Criticality Reference Checking for False Positive Safeguards
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Safety-critical systems, such as automated driving systems, face challenges in accurately triggering functionalities like emergency braking due to false positives and false negatives, which can lead to unsafe interventions or missed responses, making it difficult to demonstrate low false positive and false negative rates in release tests and potentially causing accidents.
Innovation Solution
A computer-implemented method that receives time series data to classify triggerings as false positives or near-false positives by computing error measures and identifying near-false positives, allowing for the deactivation of system functionalities and improving diagnostic capabilities, thereby enhancing safety and reducing the frequency of false positives.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If the system triggers functionality based on detected boundary conditions and sensor data, then the system responds to critical situations, but false positives may cause unjustified interventions
Solution Approach 1:
A reference system acts as an intermediary to validate the criticality assessment of the main system. The reference criticality is computed independently using the same sensor data, and its agreement with the main system's criticality serves as a mediator to filter false positives before triggering functionality.
Solution Approach 2:
The system implements feedback by comparing the main criticality assessment with the reference criticality assessment. This feedback mechanism allows the system to verify its own decisions and correct potential false positives by detecting discrepancies between the two independent assessments.
2Reliability
If extensive endurance tests are conducted to demonstrate low false positive rates, then system safety is validated, but test duration and complexity increase
Solution Approach 1:
The reference criticality computation is performed preliminarily alongside the main criticality assessment during normal operation. This preliminary action accumulates validation data throughout the system's operational life, replacing the need for lengthy post-deployment endurance tests to demonstrate low false positive rates.
3Adaptability or versatility
If the system uses generalizing algorithms like machine learning to understand surroundings, then the system handles diverse situations, but incorrect automated understanding may occur
Solution Approach 1:
The criticality assessment is segmented into two independent parallel processes: the main system's assessment and the reference system's assessment. This segmentation allows the versatile machine learning algorithms to be used in both processes, while their independent nature provides a mechanism to detect and correct errors through comparison.
Data Source
AI summary
A computer-implemented method for safeguarding a system against false positives. The method includes: receiving a time series of a criticality, the system including a functionality that is triggered when the criticality meets a first predetermined criterion; computing a time series of a reference, the reference being a comparison criticality for a triggering of the functionality; computing a time series of an error measure at least based on the time series of the criticality and the time series of the reference, a triggering of the functionality being classified as a false positive when a portion of the time series of the error measure meets a second predetermined criterion; and identifying at least one near-false positive, a non-triggering of the functionality being classified as a near-false positive when a portion of the time series of the error measure meets a third predetermined criterion, but not the second predetermined criterion.