Vehicle-mounted self-driving road book generation and execution method and system based on multi-agent cooperation

Abstract

The disclosure provides a vehicle-mounted self-driving route book generation and execution method and system based on multi-agent cooperation, relating to the fields of intelligent networked vehicles and artificial intelligence, which comprises obtaining a user's route book planning request, a first agent cluster generating a structured route book scheme containing multiple nodes, alternative schemes and complete context based on user historical data, real-time requests and external information; synchronizing the structured route book scheme with complete context to the vehicle infotainment system of the target vehicle, the vehicle infotainment system calling a second agent, binding the received route book scheme with the vehicle navigation and optimizing the vehicle environment; the second agent interacting with the user in a multi-modal manner and triggering tasks when reaching the preset or system recommended key points; generating a picture-text travelogue automatically after the trip ends and updating the user's preferences based on the user's explicit and implicit feedback during the trip to optimize future recommendations. The disclosure improves the full-process automation from self-driving planning to execution.

Description

Technical Field

[0001] This disclosure relates to the fields of intelligent connected vehicles and artificial intelligence technology, specifically to a method and system for generating and executing in-vehicle autonomous driving route books based on multi-agent collaboration. Background Technology

[0002] The statements in this section are merely background information relating to this disclosure and do not necessarily constitute prior art.

[0003] Self-driving tours have become a mainstream travel mode, leading to a growing demand for route planning (roadbook) services. Existing technological solutions include mobile internet-based travel guide apps such as "Mafengwo" and "Qyer." These apps offer a vast amount of user-generated content (UGC) and preset roadbook templates. Users can browse text and image guides and manually plan their itineraries through mobile apps. Simple roadbook functions are integrated into in-vehicle infotainment systems or navigation software, such as "Aha Roadbook" and other car infotainment applications, or the self-driving exploration functions built into some high-end vehicles. These solutions offer a preliminary combination of roadbooks and navigation, supporting one-click navigation initiation.

[0004] However, the existing solutions mentioned above still have the following limitations: (1) Planning and execution are disconnected: After completing the planning on the mobile phone, the user needs to manually input the location information into the vehicle navigation, which is cumbersome and the experience is not consistent. The existing solution lacks the ability to automatically and seamlessly transfer the planning status across devices.

[0005] (2) Lack of deep personalization and real-time adaptability: Existing in-vehicle route planning functions are mostly based on fixed rules or popular recommendations. They cannot make dynamic and intelligent route adjustments like human travel planners, taking into account users' multi-dimensional preferences (such as parent-child activities, photography, and food), real-time vehicle conditions (such as electric vehicle range), and unexpected situations during the journey (such as traffic jams, weather changes, and users' temporary interests).

[0006] (3) The human-computer interaction is inefficient and does not meet driving safety requirements: the mobile app is overloaded with information and is not suitable for browsing in driving scenarios; the existing in-vehicle interface is complex and inconvenient to operate while driving, and poses safety hazards. There is a lack of an interaction paradigm that is voice-centric and safety-oriented.

[0007] (4) The system is closed and lacks an ecosystem and data loop: The existing solution is only a one-way information push tool. It cannot learn and optimize itself through user feedback during the trip (such as stay duration and photo behavior) and sharing data after the trip. It also cannot effectively connect offline commercial services to form a service ecosystem that can evolve sustainably. Summary of the Invention

[0008] To address the aforementioned issues, this disclosure proposes a method and system for generating and executing in-vehicle autonomous driving route books based on multi-agent collaboration. It constructs a two-layer intelligent architecture of "cloud planning - vehicle execution" and achieves state synchronization and collaboration between the two ends through a unified user travel context.

[0009] According to some embodiments, the present disclosure adopts the following technical solutions: A method for generating and executing in-vehicle autonomous driving route books based on multi-agent cooperation includes: Upon receiving the user's route planning request, the first intelligent agent cluster generates a structured route plan containing multiple nodes, alternative solutions, and a complete context based on the user's historical data, real-time requests, and external information. The structured route plan with complete context is synchronized to the vehicle's infotainment system. The infotainment system then calls a second intelligent agent to bind the received route plan with the in-vehicle navigation system, thereby optimizing the in-vehicle environment. The second intelligent agent interacts with the user in a multimodal manner, triggering tasks when reaching preset or system-recommended key points; after the trip, it automatically generates a travelogue with pictures and text, and updates user preferences based on the user's explicit and implicit feedback during the trip to optimize future recommendations.

[0010] According to some embodiments, the present disclosure adopts the following technical solutions: A multi-agent collaborative in-vehicle autonomous driving route book generation and execution system includes: The route planning module is used to obtain the user's route planning request. The first intelligent agent cluster generates a structured route plan containing multiple nodes, alternative solutions and complete context based on the user's historical data, real-time requests and external information. The cross-platform execution module is used to synchronize the structured route plan with complete context to the vehicle's infotainment system. The infotainment system calls the second intelligent agent to bind the received route plan with the vehicle navigation system and optimize the vehicle environment. The interaction module is adjusted so that the second intelligent agent can interact with the user in a multimodal manner, triggering tasks when reaching preset or system-recommended key points; after the trip ends, a travelogue with pictures and text is automatically generated, and user preferences are updated based on the explicit and implicit feedback of the user during the trip to optimize future recommendations.

[0011] According to some embodiments, the present disclosure adopts the following technical solutions: A computer program product includes a computer program that, when executed by a processor, implements the aforementioned method for generating and executing in-vehicle autonomous driving route books based on multi-agent collaboration.

[0012] According to some embodiments, the present disclosure adopts the following technical solutions: A non-transitory computer-readable storage medium is provided for storing computer instructions, which, when executed by a processor, implement the aforementioned method for generating and executing vehicle-mounted autonomous driving route books based on multi-agent cooperation.

[0013] According to some embodiments, the present disclosure adopts the following technical solutions: An electronic device includes a processor, a memory, and a computer program; wherein the processor is connected to the memory, the computer program is stored in the memory, and when the electronic device is running, the processor executes the computer program stored in the memory to enable the electronic device to implement the method for generating and executing vehicle-mounted self-driving route books based on multi-agent collaboration.

[0014] Compared with the prior art, the beneficial effects of this disclosure are as follows: The disclosed method for generating and executing in-vehicle self-driving route books based on multi-agent collaboration achieves seamless integration of "mobile phone planning and vehicle execution" through a unified travel context and automatic synchronization mechanism, reducing user operation steps by more than 70% and significantly improving the smoothness of the experience.

[0015] The disclosed method for generating and executing in-vehicle self-driving route books based on multi-agent collaboration adopts a multi-agent collaborative architecture, which can mine scenic spot recommendations that match the user's deep preferences from massive information, and dynamically replan based on real-time vehicle conditions, road conditions, weather and user status. The route satisfaction rate has been improved by more than 50% in simulation tests.

[0016] The disclosed method for generating and executing in-vehicle autonomous driving route books based on multi-agent collaboration provides a second AI agent and interactive interface for the vehicle's infotainment system. It adheres to voice as the primary method and vision as a secondary method, which significantly reduces the driver's manual operation and attention distraction during driving. It meets the requirements of ASIL (Automotive Safety Integrity Level) standards for in-vehicle applications, and the interaction is safe and efficient.

[0017] This disclosed method for generating and executing in-vehicle self-driving route books based on multi-agent collaboration enables the system to continuously learn and iterate its recommendation algorithm through closed-loop collection and analysis of trip data. Simultaneously, it provides standardized interfaces for merchant service integration, building a thriving in-vehicle travel service ecosystem. Attached Figure Description

[0018] The accompanying drawings, which form part of this disclosure, are used to provide a further understanding of this disclosure. The illustrative embodiments of this disclosure and their descriptions are used to explain this disclosure and do not constitute an undue limitation of this disclosure.

[0019] Figure 1 This is a schematic diagram of the overall architecture of the in-vehicle autonomous driving route book generation and execution method based on multi-agent collaboration, according to an embodiment of this disclosure. Figure 2 This is a flowchart illustrating the collaborative process of a cloud-based multi-agent engine according to an embodiment of this disclosure. Figure 3 This is a flowchart illustrating the dynamic adjustment of the route book on the vehicle's infotainment system according to an embodiment of this disclosure. Figure 4 This is a flowchart of an automatic check-in method based on vision and geofencing, according to an embodiment of the present disclosure. Detailed Implementation

[0020] The present disclosure will be further described below with reference to the accompanying drawings and embodiments.

[0021] It should be noted that the following detailed descriptions are illustrative and intended to provide further explanation of this disclosure. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this disclosure pertains.

[0022] It should be noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the exemplary embodiments according to this disclosure. As used herein, the singular form is intended to include the plural form as well, unless the context clearly indicates otherwise. Furthermore, it should be understood that when the terms “comprising” and / or “including” are used in this specification, they indicate the presence of features, steps, operations, devices, components, and / or combinations thereof.

[0023] Example 1 One embodiment of this disclosure provides a method for generating and executing in-vehicle autonomous driving route books based on multi-agent cooperation, the method steps of which include: Step 1: Obtain the user's route planning request. The first intelligent agent cluster generates a structured route plan containing multiple nodes, alternative solutions, and complete context based on the user's historical data, real-time requests, and external information. Step 2: Synchronize the complete contextual structured route plan to the target vehicle's infotainment system. The infotainment system then calls the second intelligent agent to bind the received route plan with the in-vehicle navigation system, thereby optimizing the in-vehicle environment. Step 3: The second intelligent agent interacts with the user in a multimodal manner, triggering tasks when reaching preset or system-recommended key points; after the trip, it automatically generates a travelogue with pictures and text, and updates user preferences based on the user's explicit and implicit feedback during the trip, optimizing future recommendations.

[0024] As one embodiment, this disclosure presents a method for generating and executing in-vehicle self-driving route books based on multi-agent collaboration. It constructs a two-layer intelligent architecture of "cloud planning - vehicle execution," and achieves state synchronization and collaboration between the two ends through a unified user travel context. The overall system consists of three main parts: a mobile terminal, a cloud platform, and a vehicle terminal, which collaborate through user accounts and the network. The mobile terminal initiates requests and previews results; the core of the cloud platform is a multi-agent cluster responsible for generating route books; the vehicle terminal is responsible for final execution, interaction, and data collection, forming a closed loop of "planning-execution-feedback." The specific implementation process is as follows: Step 1: Obtain the user's route planning request. The first intelligent agent cluster generates a structured route plan containing multiple nodes, alternative solutions, and complete context based on the user's historical data, real-time requests, and external information. As one example, users initiate planning requests via natural language or a graphical interface on mobile terminals (such as mobile phones). The first AI intelligent agent cluster deployed in the cloud (including a central scheduling agent, route planning agent, point of interest mining agent, risk assessment agent, and cost and resource intelligent agent) works collaboratively to generate a structured route plan containing multiple nodes, alternative solutions, and complete context based on user historical data, real-time requests, and massive amounts of external information.

[0025] Specifically, after obtaining the user's route planning request, the central scheduling agent parses the request task, calls the path planning agent to generate a basic path framework, then calls the point of interest mining agent to fill in personalized POIs based on the path framework and the user's historical data, then calls the risk assessment agent to assess safety and compliance, and finally calls the cost and resource agent to calculate the budget and resource consumption. All the call results are then fused and conflict-resolved to generate a structured routing scheme and encapsulate it into a structured route plan with complete context.

[0026] Furthermore, the central dispatching agent receives users' route planning requests (supporting multiple input formats such as text, voice, and forms), analyzes the user's core needs through natural language processing (NLP) technology, and extracts key parameters, including departure point, destination, trip duration, budget range, interests and preferences (natural landscapes / cultural history / food and shopping), companions (elderly / children / pets), and special needs (accessibility facilities, charging stations, nucleic acid test requirements, etc.). It transforms the fuzzy needs into a structured Task Graph, which is broken down into four ordered sub-tasks: "route generation, POI filling, risk assessment, and cost accounting," clarifying the dependencies and execution priorities of each sub-task.

[0027] Based on the subtask type, the corresponding specialized intelligent agent is invoked through the routing layer. The execution status of each intelligent agent (idle / in execution / abnormal / completed) is monitored in real time to achieve load balancing (using the Least Connection strategy to select the optimal intelligent agent instance). In case of execution timeouts, exceptions, or errors, a retry mechanism or degradation strategy is triggered to ensure that tasks are not interrupted. At the same time, an intelligent agent registry center is maintained to support the dynamic access and deactivation of intelligent agents, improving system scalability.

[0028] The system receives the output results from various specialized intelligent agents, performs data cleaning, format standardization, and logical verification. Then, based on preset rules and machine learning models, it fuses the multi-dimensional results to form a unified structured routing scheme. For conflict scenarios in the output results of various intelligent agents (such as conflicts between the shortest route in path planning and the route for filling points of interest (POIs), conflicts between cost calculation exceeding the user's budget and POI priority, and conflicts between the safe route in risk assessment and the optimal route in path planning), a hybrid conflict resolution strategy of "rule engine + auction algorithm" is adopted. Combined with user demand priorities (such as security > personalization > cost), conflict arbitration and scheme adjustments are performed to ensure that the fused result meets the user's core needs. The conflict resolution process includes: conflict detection → conflict classification → policy matching → execution adjustment → result verification. Specific rules can be dynamically optimized based on user feedback.

[0029] The integrated structured routing scheme is encapsulated according to a preset structured format (including route details, POI details, risk warnings, budget details, resource suggestions, etc.) to generate a complete contextual structured route book. It also supports the visualization of the route book (map embedding, trip timeline) and provides feedback to users to receive their modification opinions and trigger a second iteration optimization.

[0030] As one example, a route planning agent is invoked. Based on the user's input of the origin and destination, and combined with high-precision map data (road network, traffic rules, lane information, etc.), the route planning agent adopts a path planning algorithm adapted to the scenario: in static known environments, the A* algorithm is used (balancing efficiency and optimality), and in dynamic environments (such as real-time traffic and temporary control), the RRT* algorithm is used (quickly responding to environmental changes and achieving probabilistic optimal path planning), generating multiple candidate paths (shortest distance, shortest time, most comfortable route, etc.). Each path includes basic information such as road segment details, estimated time, and areas traversed, forming a standardized basic path framework.

[0031] Furthermore, by integrating real-time traffic data (congestion, accident warnings, construction control), weather data (rainfall, snowfall, high temperatures, etc.), and road infrastructure data (road surface quality, bridge height, weight restrictions), the basic route is dynamically adjusted to avoid congested and dangerous sections and optimize route time. For self-driving scenarios, charging stations and reasonable stopping points in service areas are planned simultaneously. For walking / public transportation scenarios, convenient transfer points and barrier-free access are planned to ensure the feasibility and comfort of the route.

[0032] The generated basic path framework will be output in a unified format, including path ID, departure / arrival time, coordinates of nodes along the way, road segment information, and travel mode adaptation instructions, providing standardized data support for subsequent POI filling, risk assessment, and cost accounting, and ensuring data interoperability between various intelligent agents.

[0033] As one example, an Interest Point Mining Agent is invoked. Based on the basic path framework generated by the path planning agent, the Interest Point Mining Agent combines user historical data and scenario requirements to mine personalized and highly matching POIs, thereby enabling personalized filling of the route book and improving the user experience.

[0034] Specifically, it accesses users' historical route data, browsing history, favorites preferences, evaluation feedback, etc., and combines them with users' current interests and preferences to build accurate user profiles, extract core interest tags (such as preference for ancient towns, food exploration, outdoor hiking, parent-child activities, etc.), and analyzes users' preference weights for POIs (such as prioritizing POIs with a rating ≥ 4.5 points and a distance of ≤ 1 kilometer from the route).

[0035] Based on the areas traversed by the basic route, the system accesses a POI database (covering all categories including natural landscapes, historical sites, dining and accommodation, shopping and entertainment, and medical services) to discover POIs that match the user's interest tags. The system then performs multi-dimensional filtering based on factors such as POI popularity, rating, distance from the route, opening hours, and suitable scenarios (e.g., prioritizing accessible POIs for elderly users) to select the optimal POI list.

[0036] The selected POIs are logically embedded into the basic route framework to optimize the itinerary pace, avoid excessive distances between POIs and route detours, and ensure the rationality and smoothness of the itinerary. At the same time, detailed information is added to each POI, including address, opening hours, ticket price, reasons for recommendation, user reviews, and adaptation tips (such as pet-friendly, wheelchair accessible), to achieve accurate filling of personalized POIs.

[0037] The system synchronizes the latest status of POIs in real time (such as temporary closures, ticket price increases, and event information), and dynamically adjusts the POI list based on real-time user feedback to ensure the accuracy and timeliness of POI information.

[0038] As one example, a risk assessment agent is invoked. This agent is responsible for conducting a comprehensive security and compliance assessment of the basic route and personalized Points of Interest (POIs) in the travel itinerary, identifying potential risks and providing mitigation suggestions to ensure user travel safety and that the travel itinerary complies with relevant regulations, as detailed below: At the route level, assess safety hazards along the route (such as landslide risk on mountain roads, insufficient lighting at night, and security risks on remote sections), traffic risks (congestion, accident-prone sections, and the impact of severe weather), and infrastructure risks (road damage, aging bridges, and malfunctioning charging stations). At the Point of Interest (POI) level, assess the POI's safety environment (security, fire safety facilities, and emergency access) and suitability risks (such as weather suitability for outdoor POIs and safety protection for family-friendly POIs). For companions, considering special groups such as the elderly, children, and pets, assess the trip's suitability risks (such as the difficulty of the hiking route and the accessibility of the POI).

[0039] The assessment covers relevant compliance requirements, including traffic rules (traffic restrictions, speed limits, and prohibited sections), scenic area regulations (reservation requirements, ticket policies, and prohibited items), local epidemic prevention policies (nucleic acid testing requirements and quarantine regulations), and environmental regulations (such as fire bans in outdoor areas and garbage sorting requirements). This ensures that the route plan complies with national laws and regulations as well as relevant local regulations, and avoids the risk of violations.

[0040] The identified risks are categorized into three levels of severity: high, medium, and low. For each level of risk, specific avoidance suggestions are generated (such as recommending detours for high-risk road sections, alternative routes for temporarily closed POIs, and adjustments to itineraries for severe weather). The specific location, impact range, and countermeasures of the risk points are also marked to ensure that users can anticipate and avoid risks in advance.

[0041] By accessing real-time risk data (such as sudden accidents, extreme weather, and upgraded epidemic prevention and control measures), dynamic risk reviews are conducted on the generated route plans. If new high-risk points are identified, early warning information is promptly pushed out, and adjustment suggestions are provided to ensure the safety of the trip.

[0042] As one example, the cost and resource agent is invoked. This agent is responsible for calculating various costs associated with the itinerary, assessing resource consumption, and generating detailed budget breakdowns and resource recommendations to help users effectively control costs and optimize resource allocation. Specifically: Based on detailed route information, POI data, and mode of transportation, a comprehensive cost calculation is performed, covering transportation costs (fuel and tolls), accommodation costs (hotel / guesthouse fees, considering trip duration and accommodation standards), food costs (calculated based on POI restaurant recommendations and average daily spending), entrance fees (entrance fees and activity fees for each POI), and other costs (parking fees, charging fees, emergency supplies, etc.). A 10%-20% contingency reserve is also provided to handle unexpected expenses. The cost calculation formula is as follows: Total Cost = Transportation Cost + Accommodation Cost + Food Cost + Entrance Fee Cost + Other Costs + Contingency Reserve.

[0043] Based on the user's budget range, compare the calculated cost with the user's budget. If the cost exceeds the budget, provide targeted cost optimization suggestions (such as replacing high-priced POIs, adjusting accommodation standards, optimizing routes to reduce transportation costs, and choosing off-peak travel to reduce expenses). If the cost is below the budget, recommend value-added POIs or services to enhance the trip experience.

[0044] Resource Consumption Assessment: For different modes of travel, assess resource consumption, such as fuel consumption and electricity consumption (new energy vehicles) in self-driving scenarios, and physical exertion and material consumption in hiking scenarios. Generate resource consumption details and provide suggestions for reasonable resource allocation (such as planning charging nodes in self-driving scenarios and planning rest nodes in hiking scenarios).

[0045] Real-time synchronization of cost-related data (such as fuel price increases, hotel price adjustments, and ticket discounts) allows for dynamic adjustment of budget details, ensuring the accuracy of budget calculations. It also supports users manually adjusting budget allocations and simultaneously optimizing route plans (such as increasing the catering budget and reducing the ticket budget).

[0046] Ultimately, the central scheduling agent is responsible for integrating the outputs of various specialized agents, resolving conflicting results, and forming a unified and rational structured routing scheme, specifically implemented as follows: 1. Result Standardization Processing: Receive the output results from the route planning, POI mining, risk assessment, and cost accounting agents, and perform data cleaning (remove invalid data and correct errors), format standardization (organize data according to preset structured specifications), and logical verification (verify the correlation and rationality between the results, such as whether the POI is within the route range and whether the cost accounting matches the route length) to ensure data consistency and usability.

[0047] 2. Multi-dimensional result fusion: Based on user needs priority (which can be preset by the user or defaulted by the system, such as security > personalization > cost > efficiency), the route details, POI information, risk warnings, and budget details are organically integrated to generate a structured routing plan that includes "trip overview, daily trip details, POI details, risk warnings, budget details, and resource suggestions". This ensures that the plan meets the user's personalized needs while also taking into account security, compliance, and cost control.

[0048] 3. Conflict Resolution Mechanism: For conflict scenarios in the output results of various agents, a hierarchical resolution strategy is adopted: (1) Basic conflicts (such as conflicts between path and POI location): Adopt the principle of prioritizing path fine-tuning, adjust the POI embedding position without changing the core path, and avoid route reversal; (2) Core conflict (such as conflict between security risk and optimal path): Adopt the principle of security first, abandon the optimal path, choose an alternative path with controllable risk, and adjust POI and budget at the same time; (3) Budget conflict (such as accounting costs exceeding the user's budget): Adopt the priority ranking principle, prioritize the retention of high matching degree POIs and core paths, replace low priority POIs, and reduce accommodation / food standards to ensure that the budget is controllable; (4) Complex conflicts (multi-dimensional conflicts): The auction algorithm and rule engine are combined to construct a profit matrix. Pareto optimality is achieved through iterative adjustment to ensure that the conflict resolution results meet the core needs of users.

[0049] The cloud-based processing disclosed herein begins with the decomposition and scheduling of tasks by a central scheduling agent. Various specialized agents work in parallel or sequentially, ultimately merging and making decisions through the central scheduling agent to output the final solution. This embodies an intelligent architecture that combines division of labor and centralized decision-making.

[0050] Step 2: Synchronize the complete contextual structured route plan to the vehicle's infotainment system. The infotainment system then calls the second intelligent agent to bind the received route plan with the in-vehicle navigation system, thereby optimizing the in-vehicle environment. This disclosure discloses that a data packet containing a structured route plan with complete context is automatically synchronized to the vehicle's infotainment system via an account. The vehicle's infotainment system receives and parses the data packet, invokes a second intelligent agent, reads vehicle bus data to adjust the vehicle's status, executes the route plan, and continuously monitors the user's real-time planning information. When an adjustment request is detected, the second intelligent agent analyzes the impact and generates an adjustment plan, solicits user confirmation via voice or card, and updates the navigation route and route plan display after the user agrees, and records it in the decision log.

[0051] As one example, the structured route plan is encapsulated in a vehicle infotainment system adaptation format (supporting mainstream vehicle infotainment systems such as Linux, QNX, and Android Automotive OS). The data package contains core route plan information (route details, POI coordinates and details, trip sequence, risk warnings, budget details), adaptation parameters (navigation preferences, vehicle adaptation requirements), and verification information (data signature, timestamp) to ensure the integrity and security of the data package and prevent data tampering or loss during transmission.

[0052] The system automatically associates the user's bound vehicle information (VIN, vehicle device ID) and synchronously verifies the user's account permissions (ensuring the user is bound to the vehicle or is an authorized user) to prevent unauthorized accounts from synchronizing roadbook data. After successful verification, the system uses an encrypted transmission protocol (HTTPS+AES encryption) to push data packets to the vehicle system via vehicle-to-everything (V2X) or mobile network (4G / 5G). The system monitors the transmission status in real time during transmission, and if a transmission interruption occurs, it automatically triggers a breakpoint resume mechanism to ensure successful data synchronization.

[0053] After receiving the data packet, the vehicle system automatically reports the synchronization status (success / failure / partial reception) to the central dispatching agent. If synchronization fails, the system automatically analyzes the cause of failure (network error, vehicle system offline, data packet incompatibility) and triggers a retry mechanism (retrying every 30 seconds, up to 3 times). If multiple retries fail, a synchronization failure message is pushed to the user, providing manual synchronization options (such as QR code synchronization, USB synchronization) to ensure that the route plan can be synchronized to the vehicle system normally.

[0054] Furthermore, the second intelligent agent, as the core execution and decision-making node of the vehicle system, is responsible for reading vehicle bus data, adjusting vehicle status, executing route plans, and handling real-time user adjustment requests, achieving dynamic adaptation between route execution and vehicle status. Specifically, this includes: (1) Vehicle bus data reading and analysis: The vehicle's core data, including the vehicle's current location, speed, fuel / electricity, tire pressure, engine status, air conditioning status, and navigation status, is read in real time through the vehicle's CAN bus and Ethernet bus. At the same time, the vehicle's current status is analyzed in conjunction with the route requirements and POI stop requirements (such as charging and rest) in the route plan, and the vehicle status adjustment suggestions are generated.

[0055] (2) Automatic vehicle status adjustment: Based on the route plan and vehicle bus data analysis results, the vehicle control interface is automatically invoked to adjust the vehicle status to adapt to the route plan execution requirements, specifically including: 1. Navigation Integration: Synchronize the route from the map to the vehicle's navigation system, automatically start navigation, provide real-time traffic updates, turn prompts, and POI arrival reminders. Adjust the navigation broadcast rhythm according to the vehicle's current speed to ensure that the driver can clearly obtain navigation information. 2. Power and Energy Consumption Adjustment: For long-distance self-driving tours, the vehicle's power mode (such as energy-saving mode or sport mode) is adjusted based on the route's gradient and distance to optimize energy consumption; if the route map includes charging POIs, the driver is reminded in advance that the battery is low, and charging time is planned simultaneously. Upon arrival at the charging POI, the vehicle's charging system is automatically activated (for new energy vehicles). 3. Comfort Adjustment: Based on the trip duration and weather conditions in the areas traversed, the system automatically adjusts the air conditioning temperature and fan speed, and links the seat heating / ventilation functions to enhance driving and riding comfort; if a rest stop is reached, the system automatically adjusts the vehicle's parking status and shuts down unnecessary electrical equipment to reduce energy consumption. 4. Safety Status Adjustment: Based on the risk warnings in the road map (such as mountain roads and night driving), the vehicle's safety assistance functions (such as lane keeping, adaptive cruise control, and automatic headlight adjustment at night) are automatically activated to monitor the vehicle's surrounding environment in real time and avoid safety risks.

[0056] 5. Route Plan Execution and Monitoring: Execute navigation, POI stops, route switching, and other operations in real time according to the time sequence requirements of the route plan. At the same time, continuously monitor the execution status of the route plan (such as whether it deviates from the route, whether it arrives at the POI on time, and whether the vehicle status is normal). If abnormal situations such as deviation from the route or vehicle malfunction occur, promptly push notifications to the user, analyze the causes of the abnormalities, and generate preliminary adjustment suggestions.

[0057] 6. User Adjustment Request Processing: Continuously monitor real-time user planning adjustment requests (supporting voice requests such as "change destination" and "avoid congested sections" and touch operation requests such as manually dragging routes and modifying POI dwell time). Upon receiving a request, immediately analyze the impact of the adjustment request on the current route plan, vehicle status, trip duration, and budget, and generate 2-3 adjustment plans (such as optimal route adjustment, POI replacement, and trip timing adjustment), clarifying the advantages and disadvantages of each plan (such as changes in time consumption, cost changes, and vehicle adaptability).

[0058] 7. Adjustment Plan Confirmation and Execution: The generated adjustment plan will be presented to the user for confirmation via voice broadcast on the vehicle's infotainment system (clear, concise, and adaptable to driving scenarios) or via a card displayed on the vehicle's screen (visualized display, supporting one-click selection). After the user confirms via voice response (e.g., "Select Plan One") or touch operation, the second intelligent agent immediately updates the vehicle's navigation route and roadbook display, and simultaneously feeds back the adjusted roadbook plan to the central dispatch intelligent agent, which then updates the roadbook data in the cloud. If the user refuses the adjustment, the original roadbook plan will be maintained, and the reason for the user's refusal will be recorded for subsequent personalized optimization.

[0059] Step 3: The second intelligent agent interacts with the user in a multimodal manner, triggering tasks when reaching preset or system-recommended key points; after the trip ends, it automatically generates a travelogue with pictures and text, and updates user preferences based on the user's explicit and implicit feedback during the trip to optimize future recommendations.

[0060] The second intelligent agent interacts with the user in a multimodal manner, triggering tasks upon reaching preset or system-recommended key points, including: The system continuously monitors the vehicle's GPS location. If the vehicle enters a preset location, it activates the vehicle's camera to capture scene images. The system then extracts image features through a visual recognition model and matches these features with a preset database. If a match is successful, the system automatically checks in, records the time, location, and associated scene images or videos, and updates the status of the route map nodes.

[0061] As one example, it supports voice interaction (the core interaction method, adapted to driving scenarios, supporting continuous dialogue and dialect recognition) and touch interaction (an auxiliary interaction method, used to view details and confirm adjustment plans). The voice interaction response time is ≤1 second, and the touch interaction response time is ≤0.5 seconds, ensuring convenient and efficient operation.

[0062] For key points in the execution of the route plan (such as approaching a POI, deviating from the route, low vehicle battery, or approaching a risk point), prompts are pushed to the user through voice broadcasts and screen prompts. The prompts are concise and clear, avoiding redundancy. For the confirmation of the adjustment plan, the voice broadcast clearly explains the core information of the plan (such as "Avoiding congested sections, estimated to take 15 minutes longer, confirm?"), and the screen cards simultaneously display the plan details, making it convenient for users to make a quick judgment.

[0063] Furthermore, it comprehensively records key operational and decision-making information, including the synchronization time, synchronization status, and parsing results of the route data package; vehicle bus data read by the second intelligent agent, vehicle status adjustment instructions, and route execution status; user adjustment request content, adjustment scheme analysis results, and user confirmation status; route update records, abnormal situations (such as synchronization failure, route deviation, and vehicle malfunction), and processing results.

[0064] Furthermore, the log format is standardized: the log is recorded in a structured format, including timestamp, operation type, operation content, execution result, associated objects (such as route ID, vehicle VIN, user ID), and exception information (if any), to ensure the readability and traceability of the log; the log storage adopts a dual storage mode of local + cloud, with local storage retaining the logs for the most recent 3 months, and cloud storage retaining them long-term (encrypted storage to protect user privacy).

[0065] Furthermore, log application scenarios include: supporting system maintenance personnel to query logs and troubleshoot abnormal issues during route synchronization and execution (such as synchronization failure reasons or abnormal vehicle adjustment command execution); supporting the central scheduling agent to access log data, analyze user adjustment preferences and route execution pain points for model optimization and personalized recommendations; and supporting users to query their own trip logs to trace route adjustment records and vehicle status changes. Furthermore, user feedback collection: Users can provide feedback on problems encountered during the route planning process (such as navigation deviations, incorrect POI information, and unreasonable vehicle adjustments) via voice or touch. The collected feedback information is automatically synchronized to the central dispatching agent for route planning optimization and iteration of the second agent model.

[0066] Example 2 One embodiment of this disclosure provides an in-vehicle autonomous driving route book generation and execution system based on multi-agent cooperation, including: The route planning module is used to obtain the user's route planning request. The first intelligent agent cluster generates a structured route plan containing multiple nodes, alternative solutions and complete context based on the user's historical data, real-time requests and external information. The cross-platform execution module is used to synchronize the structured route plan with complete context to the vehicle's infotainment system. The infotainment system calls the second intelligent agent to bind the received route plan with the vehicle navigation system and optimize the vehicle environment. The interaction module is adjusted so that the second intelligent agent can interact with the user in a multimodal manner, triggering tasks when reaching preset or system-recommended key points; after the trip ends, a travelogue with pictures and text is automatically generated, and user preferences are updated based on the explicit and implicit feedback of the user during the trip to optimize future recommendations.

[0067] Example 3 One embodiment of this disclosure provides a computer program product, including a computer program that, when executed by a processor, implements the method for generating and executing vehicle-mounted autonomous driving route books based on multi-agent collaboration.

[0068] Example 4 One embodiment of this disclosure provides a non-transitory computer-readable storage medium for storing computer instructions. When these computer instructions are executed by a processor, they implement the method for generating and executing vehicle-mounted autonomous driving route books based on multi-agent collaboration.

[0069] Example 5 One embodiment of this disclosure provides an electronic device, including: a processor, a memory, and a computer program; wherein the processor is connected to the memory, and the computer program is stored in the memory. When the electronic device is running, the processor executes the computer program stored in the memory to enable the electronic device to execute the method for generating and executing vehicle-mounted self-driving route books based on multi-agent collaboration.

[0070] This disclosure is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this disclosure. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create a machine for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0071] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0072] While the specific embodiments of this disclosure have been described above in conjunction with the accompanying drawings, this is not intended to limit the scope of protection of this disclosure. Those skilled in the art should understand that various modifications or variations that can be made by those skilled in the art without creative effort based on the technical solutions of this disclosure are still within the scope of protection of this disclosure.

Claims

1. A method for generating and executing vehicle-mounted autonomous driving route books based on multi-agent collaboration, characterized in that, include: Upon receiving the user's route planning request, the first intelligent agent cluster generates a structured route plan containing multiple nodes, alternative solutions, and a complete context based on the user's historical data, real-time requests, and external information. The structured route plan with complete context is synchronized to the vehicle's infotainment system. The infotainment system then calls a second intelligent agent to bind the received route plan with the in-vehicle navigation system, thereby optimizing the in-vehicle environment. The second intelligent agent interacts with the user in a multimodal manner, triggering tasks when reaching preset or system-recommended key points; after the trip, it automatically generates a travelogue with pictures and text, and updates user preferences based on the user's explicit and implicit feedback during the trip to optimize future recommendations.

2. The method for generating and executing vehicle-mounted autonomous driving route books based on multi-agent collaboration as described in claim 1, characterized in that, The first intelligent agent cluster includes a central scheduling intelligent agent, a route planning intelligent agent, an interest point mining intelligent agent, a risk assessment intelligent agent, and a cost and resource intelligent agent.

3. The method for generating and executing vehicle-mounted autonomous driving route books based on multi-agent collaboration as described in claim 1, characterized in that, After obtaining the user's route planning request, the central scheduling agent parses the request task, calls the path planning agent to generate a basic path framework, then calls the point of interest mining agent to fill in personalized POIs based on the path framework and the user's historical data, then calls the risk assessment agent to assess safety and compliance, and finally calls the cost and resource agent to calculate the budget and resource consumption. All the call results are then fused and conflict-resolved to generate a structured routing scheme and encapsulate it into a structured route plan with complete context.

4. The method for generating and executing vehicle-mounted autonomous driving route books based on multi-agent collaboration as described in claim 1, characterized in that, The data packet containing the structured route plan with complete context is automatically synchronized to the vehicle's infotainment system via the account. The vehicle's infotainment system receives and parses the data packet, calls the second intelligent agent, reads the vehicle bus data to adjust the vehicle status, and executes the route plan. It continuously monitors the user's real-time planning information. When an adjustment request is detected, the second intelligent agent analyzes the impact and generates an adjustment plan. It then seeks the user's confirmation via voice or card. Once the user agrees, the navigation route and route plan display are updated and recorded in the decision log.

5. The method for generating and executing vehicle-mounted autonomous driving route books based on multi-agent collaboration as described in claim 1, characterized in that, The second intelligent agent interacts with the user in a multimodal manner, triggering tasks upon reaching preset or system-recommended key points, including: The system continuously monitors the vehicle's GPS location. If the vehicle enters a preset location, it activates the vehicle's camera to capture scene images. The system then extracts image features through a visual recognition model and matches these features with a preset database. If a match is successful, the system automatically checks in, records the time, location, and associated scene images or videos, and updates the status of the route map nodes.

6. The method for generating and executing vehicle-mounted autonomous driving route books based on multi-agent collaboration as described in claim 1, characterized in that, Once the route plan is completed, the vehicle system automatically integrates all nodes, timelines, and scene images of this route plan to generate historical route records, which are then fed back to the cloud to update the user profile for future recommendation tasks.

7. A vehicle-mounted autonomous driving route book generation and execution system based on multi-agent collaboration, characterized in that, include: The route planning module is used to obtain the user's route planning request. The first intelligent agent cluster generates a structured route plan containing multiple nodes, alternative solutions and complete context based on the user's historical data, real-time requests and external information. The cross-platform execution module is used to synchronize the structured route plan with complete context to the vehicle's infotainment system. The infotainment system calls the second intelligent agent to bind the received route plan with the vehicle navigation system and optimize the vehicle environment. The interaction module is adjusted so that the second intelligent agent can interact with the user in a multimodal manner, triggering tasks when reaching preset or system-recommended key points; after the trip ends, a travelogue with pictures and text is automatically generated, and user preferences are updated based on the explicit and implicit feedback of the user during the trip to optimize future recommendations.

8. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by the processor, it implements the method for generating and executing vehicle-mounted autonomous driving route books based on multi-agent cooperation as described in any one of claims 1-6.

9. A non-transitory computer-readable storage medium, characterized in that, The non-transitory computer-readable storage medium is used to store computer instructions, which, when executed by a processor, implement the method for generating and executing vehicle-mounted autonomous driving route books based on multi-agent cooperation as described in any one of claims 1-6.

10. An electronic device, characterized in that, include: The device includes a processor, a memory, and a computer program; wherein the processor is connected to the memory, the computer program is stored in the memory, and when the electronic device is running, the processor executes the computer program stored in the memory to enable the electronic device to perform the method for generating and executing vehicle-mounted self-driving route books based on multi-agent cooperation as described in any one of claims 1-6.

Images