People Flow Prediction Model Using Adaptive Data Assimilation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing prediction systems struggle to accurately follow the actual behavior of crowds over a desired lead time for intervention, especially when the behavior tendency changes after data assimilation.
Innovation Solution
A prediction system that estimates a behavior model reproducing observation data through data assimilation and uses this model to simulate and predict people flow, allowing the system to adapt and follow actual crowd behavior.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If data assimilation is applied to improve prediction accuracy by incorporating observation data into simulation, then prediction accuracy is improved, but the system fails to follow actual behavior up to the desired lead time for intervention
Solution Approach 1:
The patent applies dynamics by making the behavior model adaptive and changeable over time. The system updates the behavior model parameters based on observed crowd behavior, allowing the model to dynamically adjust to changing crowd tendencies. This enables the system to maintain high prediction accuracy while extending the useful lead time for intervention, as the model can follow actual behavior patterns even when they change during the prediction period.
Solution Approach 2:
The patent implements feedback mechanisms where observation data from actual crowd behavior is continuously incorporated into the simulation model. The system compares simulated behavior with observed behavior and uses this feedback to update and refine the behavior model parameters. This feedback loop enables the system to maintain accuracy over extended lead times by continuously adapting to actual crowd behavior patterns.
2Ease of operation
If agent-based simulation is used to model individual pedestrian behaviors, then detailed behavior modeling is achieved, but it remains a What-if scenario analysis that cannot predict actual situations
Solution Approach 1:
The patent transforms agent-based simulation from speculative What-if analysis to predictive modeling by incorporating feedback from real observation data. The system continuously compares simulation results with actual observed crowd behavior and uses this feedback to calibrate and update the behavior model parameters. This feedback mechanism ensures that the detailed behavior modeling capability produces reliable predictions of actual crowd situations rather than merely hypothetical scenarios.
Solution Approach 2:
The patent applies parameter changes by dynamically adjusting the behavior model parameters based on observed crowd behavior. Instead of using fixed parameters, the system updates parameters such as crowd density thresholds, movement speeds, and interaction coefficients to reflect actual observed patterns. This enables the agent-based simulation to transition from theoretical modeling to reliable prediction of real crowd behavior.
Data Source
AI summary
A non-transitory computer-readable recording medium storing a prediction program causing a computer to execute a process includes measuring a feature amount related to a plurality of options and a people flow to be reproduced based on the plurality of options, executing a simulation of the people flow by using the measured feature amount; specifying a model that determines a behavior of the people flow in accordance with an input of the measured feature amount, based on the executed simulation of the people flow and the measured people flow, and predicting the people flow based on the specified model.