant companionship expression form that is unique to each person and user-specific through dynamic transformation, thereby achieving truly personalized companionship.

[0038] The vehicle-mounted virtual assistant generation method comprises:

[0039] S1: Classify all scenarios according to their triggering forms and construct multiple virtual assistant personalities. Each virtual assistant personality has a corresponding expression in each scenario.

[0040] The present invention calls the interaction scenario between the user and the car a scene, where the scene includes but is not limited to: smart assistant basic state scene, wake-up state scene, broadcast state scene, vehicle control scene, multimedia control scene, AI recommendation scene, nearby food recommendation, music, navigation, etc. companionship scene, and self-entertainment scene.

[0041] Secondly, the above scenarios can be classified into n categories according to the triggering form of the scenarios, s1, s2, ..., s n , n is a natural number. For example: the basic state scene, awakening state scene, and broadcast state scene are divided into the first category s1, the vehicle control scene is divided into the second category s2, the multimedia control scene is divided into the third category s3, the AI ​​recommendation scene and nearby food recommendation scene are divided into the fourth category s4, and the companionship scene and self-entertainment scene are divided into the fifth category s5.

[0042] Next, build multiple virtual assistant personalities Where q is a natural number. Each virtual assistant personality has multiple personality attributes. Each virtual assistant personality has two opposite performance directions under the same personality attribute. The two opposite performance directions are upward biased personality and downward biased personality. That is, they represent two opposite directions of expression under the same personality attribute. For example, Quiet and lively, Reason and mood.

[0043] Every personality type in a virtual assistant personality (m=1 or 2) in scene s n There are corresponding forms of expression in each scene, see Table 1 below. The forms of expression include the frequency of interaction with the owner, action performance, broadcast content, broadcast voice, accessories and clothing, etc. Different personalities in scenes n There are different manifestations in each character. q Will appear in scenes with probability n The probability of each character appearing is:

[0044]

[0045] Table 1 Interaction scenarios of virtual assistant personality

[0046]

[0047] S2: Collect user behavior habits based on various data points preset in the vehicle system. User behavior habits include: vehicle purpose, voice usage frequency, voice package type selected by TTS, virtual assistant image matching style, vehicle speed frequency, navigation destination type, and car wash or maintenance appointment frequency. n After that, including: collecting user behavior habits x n Classify and assign a value (1 or -1) to each user behavior habit to indicate the impact of the user behavior habit on the personality of the virtual assistant.

[0048] S3: Based on the collected user behavior habits, the matching degree between each virtual assistant personality and the user behavior habits is calculated through a preset formula and weight distribution to obtain the virtual assistant personality that best matches the user.

[0049] For each virtual assistant personality Specifically, which one belongs to the same personality attribute? In the present invention, various data points preset in the vehicle system are collected and analyzed to obtain the personality attribute that matches the user. Will be affected by user behavior habits x1,x2,x3,x4,x5,…,x p The combined influence of one or more of the above factors is used to calculate the personality of a virtual assistant based on the correlation between user behavior habits and personality. When is behavior habit x p Different weights are assigned, where one behavior habit or several user behavior habits dominate, and the weight of each user behavior habit can be preset when the system is released.

[0050] In this embodiment, the preset formula is:

[0051] y=ax1+bx2+…+nx p (1);

[0052] Wherein, p is the number of user behavior habits that affect the personality of the virtual assistant. In this embodiment, preferably, the number of user behavior habits that affect the personality of a virtual assistant does not exceed two.

[0053] x1,x2,…,x p It is the user's behavior habit, with a value of 1 or -1, indicating the user's behavior habit's bias towards the virtual assistant's personality under this personality attribute, including upper bias and lower bias.

[0054] a, b, …, n are the preset weights of the user's daily behavior habits, indicating the degree of influence of the user's behavior habits on the character of this virtual assistant.

[0055] When y is less than 0, then Select the lower biased personality under this personality attribute. When y is greater than 0, then Select the upward-biased personality under this personality attribute.

[0056] S4: Based on the matched virtual assistant personality, the in-vehicle virtual assistant displays personalized expressions in the corresponding scenes, including but not limited to virtual assistant interactive animations, TTS broadcast voices, TTS broadcast content, and the frequency of interaction with the car owner.

[0057] The in-vehicle virtual assistant generation method of the present invention has the following beneficial effects:

[0058] 1. The core of the present invention is to be user-centric, and to build the most suitable virtual assistant personality for the user through in-depth study and analysis of the user's daily car-use behavior and the desired virtual assistant personality. This personalized virtual assistant can not only provide a more personalized expression that is more in line with the user's needs when the user interacts with the smart cockpit, such as the frequency of interaction with the car owner, action performance, broadcast content, broadcast voice, and accessories, but also dynamically adjust its service methods and content according to the user's preferences and habits. In this way, users can not only get a more convenient and efficient operating experience, but also feel more intimate and personalized services, thereby greatly improving user satisfaction and loyalty.

[0059] 2. The present invention further enhances the adaptability and interactivity of the intelligent assistant by designing multiple sets of scene-related virtual assistant action performances, broadcast content and broadcast voice for different scenarios. In different scenarios, the intelligent assistant can automatically switch to the personality performance that best suits the atmosphere and needs of the current scene, and interact with the user more naturally and smoothly. This scenario-based design not only improves the intelligence level of the virtual assistant, but also enables users to obtain a more comfortable and pleasant driving experience in different scenarios. At the same time, the multi-personality performance of the virtual assistant also increases the interactivity between the user and the smart cockpit, allowing the user to enjoy more fun and surprises during use.

[0060] 3. The present invention also collects and analyzes user behavior habits through the data points preset by the vehicle system, providing data support for the character generation of the virtual assistant. This data-driven method not only makes the character generation of the virtual assistant more scientific and accurate, but also can be continuously optimized and upgraded according to the actual usage of the user. As user behavior habits change and needs upgrade, the intelligent assistant can adjust its personality performance and service methods in a timely manner to adapt to the new needs of users. This intelligent upgrade not only enhances the competitiveness of the intelligent assistant, but also provides strong support for the intelligent upgrade of the entire smart cockpit system.

[0061] The above-described embodiments are only further explanations of the present invention, and are not intended to limit the present invention in other forms. The present invention may also have other various embodiments. Without departing from the spirit and essence of the present invention, those skilled in the art may make various corresponding modifications and changes according to the present invention, but these corresponding modifications and changes should all fall within the protection scope of the present invention.

Claims

1. A method for generating a vehicle-mounted virtual assistant, characterized in that: include: All scenarios are classified according to their triggering forms, and multiple virtual assistant personalities are constructed. Each virtual assistant personality has a corresponding manifestation in each scenario; Collect user behavior habits based on various data points preset in the vehicle system; Based on the collected user behavior habits, the matching degree between each virtual assistant personality and the user behavior habits is calculated through a preset formula and weight distribution, and the virtual assistant personality that best matches the user is obtained; Based on the matched virtual assistant personality, the in-vehicle virtual assistant displays personalized expressions in corresponding scenarios.

2. The method for generating a vehicle-mounted virtual assistant according to claim 1, characterized in that: The scenarios include: basic state scenario, wake-up state scenario, broadcast state scenario, vehicle control scenario, multimedia control scenario, AI recommendation scenario, nearby food recommendation scenario, companionship scenario, and self-entertainment scenario. The companionship scenario includes music and navigation.

3. The method for generating a vehicle-mounted virtual assistant according to claim 1, characterized in that: All scenes are classified according to the triggering form of the scene, including: The basic state scene, the awakening state scene, and the broadcast state scene are classified into the first category; The vehicle control scenarios are classified into the second category; Multimedia control scenarios are divided into the third category; AI recommendation scenarios and nearby food recommendation scenarios are divided into the fourth category; The companionship scenes and self-entertainment scenes are divided into the fifth category.

4. The method for generating a vehicle-mounted virtual assistant according to claim 1, characterized in that: In the construction of multiple virtual assistant personalities, each virtual assistant personality has multiple personality attributes, and each virtual assistant personality has two opposite expression directions under the same personality attribute, and the two opposite expression directions are an upward-biased personality and a downward-biased personality.

5. The method for generating a vehicle-mounted virtual assistant according to claim 1, characterized in that: Each virtual assistant personality has a corresponding form of expression in each scene, and the form of expression includes: frequency of interaction with the car owner, action performance, broadcast content, broadcast voice, and accessories.

6. The method for generating a vehicle-mounted virtual assistant according to claim 1, characterized in that: The user behavior habits include: vehicle purpose, voice usage frequency, voice package type selected by TTS, virtual assistant image matching style, vehicle speed high and low frequency, navigation destination type, and car washing or maintenance appointment frequency.

7. The method for generating a vehicle-mounted virtual assistant according to claim 1, characterized in that: After collecting the user behavior habits, the method includes: classifying the collected user behavior habits and assigning a value to each user behavior habit to indicate the influence of the user behavior habit on the character of the virtual assistant.

8. The method for generating a vehicle-mounted virtual assistant according to claim 1, characterized in that: The weight allocation is: based on the correlation between the user's behavior habits and personality, different weights are allocated to each user's behavior habits.

9. The method for generating a vehicle-mounted virtual assistant according to claim 1, characterized in that: The preset formula is: y=ax1+bx2+…+nx p (1)? Among them, p is the number of user behavior habits that affect the personality of the virtual assistant; x1,x2,…,x p is the user's behavior habit, with a value of 1 or -1, indicating the user's behavior habit's bias towards the virtual assistant's personality under this personality attribute, including an upward bias and a downward bias; a, b, …, n are the preset weights of the user's daily behavior habits, indicating the degree of influence of the user's behavior habits on the character of this virtual assistant; When y is less than 0, the lower-biased personality under this personality attribute is selected, and when y is greater than 0, the upper-biased personality under this personality attribute is selected.

10. The method for generating a vehicle-mounted virtual assistant according to claim 9, characterized in that: The number of user behavior habits that influence the personality of a virtual assistant does not exceed two.

Images