Generative Model Trajectory Estimation for Pedestrian Movement
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional trajectory estimation models require architecture design optimization for each estimation method, leading to inefficiencies in learning performance and accuracy when estimating pedestrian movement trajectories.
Innovation Solution
An integrated trajectory estimation model using a generative model that projects pedestrian movement trajectories onto a singular space, integrating different types of training data and calculating movement patterns to accurately predict future trajectories.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If conventional trajectory estimation models use separate architecture designs for each estimation method, then each method can be optimized independently, but the overall system complexity increases and learning performance deteriorates
Solution Approach 1:
The patent merges multiple trajectory estimation methods (stochastic prediction, deterministic prediction, momentary observation, domain adaptation, and few-shot learning) into a single integrated model architecture. This unified architecture processes different estimation methods through common components, reducing overall system complexity while maintaining the ability to perform all estimation types.
Solution Approach 2:
The integrated model employs universal components that can handle multiple estimation methods simultaneously. The model uses shared layers and parameters that serve multiple functions across different estimation approaches, eliminating the need for separate architecture designs for each method while improving learning performance through consistent training.
2Measurement precision
If conventional models require optimized architecture design for each estimation method, then learning performance can be improved for that specific method, but the preprocessing complexity and training time increase
Solution Approach 1:
The patent performs preliminary data integration and normalization in a unified preprocessing stage that prepares data for all estimation methods simultaneously. By establishing a common data representation and processing pipeline in advance, the model eliminates redundant preprocessing steps that would otherwise be required for each separate method, reducing overall preprocessing time while maintaining estimation accuracy.
Data Source
AI summary
An integrated trajectory estimation model learning method is provided. The method includes receiving a plurality of position coordinates according to a movement of a pedestrian, generating a movement trajectory set related to a movement trajectory of the pedestrian based on the plurality of position coordinates, defining a singular space having a singular space coordinate system based on the movement trajectory set, and calculating a movement pattern of the pedestrian corresponding to the movement trajectory set on the singular space, and training the trajectory estimation model by using the movement pattern to estimate a future movement trajectory from a past movement trajectory of an arbitrary pedestrian.