A path calculation method based on a large model
By fusing and extracting features from multi-source traffic data using a large Transformer-based model, a dynamic path cost function is constructed. This solves the accuracy and personalization problems of existing path planning methods in complex environments, and achieves efficient and accurate path planning.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- INSPUR TIANYUAN COMM INFORMATION SYST CO LTD
- Filing Date
- 2026-02-10
- Publication Date
- 2026-06-12
AI Technical Summary
Existing route planning methods lack the ability to dynamically respond to real-time traffic conditions, user behavior preferences, and complex environmental factors, resulting in inaccurate and unpersonalized planning results. Furthermore, the failure to fully integrate multimodal traffic data restricts the level of intelligence.
A large model based on the Transformer architecture is used to perform semantic fusion and feature extraction on multi-source traffic data, generate dynamic environment representation, construct path cost function and solve for the optimal path by combining optimization algorithm, and output real-time updated path planning results.
It significantly improves the accuracy and robustness of route planning, can adaptively respond to sudden traffic events, reduce user travel time, improve the integration of personalized preferences, and enhance user experience.
Smart Images
Figure CN122192349A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of intelligent transportation technology, and in particular to a path calculation method based on a large model. Background Technology
[0002] Existing route planning methods are mostly based on static map data and simple rules (such as shortest path and fastest path), lacking the ability to dynamically respond to real-time traffic conditions, user behavior preferences, and complex environmental factors. For example, existing navigation systems typically rely on historical average speeds or limited real-time data, making it difficult to accurately predict sudden congestion, weather effects, special events, and other abnormal situations, leading to discrepancies between the planned results and actual traffic conditions. Furthermore, traditional methods have limited ability to fuse multimodal traffic data (such as video, text reports, and sensor data), failing to fully extract semantic information and correlation patterns from the data, thus restricting the level of intelligence in route planning. Summary of the Invention
[0003] To address the above technical problems, this invention provides a path calculation method based on a large model, which solves the problems of poor adaptability to complex traffic environments, inaccurate planning results, and low personalization in existing technologies.
[0004] The technical solution of this invention is: A path calculation method based on a large model is characterized by the following steps: acquiring multi-source traffic data; using a large model to perform semantic fusion and feature extraction on the multi-source data to generate a dynamic environment representation; constructing a path cost function based on the dynamic environment representation; solving for the optimal path using an optimization algorithm; and outputting the path planning result.
[0005] Furthermore, The multi-source traffic data includes various types of data such as real-time traffic flow data, historical traffic records, weather information, event reports, and user preference data. Furthermore, The large model is a multimodal pre-trained model based on the Transformer architecture, which can perform unified encoding and inference on text, images and structured data.
[0006] Furthermore, The path cost function comprehensively considers time, distance, congestion probability, user preference factors, and environmental uncertainty indicators.
[0007] Furthermore, The optimization algorithm includes graph search algorithms or reinforcement learning strategies.
[0008] Furthermore, Real-time updates and interactive adjustments, outputting updated results.
[0009] The beneficial effects of this invention are This invention achieves deep semantic fusion and dynamic reasoning of multi-source traffic data through a large model, significantly improving the accuracy and robustness of route planning. It can adaptively respond to sudden traffic events and complex environmental changes, reducing user travel time. Simultaneously, it supports personalized preference fusion, enhancing user experience. Experimental results show that in typical urban traffic scenarios, this method improves route planning accuracy by approximately 20% and reduces average travel time by 15% compared to traditional navigation systems. Attached Figure Description
[0010] Figure 1 This is a schematic diagram of the workflow of the present invention. Detailed Implementation
[0011] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are some embodiments of the present invention, but not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.
[0012] This invention relates to a path calculation method based on a large model. This method utilizes a large model to fuse and semantically understand multi-source traffic data. By constructing a dynamic path cost function and combining real-time traffic conditions, historical traffic data, and user preferences, it achieves efficient, accurate, and personalized path planning. This invention can effectively address complex scenarios such as traffic congestion and variable routes, improving travel efficiency and user experience.
[0013] Specifically, the following steps are included: 1. Acquire multi-source traffic data, including real-time traffic flow data, historical traffic records, weather information, event reports, and user preference data; 2. Utilize pre-trained large models to extract features and fuse semantics from multi-source data to generate dynamic environment representations; 3. Construct a path cost function based on dynamic environment representation. This function comprehensively considers time, distance, congestion probability, user preference factors, and uncertainty indicators. 4. Solve for the optimal path in the road network graph using graph search algorithms or reinforcement learning strategies; 5. Output path results and provide real-time update and interactive adjustment functions. The large model adopts a Transformer-based architecture, capable of unified encoding and inference for multimodal inputs such as text, images, and structured data.
[0014] The above description is merely a preferred embodiment of the present invention and is used only to illustrate the technical solution of the present invention, and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention are included within the scope of protection of the present invention.
Claims
1. A path calculation method based on a large model, characterized in that, By leveraging large-scale models to fuse and semantically understand multi-source traffic data, and constructing dynamic path cost functions, combined with real-time traffic conditions, historical traffic data, and user preferences, path planning is achieved.
2. The method according to claim 1, characterized in that, The specific steps are as follows: Acquire multi-source traffic data; utilize a large model to perform semantic fusion and feature extraction on the multi-source data to generate a dynamic environment representation; A path cost function is constructed based on the dynamic environment representation; the optimal path is solved using an optimization algorithm; and the path planning results are output.
3. The method according to claim 2, characterized in that, The multi-source traffic data includes real-time traffic flow data, historical traffic records, weather information, event reports, and user preference data.
4. The method according to claim 2, characterized in that, The large model is a multimodal pre-trained model based on the Transformer architecture, which performs unified encoding and inference on text, images, and structured data.
5. The method according to claim 2, characterized in that, The path cost function comprehensively considers time, distance, congestion probability, user preference factors, and environmental uncertainty indicators.
6. The method according to claim 2, characterized in that, The optimization algorithm includes graph search algorithms or reinforcement learning strategies.
7. The method according to claim 2, characterized in that, Real-time updates and interactive adjustments, outputting updated results.