Implicit Neural Representation for Stochastic Traffic Trajectory Generation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for modeling human mobility in uncertain traffic conditions either require extensive trajectory data, leading to privacy concerns, or assume deterministic conditions, failing to capture individual preferences and stochastic traffic properties.
Innovation Solution
The use of an implicit neural representation (INR) model to learn continuous, latent fields of stochastic traffic properties over space and time, allowing for the generation of trajectories that reflect non-deterministic and diverse route choices in the road network.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If data-driven generative deep learning methods (GANs, VAEs) are used to synthesize trajectories, then realistic mobility patterns can be generated, but large repositories of real-world trajectory data are required causing privacy concerns
Solution Approach 1:
The patent uses activity-based models to create synthetic agents that copy realistic mobility behaviors without using actual personal trajectory data. The synthetic population reproduces aggregate mobility patterns and individual route choice behaviors observed in real data, while maintaining complete anonymity. This allows the system to generate realistic trajectories for training deep learning models without exposing sensitive personal location information.
Solution Approach 2:
The patent introduces an intermediate synthetic population layer between real-world data and the deep learning model. The activity-based model generates synthetic agents with realistic mobility patterns that serve as a mediator, allowing the system to capture individual preferences and stochastic behaviors without directly processing or storing sensitive personal trajectory data.
2Object-affected harmful factors
If activity-based models with synthetic agents are used, then privacy concerns are reduced and less fine-grained location data is needed, but location-specific patterns and individual preferences are not captured
Solution Approach 1:
The patent enhances the activity-based model by incorporating location-specific parameters such as point-of-interest attributes, land-use characteristics, and spatial constraints. These parameter changes allow synthetic agents to exhibit location-aware behaviors and individual preferences while maintaining privacy. The model calibrates these parameters against aggregate observational data to reproduce realistic mobility patterns without requiring fine-grained personal trajectory information.
3Quantity of substance
If traditional activity-based models are used, then extensive trajectory data is not required, but stochastic traffic properties and individual preferences are not captured
Solution Approach 1:
The patent transforms the traditional static activity-based model into a dynamic system where synthetic agents exhibit stochastic behaviors and individual preferences. The model incorporates random variations in route choice, travel time, and mobility patterns that reflect real-world uncertainty. This dynamic approach allows the system to capture stochastic traffic properties without requiring extensive trajectory data for calibration.
Data Source
AI summary
Generating trajectories from an implicit neural representation (INR) model to predict human mobility in uncertain traffic conditions includes receiving geocoordinate data representing vehicle motion observations of a traffic pattern; receiving a road network based on the geocoordinate data; training the INR model to learn continuous, latent fields of stochastic traffic properties over space and time based on the geocoordinate data; utilizing the INR model to extract spatio-temporal speed distributions from the geocoordinate data; applying a near-shortest-path, heuristic algorithm, weighted by predictions of the INR model, to produce real-world routing choices for traversing the road network; generating trajectories for transportation between an origin and destination in the road network using the algorithm and the predictions of the INR model, wherein the trajectories reflect non-deterministic and diverse route choices in the road network; and outputting generated trajectories to improve routing choices for a GPS and to provide the route choices for selection.