Vehicle Incident Learning Pipeline for Continuous Driving Updates
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Vehicle applications and machine learning models are not updated in a timely manner to address new issues, weaknesses, or edge cases, requiring manual developer intervention for updating requirements and testing.
Innovation Solution
A system that automatically collects incident samples from the Internet, clusters them using a clustering machine-learning model, and defines vehicle application compliance requirements using a requirement-defining machine-learning model, enabling continuous update of vehicle applications and machine learning models without developer notification or manual input.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If manual developer intervention is used to update requirements and test vehicle applications, then accuracy and control of updates are improved, but update speed and productivity deteriorate
Solution Approach 1:
The system enables self-service by automatically collecting incident samples from the Internet, clustering them using machine learning models, and generating updated requirements without manual developer intervention. The automated pipeline includes incident collection, clustering, requirement generation, and testing phases, all executed autonomously to maintain continuous improvement of vehicle applications.
Solution Approach 2:
The patent replaces manual mechanical processes with automated electronic systems. Manual requirement writing and testing are substituted by machine learning models that automatically generate requirements from clustered incident data. The system uses automated pipelines to collect, process, and validate updates, eliminating the need for manual developer involvement in routine updates.
2Adaptability or versatility
If vehicle applications are updated frequently to address new issues, then adaptability and safety are improved, but system stability and reliability deteriorate
Solution Approach 1:
The system performs preliminary action by proactively collecting incident samples from the Internet before they become critical issues. The automated pipeline continuously monitors for new incidents, clusters them by similarity, and generates requirements in advance. This allows the system to prepare updates before actual problems occur, maintaining both adaptability and stability through anticipatory updates rather than reactive fixes.
Solution Approach 2:
The patent implements feedback mechanisms where the system continuously monitors vehicle application performance and incident occurrence rates. The automated pipeline uses this feedback to determine when updates are needed, ensuring that frequent updates only occur when actual incidents indicate a problem. The feedback loop includes incident collection, clustering by severity, and conditional requirement generation, preventing unnecessary updates while maintaining system reliability.
3Productivity
If automated requirement generation is implemented, then productivity and update speed are improved, but complexity of the system deteriorates
Solution Approach 1:
The patent applies segmentation by dividing the complex automated system into distinct modular components: incident collection module, clustering module, requirement generation module, and testing module. Each component handles a specific phase of the automated pipeline, making the overall complex system manageable through modular architecture. The segmentation allows independent development and maintenance of each module while achieving automated end-to-end operation.
Data Source
AI summary
Continuous update of driving system for incident avoidance is performed by collecting a plurality of incident samples from an Internet, the plurality of incident samples identified by an identification machine-learning model to involve one or more vehicles, clustering, by a clustering machine-learning model, the plurality of incident samples into a plurality of incident clusters, and defining, by a requirement defining machine-learning model, a vehicle application compliance requirement according to an incident cluster among the plurality of incident clusters.