Radar Data Simulation Using Real-World Point Cloud Merging
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Generating synthetic simulations for autonomous vehicles that accurately reflect real-world scenarios is challenging due to noisy, inconsistent, and incomplete data, which can lead to inaccurate testing of vehicle controllers when using synthetically generated radar data.
Innovation Solution
Utilizing a simulation radar data store that includes real-world object and background radar data captured from physical environments, and applying probabilistic merging techniques to generate simulation radar data for synthetic simulations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If synthetic radar data is generated from real-world sensor data, then the simulation can be created, but the data becomes noisy, inconsistent, and incomplete leading to inaccurate testing
Solution Approach 1:
The patent captures real-world radar sensor data from physical environments and stores it in a simulation radar data store. This real captured data is then used to generate simulation radar data, creating an accurate copy of real-world scenarios for testing autonomous vehicle perception models without the noise and inconsistencies of purely synthetic generation
Solution Approach 2:
The system performs preliminary capture and storage of real-world radar data in a simulation radar data store before simulation needs arise. This pre-captured data is then retrieved and used to generate accurate simulation radar data, ensuring high fidelity testing data is available when needed
2Manufacturing precision
If real-world sensor data is captured and stored, then accurate simulation data can be generated, but the data store requires significant storage space and processing resources
Solution Approach 1:
The patent extracts only the essential radar sensor data from real-world sensor captures and stores it in a simulation radar data store. By taking out only the relevant radar information needed for simulation testing rather than storing complete sensor datasets, the system reduces storage requirements while maintaining simulation accuracy
Solution Approach 2:
The system segments the captured sensor data to separate and store only the radar-specific simulation data needed for autonomous vehicle testing. This segmentation allows the data store to contain only the necessary information rather than complete sensor suites, reducing overall storage requirements
3Adaptability or versatility
If multiple different systems and components interact in real-time simulated environment, then comprehensive testing is achieved, but the simulation becomes resource and computationally expensive
Solution Approach 1:
The patent uses captured real-world radar data that inherently contains the complex interactions of multiple systems and components from actual autonomous vehicle operation. This self-service approach leverages the data's natural complexity rather than requiring computationally expensive recreation of these interactions in simulation, achieving comprehensive testing with reduced resource requirements
Data Source
AI summary
A simulation system may generate radar data for synthetic simulations of autonomous vehicles, by using a data store of object radar data and attributes determined from sensor data captured in real-world physical environments. The radar data store may include radar point clouds representing real-world objects and associated object attributes, as well as radar background data captured for a number of physical environments. The simulation system may construct radar data for use in a simulation based on radar object data and/or radar background data, including using various probabilities within various overlay regions to determine subsets of object and background radar points to be rendered. During a simulation, the generated radar data may be provided to a simulated radar sensor of a simulated vehicle configured to execute trained perception models based on radar data input.