Road Safety Control Signals for Intersection Accident-Risk Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing systems struggle to accurately predict and prevent accidents at intersections and crosswalks due to difficulties in estimating vehicle trajectories, pedestrian intentions, and the severity of dangerous situations, particularly influenced by traffic lights and spatial characteristics.
Innovation Solution
A method and apparatus that utilize a prediction model trained with a loss function to infer dangerous situation prediction information from road state information, generating a safety control signal by inputting dynamic information from sensors like cameras and LiDAR, and updating the model based on measurement information to improve prediction accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional trajectory estimation methods (Kalman filter) are used, then the system is simple to implement, but the accuracy of predicting dangerous situations is insufficient
Solution Approach 1:
The prediction model is trained in advance using historical road state information and dangerous situation data. This preliminary training phase allows the model to learn complex patterns and relationships before actual deployment, enabling accurate predictions without requiring complex real-time calculations during operation.
Solution Approach 2:
The system uses a loss function that compares predicted dangerous situation information with actual measurement information to generate feedback signals. This feedback is used to continuously update and refine the prediction model, improving its accuracy over time while maintaining a relatively simple operational structure.
2Reliability
If individual model improvement is implemented for different environments, then the prediction accuracy for specific locations improves, but the system complexity and training requirements increase
Solution Approach 1:
The system implements environment-specific prediction models trained on local road state information and dangerous situation data. Each location (intersection, crosswalk, etc.) has its own optimized model that accounts for local characteristics such as traffic patterns, road geometry, and pedestrian behavior, thereby improving prediction reliability for each specific environment.
Solution Approach 2:
Individual models for different environments are trained in advance using location-specific historical data. This preliminary training allows each model to adapt to its specific environment before deployment, achieving high local reliability without requiring complex real-time adaptation mechanisms during operation.
3Measurement precision
If real-time video processing is performed, then the system can detect current road conditions, but the processing time and computational load increase
Solution Approach 1:
The prediction model is pre-trained with extensive road state information and dangerous situation patterns during an offline phase. This preliminary action transfers complex processing requirements from real-time operation to the training phase, enabling the system to make accurate predictions quickly during actual deployment without heavy computational burden.
Solution Approach 2:
Instead of processing raw video data in real-time, the system uses the trained prediction model to generate predicted dangerous situation information based on input road state data. This copying approach replaces complex real-time video analysis with efficient model inference, reducing processing time while maintaining detection accuracy.
Data Source
AI summary
Provided are a method and apparatus for generating a safety control signal of a road. The method includes inputting road state information for a first time point, including a safety control signal for the first time point and dynamic information for the first time point obtained from a video of a road, to a prediction model, inferring dangerous situation prediction information for a second time point after the first time point, by using the prediction model, and generating a safety control signal notifying a risk of accident on the road for the second time point, based on the inferred dangerous situation prediction information, wherein the prediction model is trained by using a loss function configured by dangerous situation prediction information inferred for a specific time point from road state information before the specific time point, and dangerous situation measurement information calculated from road state information for the specific time point.