Virtual Sensor Model for Autonomous Driving Fault Injection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for acquiring sensor data for autonomous driving systems require real-world testing, which is resource-intensive, inefficient, and poses safety hazards, especially in adverse weather conditions, and does not effectively simulate various scenes.
Innovation Solution
A method and system for generating sensor data that includes receiving fault type information, generating instructions with fault parameters such as injection mode and occurrence probability, and simulating target sensor data to recreate real-world scenarios in a controlled environment, using a computing device and simulation device to create a virtual sensor model that mimics real sensors, allowing for efficient and safe data generation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If real-world testing is used to acquire sensor data, then the authenticity of sensor data is improved, but the time consumption and resource usage increase significantly
Solution Approach 1:
The patent creates a virtual sensor model that copies the characteristics and behavior of real sensors. This virtual model generates sensor data that mimics real sensor outputs, including fault characteristics, without requiring physical presence in real-world scenarios. The virtual sensor model is configured to reproduce authentic sensor data patterns while eliminating the need for time-consuming real-world testing.
Solution Approach 2:
The system performs preliminary configuration of the virtual sensor model with fault type information and fault occurrence probabilities before actual data generation. By pre-configuring the simulation environment with expected fault scenarios and their likelihoods, the system can efficiently generate authentic sensor data without needing to conduct extensive real-world testing to encounter various fault conditions.
2Adaptability or versatility
If real-world testing is conducted in adverse weather conditions, then the comprehensiveness of test scenes is improved, but safety hazards increase
Solution Approach 1:
The virtual sensor model reproduces sensor data characteristics from various weather conditions and fault scenarios without requiring the testing vehicle to physically experience these conditions. By copying the effects of adverse weather and sensor faults in a virtual environment, the system achieves comprehensive scene testing while eliminating safety hazards associated with real-world adverse conditions.
3Manufacturing precision
If more real-world testing is performed to cover various scenes, then the quality of sensor data is improved, but the cost and resource consumption increase
Solution Approach 1:
The system uses a virtual sensor model to copy and generate sensor data that reflects various fault conditions and scenes. This approach produces high-quality test data with diverse fault scenarios without requiring proportional increases in real-world testing resources. The virtual model efficiently generates authentic sensor data patterns including faults, eliminating the need for expensive and resource-intensive physical testing campaigns.
4Productivity
If fault injection is implemented in simulation, then the efficiency of data generation is improved, but the complexity of the system increases
Solution Approach 1:
The virtual sensor model copies the essential fault injection mechanisms needed to generate diverse sensor data efficiently. By implementing a simplified virtual representation of fault injection capabilities rather than complex physical fault induction systems, the patent achieves high data generation efficiency while keeping system complexity manageable through software-based simulation.
Data Source
AI summary
A method includes: receiving information about a fault type of a sensor; generating an instruction corresponding to the fault type of the target sensor, the instruction including a fault parameter, and the fault parameter including an injection mode of the fault type and a fault occurrence probability; and obtaining target sensor data having the fault type based on the injection mode of the fault type and the fault occurrence probability. In some embodiments, an injection mechanism of the fault types of the target sensors is increased, and the target sensor data corresponding to different scenes are rendered on the basis of the fault types, thus reducing occupation of real test resources.