Hub ai travel assistant working method based on multi-modal fusion and maas collaboration

Abstract

The application discloses a hub AI travel small assistant working method based on multi-modal fusion and MaaS cooperation, which comprises the following steps: 1, collecting the demand of a user through a man-machine interaction interface, understanding the intention of the user through a natural language processing (NLP) module, and using a service arrangement engine to disassemble the demand of the user into several sub-services; 2, collecting the individual preference of the user according to all the sub-services; 3, automatically calling internal algorithm modules and external service provider interfaces in logical order; 4, using an AI large model to perform fusion analysis on multi-dimensional data obtained by using the internal algorithm modules and the external service provider interfaces, calculating the comprehensive utility of each travel scheme in real time, and performing sorting and recommendation; the sorting and recommendation is a recommended travel service mode combined with the individual preference; 5, realizing the corresponding sub-service through the external service provider interface. The application solves the fundamental contradiction between service fragmentation and demand diversification.

Description

Technical Field

[0001] This invention relates to the field of artificial intelligence service technology for transportation hubs, and in particular to a working method for a hub AI travel assistant based on multimodal fusion and MaaS collaboration. Background Technology

[0002] With the large-scale development of integrated transportation hubs, passenger travel service demands are becoming increasingly diversified and personalized, making it difficult for the existing hub travel service system to meet the demand for efficient and precise services. Current mainstream hub travel assistance tools suffer from three major pain points:

[0003] 1. The multi-modal travel recommendation is based on a single dimension. For example, the existing technology with the public number CN202511465K is based on historical trajectory for vehicle scheduling, focusing only on operational efficiency optimization. It does not take into account the real-time personalized needs of passengers, such as luggage volume, time sensitivity, cost preference, etc., to compare multiple modes, resulting in insufficient matching of recommendations for ride-hailing, taxis, customized buses and rail transit.

[0004] For example, the prior art disclosed in CN112785026A focuses on the scheduling optimization or queue prediction of a single mode of transportation (such as taxis) within a hub.

[0005] Furthermore, existing technologies, such as the one disclosed in CN113112114B, focus on static route planning or shop information queries based on indoor hub maps. These solutions fail to dynamically integrate various transportation modes (ride-hailing, taxis, customized buses, and rail transit) inside and outside the hub for real-time comparison and recommendation, nor can they achieve one-click seamless service across different modes of transportation (e.g., rail transit + taxi).

[0006] 2. Fragmented service modules: For example, the existing technology disclosed in CN115880864A, while mentioning various modes of transportation, often presents its recommendation system, ticketing system, and navigation system as independent modules, with data and processes not integrated. Passengers need to use different applications to complete steps such as querying, comparing prices, purchasing tickets, booking rides, and navigating, resulting in a fragmented experience and cumbersome operation. Existing solutions mostly remain at the level of information integration (such as Level 1 services) or single-stage closed-loop solutions (such as only covering ride booking or ticketing), lacking a seamless end-to-end process from problem querying and itinerary planning to ticket booking and indoor / outdoor navigation, and failing to achieve dynamic updates of service status.

[0007] 3. The interaction methods are simplistic and lack proactive intelligence. Existing technologies mostly respond passively to user-inputted destinations, lacking proactive and predictive service recommendations and adjustments based on real-time contexts (such as flight / train delays, hub congestion, weather, and personalized passenger preferences and historical behavior). Furthermore, there is a lack of a unified intelligent Q&A portal supporting natural language interaction to answer various questions throughout the travel process. Moreover, intelligent Q&A often relies on rule-based matching (such as fixed-script responses), failing to address dynamic issues in complex hub scenarios (such as inquiries about itinerary adjustments due to temporary road closures or train delays).

[0008] Therefore, how to achieve end-to-end intelligent processing from "problem / requirement" to "service completion" and resolve the fundamental contradiction between service fragmentation and demand diversification has become a technical problem that urgently needs to be solved by those skilled in the art. Summary of the Invention

[0009] In view of the above-mentioned deficiencies of the prior art, the present invention provides a working method for a hub AI travel assistant based on multimodal fusion and MaaS collaboration. The purpose is to achieve end-to-end intelligent processing from "problem / requirement" to "service completion" through a unified AI assistant interaction interface, thereby resolving the fundamental contradiction between service fragmentation and demand diversification.

[0010] To achieve the above objectives, this invention discloses a working method for a hub AI travel assistant based on multimodal fusion and MaaS collaboration, comprising the following steps:

[0011] Step 1: Collect user needs through the human-computer interaction interface, understand the user's intent through the natural language processing (NLP) module, and use the service orchestration engine to decompose the user's needs into several sub-services;

[0012] Step 2: Collect the user's personalized preferences based on all the sub-services;

[0013] Step 3: Based on the different sub-services and the user's personalized preferences, automatically call the internal algorithm module and the external service provider interface in a logical order;

[0014] Step 4: Utilize the AI ​​big data model to perform fusion analysis on the multi-dimensional data obtained from internal algorithm modules and external service provider interfaces, calculate the comprehensive utility of each travel option in real time, and rank and recommend it.

[0015] The ranking recommendation is based on the personalized preferences, providing recommended travel service options.

[0016] Step 5: After the user confirms the travel service method through the human-computer interaction interface, the corresponding sub-service is implemented through the external service provider interface.

[0017] Preferably, in step 1, the user's audio data is collected via a microphone, and the language in the audio data is converted into text information. The Natural Language Processing (NLP) module then interprets the user's intent.

[0018] Alternatively, video data of the user can be captured via a camera, and the sign language in the video data can be converted into text. The natural language processing (NLP) module then interprets the user's intention.

[0019] Alternatively, the text information can be collected from the user via a keyboard or handwriting tablet, and then input into the Natural Language Processing (NLP) module to understand the user's intent.

[0020] Preferably, in step 1, the sub-services include query service, price comparison service, seat locking service, order placement service, payment service, navigation service, and / or pace guidance service.

[0021] Preferably, in step 2, the personalized preferences include personalized user profiles and user travel origin / destination (OD).

[0022] The personalized user profile includes the user's time sensitivity, cost preference, historical choices, number of luggage items, number of companions, and / or payment method.

[0023] Preferably, step 3 includes:

[0024] The user's real-time location and current time are obtained by calling the Amap location API, Baidu Map location API, and system time interface.

[0025] Weather conditions are obtained by calling the interfaces of weather forecast software, including MoWeather.

[0026] The data on pedestrian density within the hub, the map within the hub, and the time to reach various modes of transportation within the hub are obtained by calling the indoor map API deployed within the hub.

[0027] Real-time dynamic information on transportation modes within the hub, including ride-hailing, taxis, rail transit, regular buses, passenger buses, and maglev, is obtained by calling the corresponding travel software of the hub;

[0028] The real-time dynamic information of the transportation mode includes queuing status, congestion level, real-time traffic conditions, departure timetable, frequency of service, and fare rules;

[0029] The estimated travel time for various modes of transportation is obtained by calling the interface of location software, including Amap.

[0030] Preferably, the travel plan includes various single or combined modes of transportation; the comprehensive utility value includes travel time, cost, queuing time, congestion level, and comfort.

[0031] Preferably, some of the sub-services include navigation guidance;

[0032] The navigation guidance specifically refers to:

[0033] After the user confirms and completes the booking and payment for the travel service, based on the user's real-time location, target mode of transportation / destination, and combined with information such as the density of people in the hub, route planning, and walking speed suitability, a precise navigation guidance service is provided through the human-computer interaction interface.

[0034] The precise navigation and guidance service includes walking route guidance within the hub, transfer node prompts, and pace-adaptive guidance, ensuring that users can efficiently and smoothly reach their target mode of transportation or travel destination.

[0035] More preferably, during the navigation guidance process, the internal algorithm module and external interface are continuously invoked to monitor the user's multidimensional information in real time;

[0036] The user's multidimensional information includes location status and dynamic changes in the target mode of transportation, specifically including: service delays, queue time updates, and road condition fluctuations.

[0037] The density of people in the hub and emergencies, including: abnormal weather, facility malfunctions;

[0038] If any abnormal situations are detected, specifically including flight cancellations, route congestion, or users deviating from the navigation route,

[0039] This triggers the exception handling mechanism, which backtracks to the multi-dimensional comparison and service recommendation process to regenerate a suitable travel plan;

[0040] If the monitoring status is normal, continue to advance the navigation guidance process.

[0041] More preferably, once the dynamic monitoring module confirms that the user has successfully arrived at the preset travel destination, it completes the closed loop of this travel service, records the entire process data of this trip, including travel time, service selection, and handling of emergencies, and simultaneously updates the corresponding personalized preferences and travel profile of the user to provide data support for subsequent intelligent service recommendations.

[0042] The beneficial effects of this invention are:

[0043] The application of this invention enables end-to-end intelligent processing from "problem / requirement" to "service completion", resolving the fundamental contradiction between service fragmentation and demand diversification.

[0044] The following will further explain the concept, specific structure, and technical effects of the present invention in conjunction with the accompanying drawings, so as to fully understand the purpose, features, and effects of the present invention. Attached Figure Description

[0045] Figure 1 A flowchart of an embodiment of the present invention is shown. Detailed Implementation

[0046] Example: Figure 1 As shown, the working method of the hub AI travel assistant based on multimodal fusion and MaaS collaboration includes the following steps:

[0047] Step 1: Collect user needs through the human-computer interaction interface, understand user intent through the natural language processing (NLP) module, and use the service orchestration engine to break down user needs into several sub-services;

[0048] Step 2: Collect users' personalized preferences based on all sub-services;

[0049] Step 3: Based on different sub-services and the user's personalized preferences, automatically call the internal algorithm module and the external service provider interface in a logical order, i.e., in the form of MCP or Functioncall;

[0050] In practical applications, the core of the internal algorithm module is multi-dimensional data acquisition and intelligent decision support. The internal algorithm module integrates existing technologies such as path planning and navigation algorithms, multi-objective optimization algorithms, real-time prediction and scheduling algorithms, and multi-modal data fusion algorithms.

[0051] Step 4: Utilize the AI ​​big data model to perform fusion analysis on the multi-dimensional data obtained from internal algorithm modules and external service provider interfaces, calculate the comprehensive utility of each travel option in real time, and rank and recommend it.

[0052] The ranking recommendation is based on personalized preferences, providing recommended travel service options; for example, when a flight is delayed and arrives at night, "customized bus" or "private car" options are recommended for families with large luggage, rather than the subway.

[0053] In practical applications, the core of AI big data models is multimodal data fusion analysis and intelligent recommendation. AI big data models are adapted and selected according to the needs of the scenario, such as some general multimodal big data models, transportation-specific big data models, and open-source multimodal big data models.

[0054] Step 5: After the user confirms the travel service method through the human-computer interaction interface, the corresponding sub-services are implemented through the external service provider interface. These include integrated services such as ticketing, order placement, navigation, and pace guidance.

[0055] This invention breaks down the barriers between the various stages of inquiry, reservation, payment, and navigation, realizing "one-sentence request, full-process service" and achieving integrated service execution and closed loop of "AI + Maas". Users can make complex requests through natural language (such as "help me find the fastest way to get to XX Hotel without spending more than 50 yuan").

[0056] This invention understands the user's intent through a Natural Language Processing (NLP) module, uses a service orchestration engine to break down the user's needs into several sub-services, and then automatically calls internal algorithm modules and external service provider interfaces in a logical order according to different sub-services, linking various service provider APIs to complete the entire chain from solution recommendation to ticket booking, vehicle reservation, electronic ticket generation, and payment with one click, automatically realizing integrated service execution and presenting the user with a coherent closed-loop service result.

[0057] In terms of algorithms, this invention differs from existing recommendation algorithms based on a single dimension by constructing a three-dimensional comparison system that includes personalized preferences, real-time environment, and hub features. It introduces hub-specific parameters such as baggage carrying coefficient and transfer comfort weight, and optimizes the recommendation priority of ride-hailing, taxis, customized buses, and rail transit in real time through a reinforcement learning model, thus solving the problem of insufficient adaptability of existing technologies.

[0058] This invention constructs a closed-loop system centered on "AI-driven one-stop, full-chain, proactive travel services" through the aforementioned technical means. Its innovation lies in the deep integration of "multi-mode dynamic intelligent comparison (AI comparison)" and "integrated service execution (AI+Maas)," enabling a unified AI assistant interface for end-to-end intelligent processing from "problem / requirement" to "service completion," thus resolving the fundamental contradiction between service fragmentation and diversified demands.

[0059] In some embodiments, in step 1, the user's audio data is collected via a microphone, and the language in the audio data is converted into text information and then the user's intent is understood by a natural language processing (NLP) module.

[0060] Alternatively, video data of the user can be captured via camera, and the sign language in the video data can be converted into text and then the user's intentions can be understood through a natural language processing (NLP) module.

[0061] Alternatively, text information can be collected from the user via keyboard or handwriting tablet, and then input into a natural language processing (NLP) module to understand the user's intent.

[0062] In some embodiments, in step 1, the sub-services include query service, price comparison service, seat locking service, order placement service, payment service, navigation service, and / or pace guidance service.

[0063] In some embodiments, in step 2, personalized preferences include personalized user profiles and user travel origin / destination (OD).

[0064] Personalized user profiles include the user's time sensitivity, cost preference, historical choices, amount of luggage, number of companions, and / or payment method.

[0065] In some embodiments, step 3 includes:

[0066] The user's real-time location and current time are obtained by calling the Amap location API, Baidu Map location API, and system time interface.

[0067] Weather conditions are obtained by calling the interfaces of weather forecast software, including MoWeather.

[0068] The data on pedestrian density within the hub, the map within the hub, and the time to reach various modes of transportation within the hub are obtained by calling the indoor map API deployed within the hub.

[0069] Real-time dynamic information on transportation modes within the hub, including ride-hailing, taxis, rail transit, regular buses, passenger buses, and maglev, is obtained by calling the corresponding travel software of the hub;

[0070] In practical applications, the travel software corresponding to the hub includes various software covering a country or a region, such as Shanghai Metro API, Didi API, or Suishenxing (Shanghai).

[0071] Real-time information on transportation modes includes queuing status, congestion levels, real-time traffic conditions, departure schedules, service frequency, and fare rules.

[0072] The estimated travel time for various modes of transportation is obtained by calling the interface of location software, including Amap.

[0073] In some embodiments, the travel plan includes various single or combined modes of transportation; the overall utility value includes travel time, cost, queue time, congestion level, and comfort.

[0074] In some embodiments, navigation guidance is included in several of the sub-services;

[0075] The navigation guidance specifically refers to:

[0076] After the user confirms and completes the booking and payment for the travel service, based on the user's real-time location, target mode of transportation / destination, and combined with information such as the density of people in the hub, route planning, and walking speed suitability, a precise navigation guidance service is provided through the human-computer interaction interface.

[0077] The precise navigation and guidance service includes walking route guidance within the hub, transfer node prompts, and pace-adaptive guidance, ensuring that users can efficiently and smoothly reach their target mode of transportation or travel destination.

[0078] In some embodiments, during the navigation guidance process, internal algorithm modules and external interfaces are continuously invoked to monitor the user's multidimensional information in real time;

[0079] The user's multidimensional information includes location status and dynamic changes in the target mode of transportation, specifically including: service delays, queue time updates, and road condition fluctuations.

[0080] The density of people in the hub and emergencies, including: abnormal weather, facility malfunctions;

[0081] If any abnormal situations are detected, specifically including flight cancellations, route congestion, or users deviating from the navigation route,

[0082] This triggers the exception handling mechanism, which backtracks to the multi-dimensional comparison and service recommendation process to regenerate a suitable travel plan;

[0083] If the monitoring status is normal, continue to advance the navigation guidance process.

[0084] In some embodiments, once the dynamic monitoring module confirms that the user has successfully arrived at the preset travel destination, the service loop for this trip is completed, and the entire process data of this trip is recorded, including the trip time, service selection, and handling of emergencies. The corresponding personalized preferences and travel profile of the user are updated simultaneously to provide data support for subsequent intelligent service recommendations.

[0085] The preferred embodiments of the present invention have been described in detail above. It should be understood that those skilled in the art can make numerous modifications and variations based on the concept of the present invention without creative effort. Therefore, all technical solutions that can be obtained by those skilled in the art based on the concept of the present invention through logical analysis, reasoning, or limited experimentation on the basis of existing technology should be within the scope of protection defined by the claims.

Claims

1. A working method for a hub AI travel assistant based on multimodal fusion and MaaS collaboration; characterized in that, Includes the following steps: Step 1: Collect user needs through the human-computer interaction interface, understand the user's intent through the natural language processing (NLP) module, and use the service orchestration engine to decompose the user's needs into several sub-services; Step 2: Collect the user's personalized preferences based on all the sub-services; Step 3: Based on the different sub-services and the user's personalized preferences, automatically call the internal algorithm module and the external service provider interface in a logical order; Step 4: Utilize the AI ​​big data model to perform fusion analysis on the multi-dimensional data obtained from internal algorithm modules and external service provider interfaces, calculate the comprehensive utility of each travel option in real time, and rank and recommend it. The ranking recommendation is based on the personalized preferences, providing recommended travel service options. Step 5: After the user confirms the travel service method through the human-computer interaction interface, the corresponding sub-service is implemented through the external service provider interface.

2. The working method of the hub AI travel assistant based on multimodal fusion and MaaS collaboration according to claim 1, characterized in that, In step 1, the user's audio data is collected via microphone, and the language in the audio data is converted into text information. Then, the Natural Language Processing (NLP) module understands the user's intent. Alternatively, video data of the user can be captured via a camera, and the sign language in the video data can be converted into text. The natural language processing (NLP) module then interprets the user's intention. Alternatively, the text information can be collected from the user via a keyboard or handwriting tablet, and then input into the Natural Language Processing (NLP) module to understand the user's intent.

3. The working method of the hub AI travel assistant based on multimodal fusion and MaaS collaboration according to claim 1, characterized in that, In step 1, the sub-services include query service, price comparison service, seat locking service, order placement service, payment service, navigation service and / or pace guidance service.

4. The working method of the hub AI travel assistant based on multimodal fusion and MaaS collaboration according to claim 1, characterized in that, In step 2, the personalized preferences include personalized user profiles and user travel origin / destination (OD). The personalized user profile includes the user's time sensitivity, cost preference, historical choices, number of luggage items, number of companions, and / or payment method.

5. The working method of the hub AI travel assistant based on multimodal fusion and MaaS collaboration according to claim 1, characterized in that, Step 3 includes: The user's real-time location and current time are obtained by calling the Amap location API, Baidu Map location API, and system time interface. Weather conditions are obtained by calling the interfaces of weather forecast software, including MoWeather. The data on pedestrian density within the hub, the map within the hub, and the time to reach various modes of transportation within the hub are obtained by calling the indoor map API deployed within the hub. Real-time dynamic information on transportation modes within the hub, including ride-hailing, taxis, rail transit, regular buses, passenger buses, and maglev, is obtained by calling the corresponding travel software of the hub; The real-time dynamic information of the transportation mode includes queuing status, congestion level, real-time traffic conditions, departure timetable, frequency of service, and fare rules; The estimated travel time for various modes of transportation is obtained by calling the interface of location software, including Amap.

6. The working method of the hub AI travel assistant based on multimodal fusion and MaaS collaboration according to claim 1, characterized in that, The travel options include various single or combined modes of transportation; the overall utility value includes travel time, cost, queuing time, congestion level, and comfort.

7. The working method of the hub AI travel assistant based on multimodal fusion and MaaS collaboration according to claim 1, characterized in that, Some of the sub-services include navigation guidance; The navigation guidance specifically refers to: After the user confirms and completes the booking and payment for the travel service, based on the user's real-time location, target mode of transportation / destination, and combined with information such as the density of people in the hub, route planning, and walking speed suitability, a precise navigation guidance service is provided through the human-computer interaction interface. The precise navigation and guidance service includes walking route guidance within the hub, transfer node prompts, and pace-adaptive guidance, ensuring that users can efficiently and smoothly reach their target mode of transportation or travel destination.

8. The working method of the hub AI travel assistant based on multimodal fusion and MaaS collaboration according to claim 7, characterized in that, During the navigation guidance process, the internal algorithm module and external interface are continuously invoked to monitor the user's multidimensional information in real time; The user's multidimensional information includes location status and dynamic changes in the target mode of transportation, specifically including: service delays, queue time updates, and road condition fluctuations. The density of people in the hub and emergencies, including: abnormal weather, facility malfunctions; If any abnormal situations are detected, specifically including flight cancellations, route congestion, or users deviating from the navigation route, This triggers the exception handling mechanism, which backtracks to the multi-dimensional comparison and service recommendation process to regenerate a suitable travel plan; If the monitoring status is normal, continue to advance the navigation guidance process.

9. The working method of the hub AI travel assistant based on multimodal fusion and MaaS collaboration according to claim 7, characterized in that, Once the dynamic monitoring module confirms that the user has successfully arrived at the preset travel destination, the service loop for this trip is completed, and all process data for this trip is recorded, including travel time, service selection, and handling of emergencies. The module also updates the user's personalized preferences and travel profile accordingly, providing data support for subsequent intelligent service recommendations.

Images