Scenario-Based Event Trigger for Autonomous Vehicles
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current event identification systems for autonomous vehicles rely on rule-based triggers, which are inadequate for capturing complex scenarios, such as lane changes and interactions with pedestrians, and result in inefficient data recording and uploading to cloud storage, especially due to the large volume of data generated.
Innovation Solution
A scenario-based event triggering system using machine-learning models to generate feature vectors encoding driving scenarios and detect unique events outside pre-programmed triggers, allowing for the upload of relevant data to a central server for improved data collection and analysis.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If rule-based triggers are used for event identification, then the system is simple to implement, but it cannot capture complex driving scenarios effectively
Solution Approach 1:
The patent replaces rule-based mechanical event identification with machine learning models that automatically learn and detect complex driving scenarios from sensor data, enabling the system to capture nuanced events like lane changes and pedestrian interactions without explicit programming
Solution Approach 2:
The system transforms raw sensor data into feature vectors that encode driving scenarios, changing the parameter representation to enable more effective detection of complex events while managing computational complexity
2Loss of information
If all driving data is uploaded to cloud storage, then complete data is available for analysis, but storage burden and transmission time increase significantly
Solution Approach 1:
The system extracts only the most relevant driving events identified by the machine learning model and uploads selectively to cloud storage, removing unnecessary data transmission while preserving critical information for analysis
Solution Approach 2:
The patent implements local event detection and filtering at the vehicle level before cloud upload, allowing the system to process data locally and upload only high-value scenarios, reducing overall transmission burden
3Measurement precision
If machine learning models are used to detect unique driving scenarios, then rare and impactful events are identified more effectively, but computational resources and processing time increase
Solution Approach 1:
The system applies machine learning models selectively to detect only unusual or critical driving scenarios rather than processing all events with full computational power, using partial action to balance detection accuracy with energy consumption
Data Source
AI summary
A method for scenario-based event triggers is described. The method includes generating, by a first machine-learning (ML) model, feature vectors encoding driving scenarios surrounding an ego vehicle. The method also includes detecting, by a second machine-learning (ML) model, a unique driving scenario outside of pre-programmed event triggers corresponding to one of the feature vectors encoding driving scenarios surrounding the ego vehicle. The method further includes triggering uploading of the unique driving scenario outside of pre-programmed event triggers to a central scenario-based event control server.