Autonomous Driving ML Rationalization Check for Safe Real-Time Validation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional systems for verifying machine learning algorithms in autonomous vehicles are computationally intensive and complex, posing a need for a more efficient and safer method to validate these algorithms in real-time during vehicle operation.
Innovation Solution
A testing system and method that involves a sensor system capturing external environment data, a controller processing this data by inserting known input data into a target portion to generate modified data, and determining the accuracy of autonomous driving features by comparing output and known data, with the capability to enable or disable the feature based on accuracy, specifically applicable during vehicle start-up or when the feature is disengaged.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If conventional checker-doer systems are used to verify machine learning algorithm outputs, then verification accuracy is improved, but computational complexity and system design complexity increase
Solution Approach 1:
The patent extracts the verification function from a separate complex checker-doer system and integrates it into the existing autonomous driving system as a rationalization check module. This allows verification to be performed using the same neural network processor and memory structures already present in the system, eliminating the need for separate verification hardware and reducing overall system complexity while maintaining verification accuracy
Solution Approach 2:
The patent makes the existing neural network processor and memory structures serve dual purposes: they process both the original input data for autonomous driving decisions and the rationalization data for verification purposes. By using the same hardware resources for both verification and primary processing, the system avoids adding separate verification components, thereby reducing device complexity while maintaining reliability
2Reliability
If conventional checker-doer systems are used to verify machine learning algorithm outputs, then verification accuracy is improved, but computational resources and processing load increase
Solution Approach 1:
The patent merges the verification processing with the primary autonomous driving processing by using the same neural network processor and memory structures for both rationalization check and object detection/classification tasks. This consolidation allows the system to perform verification without adding separate computational resources, thereby reducing energy consumption and processing load while maintaining verification accuracy
Solution Approach 2:
The patent creates a simplified copy of the rationalization data in memory that mirrors the structure of the original input data. This copy is then processed by the existing neural network processor using the same algorithms, allowing verification to be performed with minimal additional computational resources since the processing pipeline is already in place for the primary function
3Adaptability or versatility
If machine learning algorithms are used for autonomous driving features, then functionality and adaptability are improved, but unexpected outputs and safety concerns increase
Solution Approach 1:
The patent implements a feedback mechanism where the rationalization check continuously monitors the neural network outputs and compares them against expected rationalization data stored in memory. When discrepancies are detected, the system can trigger safety protocols or alert the driver, providing continuous safety assurance that feedbacks on the machine learning algorithm outputs without limiting the algorithm's adaptability for various autonomous driving scenarios
Data Source
AI summary
Systems and methods for testing a machine learning algorithm or technique of an autonomous driving feature of a vehicle utilize a sensor system configured to capture input data representative of an environment external to the vehicle and a controller configured to receive the input data from the sensor system and perform a testing procedure for the autonomous driving feature that includes inserting known input data into a target portion of the input data to obtain modified input data, processing the modified input data according to the autonomous driving feature to obtain output data, and determining an accuracy of the autonomous driving features based on a comparison between the output data and the known input data.