Shadow Model Deviation Detection for Autonomous Driving Data Collection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for acquiring sample deviation data in automatic driving vehicles are inefficient, as they require manual construction of test environments and extensive data recording, making it difficult to collect targeted data quickly and accurately.
Innovation Solution
A method and apparatus that acquire sample deviation data by comparing driving behavior parameters in manual and simulated automatic driving modes, automatically collecting data when deviations occur, using a 'shadow model' to simulate human driving behavior and identify discrepancies, thereby improving data collection efficiency and accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If manual construction of test environments and extensive data recording is used to acquire sample deviation data, then data completeness can be ensured, but data collection efficiency and speed deteriorate
Solution Approach 1:
The system performs preliminary actions by pre-defining deviation conditions and thresholds before actual driving occurs. The shadow model is pre-configured with deviation detection rules, allowing the system to automatically identify and collect sample data when deviations occur, eliminating the need for manual test environment construction and extensive preliminary data recording
Solution Approach 2:
The patent creates a virtual copy of the automatic driving model called a 'shadow model' that runs parallel to the actual model. This shadow model copies the decision-making logic and outputs of the automatic driving model, allowing comparison between actual and simulated driving behaviors without requiring separate manual test environments, thus improving data collection efficiency while maintaining data completeness
2Measurement precision
If manual construction of test environments is used to acquire sample deviation data, then targeted data collection can be achieved, but manual labor and costs increase
Solution Approach 1:
The system implements self-service by enabling automatic deviation detection and data collection without manual intervention. The shadow model automatically compares its outputs with the actual automatic driving model, identifies deviations based on pre-set conditions, and triggers sample data collection autonomously, eliminating manual test environment construction and reducing labor costs while maintaining targeted data collection
Solution Approach 2:
The patent establishes a feedback mechanism where the shadow model continuously monitors and compares the actual automatic driving model's outputs. When deviations are detected through this feedback loop, the system automatically triggers data collection. This closed-loop feedback system replaces manual monitoring and test environment construction, reducing operational complexity and costs while ensuring accurate targeted data collection
3Quantity of substance
If extensive data recording is used to acquire sample deviation data, then data coverage can be ensured, but data collection time and resources increase
Solution Approach 1:
The system extracts only the necessary sample data that exhibits deviations between the shadow model and the actual automatic driving model. Instead of recording extensive data from all driving scenarios, the system selectively collects data points where deviations occur, defined by pre-set conditions and thresholds. This extraction approach ensures adequate data coverage for model improvement while significantly reducing data collection time and resource consumption
Data Source
AI summary
The present application discloses a method and an apparatus for acquiring sample deviation data and an electric device, which relate to the fields of artificial intelligence technology, automatic driving technology, intelligent transportation technology and deep learning technology. The specific implementation solution is: in case of acquiring the sample deviation data, a first driving behavior parameter of a vehicle and a second driving behavior parameter of the vehicle in a simulated automatic driving mode are respectively acquired in the manual driving mode; and it is determined whether there is a deviation between the first driving behavior parameter and the second driving behavior parameter, and if there is a deviation, the vehicle is controlled to acquire the sample deviation data within a preset time period.