Autonomous Vehicle Subsystem Supervision for Fault Isolation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for evaluating the performance of autonomous vehicles do not effectively pinpoint the sub-systems responsible for malfunctions, especially in critical situations.
Innovation Solution
A method and device for supervising the operation of a motor vehicle equipped with multiple automated systems, which involves activating supervision, iteratively measuring and scoring the performance of each automated system, and updating the systems based on their scores.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If overall behavior evaluation is used to assess autonomous vehicle performance, then the evaluation covers all systems comprehensively, but the ability to pinpoint specific malfunctioning sub-systems is lost
Solution Approach 1:
The patent divides the autonomous vehicle into multiple independent automated systems (perception system, decision-making system, control system) and evaluates each separately through iterative supervision. This segmentation allows precise identification of which specific subsystem caused a malfunction while maintaining comprehensive coverage of all systems through the ordered evaluation sequence.
2Measurement precision
If iterative supervision of each automated system is implemented, then the malfunctioning sub-system can be identified precisely, but the complexity of the supervision process increases
Solution Approach 1:
The patent establishes a predetermined ordered set of automated systems to be evaluated in sequence (perception → decision-making → control). This preliminary arrangement of evaluation order simplifies the supervision process by providing a clear, structured workflow that reduces complexity while maintaining precise malfunction identification through systematic iteration through each subsystem.
3Extent of automation
If reinforcement learning algorithm is used for overall behavior evaluation, then the evaluation can be automated, but the specific sub-systems causing malfunctions cannot be pinpointed
Solution Approach 1:
The patent introduces an intermediary evaluation mechanism that sits between the automated reinforcement learning evaluation and the specific subsystem identification. The iterative supervision process acts as an intermediary that translates the overall automated evaluation into specific subsystem-level diagnostics, allowing both automation and precise identification to coexist.
Data Source
AI summary
A method supervises the operation of a motor vehicle included an ordered set of at least two automated systems. The method includes activating the supervision of the ordered set of automated systems and iterations of supervising an automated system. Each iteration is applied successively, according to the order of the ordered set, to one of the automated systems. The supervising including measuring a performance of the automated system being supervised, and optionally assigning a positive, negative, or zero score to the automated system being supervised on the basis of the measured performance. The method also includes updating at least one automated system of the ordered set on the basis of the score assigned to the at least one automated system