Method for driving behavior modeling based on spatio-temporal information fusion
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing driving behavior models in virtual simulation environments for autonomous driving lack sufficient realism and intelligence, making it difficult to accurately replicate human driving behavior and improve trajectory control accuracy.
Innovation Solution
A method for driving behavior modeling based on spatio-temporal information fusion, which includes obtaining a training sample set, constructing a driving behavior model with spatial and temporal information encoding networks, a feature fusion network, and a feature decoding network, and training the model to determine a future trajectory sequence of a target vehicle.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional driving behavior models are used in virtual simulation environments, then the model structure is simple, but the realism and intelligence of driving behavior are insufficient
Solution Approach 1:
The patent segments the driving behavior model into multiple specialized components: spatial information encoding network, temporal information encoding network, feature fusion network, and feature decoding network. Each component processes specific aspects of driving behavior (spatial relationships, temporal sequences, feature integration), allowing the system to achieve high realism through modular specialized processing while maintaining manageable complexity through clear functional separation.
Solution Approach 2:
The patent introduces multi-dimensional information processing by incorporating both spatial dimensions (lane markings, vehicle positions, trajectories) and temporal dimensions (historical trajectory sequences, time-stamped observations). This dimensional expansion enables the model to capture complex driving patterns and human-like behavior decisions that single-dimension models cannot achieve.
2Loss of information
If spatial and temporal information are not fused, then the model structure is simple, but the understanding of driving scenarios is insufficient
Solution Approach 1:
The patent merges spatial information encoding and temporal information encoding through a feature fusion network that integrates both dimensions. The spatial encoding network processes lane markings, vehicle positions, and trajectories, while the temporal encoding network processes historical sequence data. The fusion network combines these encoded features to create a comprehensive scenario representation, ensuring no information loss from either dimension while managing complexity through structured integration.
3Measurement precision
If simple trajectory prediction methods are used, then the computational process is fast, but the trajectory control accuracy is low
Solution Approach 1:
The patent performs preliminary encoding of spatial and temporal information separately before final trajectory prediction. The spatial information encoding network pre-processes lane markings, vehicle positions, and trajectories into compact representations. The temporal information encoding network pre-processes historical trajectory sequences. This preliminary encoding reduces the complexity of the final prediction step, achieving high accuracy through structured pre-processing while maintaining computational efficiency.
Data Source
AI summary
Provided is a method for driving behavior modeling based on spatio-temporal information fusion, relating to the field of driving behavior simulations. The method includes: constructing a driving behavior model, where the driving behavior model includes a spatial information encoding network, a temporal information encoding network, a feature fusion network, and a feature decoding network, with the feature fusion network being connected to both the spatial information encoding network and the temporal information encoding network, and the feature decoding network being connected to the feature fusion network; determining a future trajectory sequence of a target main vehicle at future time points based on the trained driving behavior model according to spatial information and temporal information of the target main vehicle, where the target main vehicle is controlled to travel according to the future trajectory sequence at the future time points.