AV Simulation Scenario Generation for Rare Event Coverage
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for generating simulation driving scenarios for autonomous vehicles are limited by the lack of sufficient real-world data for rare events, leading to inadequate testing and validation, and manual crafting of scenarios is time-consuming and unrealistic.
Innovation Solution
A method is introduced to generate simulation driving scenarios by identifying vehicles in on-road driving data and swapping the perspective of the autonomous vehicle with the vehicle in the simulation, allowing it to navigate along the vehicle's trajectory, thereby expanding the range of test scenarios and improving realism.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If manual crafting of simulation scenarios is used, then scenario diversity can be improved, but time consumption and cost increase significantly
Solution Approach 1:
The patent copies real-world driving scenarios captured by autonomous vehicles and reproduces them in simulation environments. By capturing actual sensor data, vehicle states, and environmental conditions from real drives, the system creates authentic simulation scenarios without manual crafting, thereby diversifying scenario types while reducing time and cost.
Solution Approach 2:
The system enables autonomous vehicles to self-generate training data by capturing their own driving experiences. The vehicle's sensors and onboard systems automatically record and structure scenario data during normal operation, which is then fed back into the simulation environment for continued training, eliminating the need for external manual scenario creation.
2Reliability
If real-world driving data is used for simulation, then realism of scenarios is improved, but coverage of rare events remains insufficient
Solution Approach 1:
The patent inverts the traditional approach by using rare events observed in real-world data to generate additional simulation scenarios. Instead of only simulating common scenarios, the system identifies and amplifies rare but critical events from captured driving data, creating specialized simulation cases that focus on edge cases and safety-critical situations.
Solution Approach 2:
The system performs preliminary analysis of real-world driving data to identify rare events and critical scenarios before generating simulation cases. By pre-processing and categorizing captured data to detect uncommon patterns, the system proactively creates targeted simulation scenarios for rare events, ensuring comprehensive coverage before training begins.
3Device complexity
If sensors are mounted at fixed locations on autonomous vehicles, then system complexity is reduced, but ability to capture diverse perspectives is limited
Solution Approach 1:
The patent adds a temporal dimension to fixed sensor data by capturing scenarios over time and reconstructing multi-perspective views from a single fixed sensor suite. By processing sequential sensor readings and tracking object movements through time, the system derives diverse perspective information without adding physical sensors at multiple locations.
Data Source
AI summary
The disclosed technology provides solutions for improving simulation scenario generation and in particular, provides methods for improving the generation of simulation scenarios based on autonomous vehicle (AV) driving data. A method of the disclosed technology can include steps for receiving driving data, which includes sensor data from an AV that is descriptive of an environment around the AV, identifying a vehicle in the environment based on the driving data, generating a synthetic driving scenario emulating the environment based on the driving data, and identifying a trajectory of the vehicle in the synthetic driving scenario. The method can further include steps for replacing the vehicle with the AV in the synthetic driving scenario by simulating navigation of the AV in the synthetic driving scenario along the identified trajectory. Systems and machine-readable media are also provided.