Autonomous Vehicle Safety Analysis via Virtual Human Driver Correlation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current systems for autonomous vehicles lack effective methods to classify and analyze safety events, making it difficult to evaluate and improve the safety performance of autonomous vehicles compared to human-operated vehicles under equivalent driving scenarios.
Innovation Solution
A system and method that correlate sensor data from human-operated vehicles with autonomous vehicles using similarity-based analysis, employing machine learning techniques and safety event processing to generate safety-related analysis and performance metrics, enabling objective evaluation of safety between human and autonomous vehicle operations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If sensor data from autonomous vehicles is collected and analyzed, then safety performance evaluation is enabled, but lack of effective classification and analysis methods makes it difficult to evaluate safety compared to human-operated vehicles
Solution Approach 1:
The patent creates a virtual human driver model that replicates human driving behavior patterns by processing sensor data through machine learning algorithms. This virtual model serves as a copy of human driving decisions, enabling comparison between autonomous vehicle performance and human-driven safety standards without requiring actual human drivers in every test scenario
Solution Approach 2:
The patent introduces a virtual human driver as an intermediary between sensor data collection and safety evaluation. This virtual driver acts as a mediator that translates raw sensor data into comparable safety metrics by simulating human decision-making processes, bridging the gap between autonomous vehicle data and human-operated vehicle safety standards
2Measurement precision
If similarity-based analysis is used to correlate human-operated vehicle events with autonomous vehicle events, then objective safety evaluation is enabled, but complex processing of sensor data and event correlation is required
Solution Approach 1:
The patent segments the complex safety evaluation process into distinct modules: sensor data acquisition, event detection, virtual human driver simulation, event correlation, and safety metric generation. Each module handles a specific aspect of the analysis, making the overall complex system more manageable and interpretable while maintaining high measurement precision through specialized processing at each stage
Solution Approach 2:
The patent transforms raw sensor data into standardized safety event parameters that can be directly compared between human-operated and autonomous vehicles. By changing the parameter representation from raw sensor readings to normalized safety metrics, the system enables precise comparison while reducing the complexity of direct raw data analysis
Data Source
AI summary
Systems and methods are provided for performing autonomous vehicle operation analysis. A method includes sensor data being received from sensor devices that is representative of an environment and operation of a human-operated vehicle. Human event data is determined based upon the received sensor data. Event data associated with operation of an autonomous vehicle is received. The human event data is correlated with the autonomous vehicle event data based upon degree of similarity of human-operated vehicle events and autonomous vehicle events with respect to driving scenarios. Safety-related analysis is generated, by the one or more data processors, based upon the correlated human-operated vehicle events and autonomous vehicle events.