Forecasting Traffic Signal Switching Behavior Using Interpretable Machine Learning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing light signaling systems fail to accurately and transparently forecast future switching behavior, leading to inadequate modeling of hierarchical processes and insufficient interpretability, making it difficult to adapt to changing traffic conditions and identify errors in forecasting.
Innovation Solution
A computer-implemented method using machine learning models, specifically rule-based approaches like decision trees for deterministic switching behavior and stochastic approaches like Poisson processes for sensor units, to configure system and sensor models that provide reliable forecasts of switching behavior and probability distributions, enabling better decision-making in traffic management.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If black box machine learning methods (neural networks, random forests) are used to forecast remaining time or probability of green light, then the forecasting capability is improved, but interpretability and transparency are worsened
Solution Approach 1:
The patent segments the machine learning model into interpretable components by representing the hierarchical process as a directed acyclic graph (DAG) with discrete states and transitions. This segmentation allows each component of the forecasting system to be individually understood and validated, while maintaining overall forecasting accuracy.
Solution Approach 2:
The patent introduces an intermediary layer between the black box model and the final forecast output. The hierarchical state machine acts as a mediator that translates complex model predictions into interpretable sequences of states and transitions, preserving both accuracy and understandability.
2Productivity
If conventional machine learning models are used, then forecasting is performed, but the hierarchical process of state changes is insufficiently modeled
Solution Approach 1:
The patent implements a dynamic hierarchical state machine that adapts to changing traffic conditions. The system transitions between discrete states based on observed patterns, allowing the model to dynamically adjust its forecasting approach while maintaining computational efficiency and reliability.
Solution Approach 2:
The patent changes the parameter representation from continuous predictions to discrete state transitions. By modeling the hierarchical process as sequences of state changes with associated probabilities, the system achieves both efficiency in computation and reliability in capturing the true nature of traffic signal behavior.
3Reliability
If expert knowledge about switching behavior is incorporated, then the modeling accuracy should improve, but the model complexity increases
Solution Approach 1:
The patent segments expert knowledge into discrete, manageable components represented as states and transitions in a hierarchical state machine. This segmentation allows expert knowledge to be systematically incorporated without creating an intractably complex model, as each state and transition can be independently defined and validated.
Data Source
AI summary
A computer-implemented method for configuring a system model and a computer-implemented method for configuring a sensor model. There is also described a computer-implemented method for determining future switching behavior of a system unit, with the following steps: a) receiving the configured system model; b) receiving the configured sensor model, c) the configured sensor model being a probability distribution regarding how the sensor unit will behave in the specific time period; d) establishing at least one random sample of behavior of a sensor unit by sampling from the probability distribution; and e) determining the future switching behavior of the system unit and/or at least one associated statistical value on the basis of the established random sample by means of the trained system model. There is also described a corresponding computer program product.