Reconfigurable ML Model for Ultrasonic Obstacle Classification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current ultrasonic sensor systems in vehicles rely on rule-based empirical classifiers that generate weak true positives and false positives, are difficult to adapt to specific vehicle variants, and have high dependence on object classes, leading to inaccurate obstacle classification and potential safety issues during parking and maneuvering.
Innovation Solution
A reconfigurable machine learning model is created using a decision tree arrangement with fixed and configurable parameter models, allowing for updates via wireless or wired communication, which includes feature values, split thresholds, and handling of invalid and missing values to improve obstacle classification accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If rule-based empirical classifiers are used for obstacle detection, then the system structure is simple, but the classification accuracy is poor with weak true positives and false positives
Solution Approach 1:
The system transitions from fixed rule-based parameters to configurable machine learning parameters that can be adjusted and optimized. The configurable parameter model allows modification of classification thresholds, feature weights, and decision boundaries to improve accuracy while maintaining system structure through parameter reconfiguration rather than structural changes.
Solution Approach 2:
The classifier evolves from static rule-based logic to a dynamic machine learning model that can adapt its parameters based on training data and operational conditions. The system dynamically adjusts classification behavior through the configurable parameter model, enabling it to learn from experience and improve performance over time.
2Adaptability or versatility
If rule-based classifiers are used, then the system is easy to implement, but it is difficult to adapt to specific vehicle variants
Solution Approach 1:
Different vehicle variants are accommodated by configuring different parameter sets within the same machine learning model framework. The configurable parameter model allows each vehicle variant to have optimized parameters trained on variant-specific data, enabling adaptation without requiring separate classification systems for each variant.
Solution Approach 2:
A single machine learning model framework serves multiple vehicle variants through parameter configuration. The system maintains universality by using the same underlying model structure while adapting to different variants through configurable parameters, reducing the need for variant-specific implementations.
3Adaptability or versatility
If rule-based classifiers with many object classes are used, then comprehensive coverage is achieved, but the system complexity increases
Solution Approach 1:
The machine learning model uses configurable parameters to manage multiple object classes efficiently. By adjusting parameters such as class weights, decision thresholds, and feature importance, the system can handle comprehensive object class coverage without linearly increasing complexity. The configurable parameter model allows optimization for multi-class classification scenarios.
4Measurement precision
If the machine learning model is updated with configurable parameter model, then the classification accuracy is improved, but the model reconfiguration complexity increases
Solution Approach 1:
The model reconfiguration process is designed to be dynamic and automated. The configurable parameter model can be updated through wireless or wired communication protocols, allowing parameter adjustments without manual reconfiguration. This dynamic update capability reduces the practical complexity of model reconfiguration while maintaining improved classification accuracy.
Data Source
AI summary
A method and system is disclosed for creating a machine learning model that is reconfigurable. A fixed parameter model is created to include fixed feature values obtained during a training process for the machine learning model. The fixed parameter model may include a fixed base classifier used by the machine learning model to classify objects detected by an ultra-sonic system within a vicinity of a vehicle. A configurable parameter model may be created to include feature values that are different from the fixed feature values, the configurable parameter model including a modified base classifier. A vehicle controller may receive and update the fixed parameter model with the configurable parameter model. The machine learning model may be updated to use the configurable parameter model to classify the objects detected by the ultra-sonic system.