Logged Sensor Data Augmentation for Autonomous Vehicle Simulation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Autonomous vehicle technology faces challenges in acquiring sufficient and high-quality training data to accurately represent various driving conditions, as real-world sensor data is insufficient and simulation data from video game-like simulators lacks realism.
Innovation Solution
Generating simulation data from logged sensor data of autonomous vehicles, which includes receiving sensor data, generating augmented data describing actors in the vehicle's environment, and creating simulation scenarios by varying actor characteristics and scenarios to enhance data quality and variety.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Quantity of substance
If simulation data from video game-like simulators is used, then the quantity of training data is increased, but the quality and realism of the data deteriorates
Solution Approach 1:
The patent creates realistic simulation scenarios by copying and reconstructing real-world sensor data into simulated environments. Instead of using generic video game simulations, the system captures actual sensor data from real vehicles and uses it to generate authentic simulation scenarios that preserve the statistical properties and realism of real-world driving conditions while enabling scalable data generation.
2Measurement precision
If only real-world sensor data from autonomous vehicle operation is captured, then the quality of training data is maintained, but the quantity of training data is insufficient
Solution Approach 1:
The system performs preliminary capture and storage of real-world sensor data during normal autonomous vehicle operation, preparing this data in advance for later use in generating simulation scenarios. By capturing and archiving real sensor data beforehand, the system creates a foundation that can be reused to generate multiple simulation scenarios, thereby multiplying the effective training data quantity from limited real-world captures.
3Adaptability or versatility
If simulation scenarios are generated with varied actor characteristics, then the variety and representativeness of training data is improved, but the complexity of data processing increases
Solution Approach 1:
The patent implements dynamic variation of actor characteristics in simulation scenarios by systematically modifying parameters such as actor type, size, speed, and behavior based on real-world data distributions. The system dynamically adjusts these characteristics to create diverse yet statistically representative scenarios, balancing variety with processing efficiency through algorithmic generation rather than manual creation.
Data Source
AI summary
Logged data from an autonomous vehicle is processed to generate augmented data. The augmented data describes an actor in an environment of the autonomous vehicle, the actor having an associated actor type and an actor motion behavior characteristic. The augmented data may be varied to create different sets of augmented data. The sets of augmented data can be used to create one or more simulation scenarios that in turn are used to produce machine learning models to control the operation of autonomous vehicles.