Procedural Scenario Generation for Vehicle Safety Prediction Validation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Autonomous and semi-autonomous vehicle planning systems face challenges in verifying the reliability of their predictions due to complexity, especially in intricate scenarios, as they rely on sensor data to avoid objects but struggle to assess the accuracy of these predictions effectively.
Innovation Solution
A procedural generation model that receives data on object and vehicle states, including trajectories and behaviors, to predict intersection probabilities at different locations, allowing for the validation and verification of safety system predictions by generating scenarios and testing them against potential future interactions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If planning systems utilize sensor data and perception systems to determine actions for avoiding objects, then safety and collision avoidance capability are improved, but system complexity increases making reliability inspection difficult
Solution Approach 1:
The patent creates a simplified procedural generation model that copies the essential functionality of complex safety systems to verify their predictions. Instead of inspecting the complex perception and planning systems directly, the invention generates simplified test scenarios that replicate critical safety situations, allowing reliability verification without dealing with the full system complexity.
Solution Approach 2:
The procedural generation model acts as an intermediary between the complex safety system and the verification process. It generates test scenarios and expected outcomes that mediate the reliability inspection, translating complex system behavior into verifiable predictions without requiring direct analysis of the complex perception and planning components.
2Adaptability or versatility
If planning systems explore larger parameter spaces to handle complicated scenarios, then adaptability and scenario coverage are improved, but verification and validation of predictions become more difficult
Solution Approach 1:
The patent segments the verification process into two parts: a procedural generation model that handles scenario generation and a verification component that checks predictions. This segmentation allows the system to explore large parameter spaces through procedural generation while keeping the verification task manageable by comparing against generated expected outcomes.
Solution Approach 2:
The procedural generation model performs preliminary action by generating test scenarios and expected outcomes before actual verification occurs. This pre-computation of test cases and expected results enables systematic verification of safety system predictions across diverse scenarios without making the verification process itself complex.
Data Source
AI summary
Techniques for procedurally generating scenarios that verify and validate predictions from or operation of a vehicle safety system are discussed herein. Sensors of a vehicle may detect one or more objects in the environment. A model may determine intersection values indicative of probabilities that the object will intersect with a trajectory of the vehicle. A vehicle may receive one or more intersection values from a model usable by a computing device to validate a prediction from a vehicle safety system.