Machine Learning Scenario Clustering for Automated Driving Validation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for determining similar scenarios based on sensor data in automated and autonomous driving systems are computationally intensive and inefficient, requiring significant resources to identify relevant scenarios.
Innovation Solution
A computer-implemented method using machine learning algorithms to generate dimension-reduced feature representations of sensor data from multiple augmentations, applying optimization algorithms to approximate and cluster similar scenarios, thereby reducing computational requirements and identifying relevant interaction maneuvers.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional methods are used to determine similar scenarios based on sensor data, then comprehensive scenario coverage is achieved, but computational resources and processing time are excessively consumed
Solution Approach 1:
The system performs preliminary actions by generating multiple augmentations of the sensor data set before the actual scenario determination process. These augmentations are created in advance and stored for later use, allowing the optimization algorithm to work with pre-prepared data rather than processing raw sensor data from scratch during validation, thus reducing computational load while maintaining scenario coverage
Solution Approach 2:
The invention changes parameters by applying different augmentation transformations to the sensor data set, creating varied versions of the same underlying scenario data. This allows the system to explore multiple scenario variations without requiring equivalent computational resources, as the augmentations are generated efficiently and reused across multiple analyses
2Measurement precision
If multiple augmentations of sensor data are generated and processed through machine learning algorithms, then identification of relevant scenarios is improved, but computational complexity increases
Solution Approach 1:
The system creates multiple copies (augmentations) of the sensor data set, where each augmentation is a transformed version of the original data. These copies allow the machine learning algorithm to be trained and evaluated on varied data representations without requiring equivalent computational resources for each unique scenario, as the augmentations are generated through efficient transformations rather than full scenario simulations
Solution Approach 2:
The augmentations serve multiple functions: they are used for training the machine learning algorithm, for validating scenario determination, and for generating diverse scenario variations. This multi-functionality reduces overall computational complexity by reusing the same augmented data sets across multiple purposes rather than generating separate data sets for each function
Data Source
AI summary
A computer-implemented method for providing a machine learning algorithm for determining similar scenarios based on scenario data of a data set of sensor data, wherein an optimization algorithm is applied to the feature representation, output by the first machine learning algorithm, of the first augmentation of the data set of sensor data, wherein the optimization algorithm approximates the feature representation, output by the second machine learning algorithm, of the second augmentation of the data set of sensor data. The invention further relates to a method for determining similar scenarios based on scenario data of a data set of sensor data and to a training controller.