ADS Policy Uncertainty Mapping for Safety Proof Data Collection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Ensuring the safety requirements of Autonomous Driving Systems (ADS) are effectively proven, particularly in the absence and presence of faults, remains a significant challenge due to the complexity of modeling and predicting their behavior, which hinders widespread market adoption.
Innovation Solution
An ADS development system that derives parameter-specific safe driving policies based on safety requirements and operational parameters, identifies uncertainties that need further observation, and generates mappings associating geographical locations with operational parameters requiring further observance to relax safety requirements, thereby optimizing performance and data collection.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If more data collection and operational hours are gathered to improve ADS performance and safety proof, then ADS performance and safety fulfillment increase, but development time and complexity increase
Solution Approach 1:
The system performs preliminary analysis of operational parameters and their uncertainties before actual data collection. By identifying which parameters need further observance and where to observe them in advance, the system prepares targeted data collection strategies that avoid unnecessary data gathering, thus reducing development time while ensuring safety requirements are met.
Solution Approach 2:
The ADS development system automatically derives safe driving policies, identifies uncertainties, selects parameters needing observance, and generates location mappings without requiring extensive manual analysis. This self-service capability accelerates the safety proof process by automating the identification of critical data collection needs.
2Reliability
If comprehensive models of operational parameters are created to prove safety requirements, then safety proof capability improves, but system complexity increases
Solution Approach 1:
The system segments the complex safety proof task into distinct components: deriving safe driving policies from safety requirements, analyzing operational parameters and their uncertainties, identifying parameters needing further observance, and generating location mappings. This segmentation simplifies the overall complexity by breaking down the monolithic safety proof process into manageable, automated steps.
Solution Approach 2:
The system introduces an intermediary analysis layer that processes operational parameters and their uncertainties to identify which parameters need further observance. This intermediary step acts as a bridge between the complex safety requirements and the actual data collection needs, simplifying the path to safety proof by filtering and prioritizing critical parameters.
3Productivity
If targeted data collection is implemented to improve specific operational parameters, then performance growth efficiency increases, but identification accuracy of critical parameters is required
Solution Approach 1:
The system establishes a feedback loop where safe driving policies are derived from safety requirements, operational parameters are analyzed with their uncertainties, and parameters needing further observance are identified based on this analysis. The generated location mappings guide targeted data collection, which in turn provides feedback to refine the models and improve policy accuracy, creating a continuous improvement cycle.
Solution Approach 2:
The system dynamically adjusts the focus of data collection by changing which operational parameters are prioritized based on their identified uncertainties. By deriving safe driving policies that are parameter-specific and understanding which parameters need further observance, the system adapts its data collection strategy to target the most critical parameters, improving identification accuracy over time.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
The present disclosure relates to a method performed by an ADS development system (1) for supporting proving fulfilment of safety requirements imposed on Automated Driving Systems, ADSs. The ADS development system derives (1001) based on the safety requirements and predeterminable models for a set of operational parameters of an ADS (21), a corresponding set of parameter-specific ADS safe driving policies, each safe driving policy exhibiting a respective uncertainty inflicted by an uncertainty of the corresponding operational parameter. The ADS development system further identifies (1002) at least a first parameter-specific ADS safe driving policy inflicted with an uncertainty fulfilling predeterminable criteria, which criteria filters out uncertainties indicating, respectively, that the corresponding operational parameter needs to be further observed in order to relax the ADS's currently allowed safety requirements-fulfilled ADS safe driving policy. Moreover, the ADS development system identifies (1003) at least a first geographical location, e.g. of an ADS-compliant digital map, exhibiting conditions allowing the operational parameter(s) in need of further observance, to be observed. The ADS development system further generates (1004) a mapping associating the identified at least first geographical location with the operational parameter(s) in need of further observance. The disclosure also relates to an ADS development system in accordance with the foregoing, an arrangement - for instance an offboard system (114) and/or a vehicle (2) comprising such an ADS development system, and a respective corresponding computer program product and non-volatile computer readable storage medium.