Information Generation Device for Autonomous Driving Test Cases
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional test cases for autonomous driving technology fail to sufficiently reproduce actual road traffic conditions, particularly missing scenarios that did not lead to accidents and cannot generate test cases for unexpected events like traffic accidents or landslides based on sensor detection signals.
Innovation Solution
An information generation device that simulates road traffic conditions by storing moving-object information from roadside sensors, determining incident behavior, extracting relevant data, and generating test cases based on this information to reproduce actual environments, including weather considerations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If test cases are generated based on accident investigation results, then test cases can be created using available data, but actual road traffic environment cannot be sufficiently reproduced
Solution Approach 1:
The patent uses sensor data from the actual road environment to create digital copies of real traffic scenarios. By capturing and storing moving object information from roadside sensors, the system reproduces actual road conditions including non-accident scenarios, replacing reliance on accident investigation results with direct environmental copies.
Solution Approach 2:
The system performs preliminary data collection and storage of moving object information before test case generation is needed. By continuously capturing sensor data and storing it in advance, the system prepares a comprehensive database of real traffic conditions that can be used to generate test cases for various scenarios including those that never resulted in accidents.
2Adaptability or versatility
If test cases are generated only from accident data, then available accident records can be utilized, but cases that did not lead to accidents cannot be generated
Solution Approach 1:
The system creates digital replicas of normal traffic conditions by copying sensor data from the road environment. This allows generation of test cases representing typical driving scenarios that did not result in accidents, expanding the variety and quantity of available test cases beyond accident records.
Solution Approach 2:
The patent segments traffic scenarios into different categories by analyzing moving object information from sensors. By dividing the continuous sensor data into discrete event types and normal conditions, the system can generate separate test cases for various scenarios, increasing both the quantity and adaptability of test cases.
3Measurement precision
If sensor detection signals are used to determine unexpected events, then detection capability is provided, but test cases reproducing unexpected incidents cannot be generated
Solution Approach 1:
The system captures and stores complete sensor detection data including unexpected events as they occur in the real environment. By copying the full sensor information at the moment of detection, the system preserves detailed incident data that can be used to generate accurate test case reproductions of unexpected events.
Solution Approach 2:
The system uses sensor detection signals as feedback to identify unexpected events and triggers test case generation based on this detected information. The detection results feed back into the test case creation process, ensuring that actual detected incidents are accurately reproduced as test cases.
Data Source
AI summary
An information generation device generating a test case being a simulation model for reproducing a road traffic condition in an area on a road including a target point, the information generation device including: a first storage unit that stores moving-object information being information regarding a moving object existing in the area; a determination unit that determines whether or not an incident in which the moving object existing in the area shows a behavior that leads to occurrence of an accident has occurred, on the basis of the moving-object information; an extraction unit that extracts, as target information, moving-object information in a target period being a predetermined time period including a time point at which the incident occurred; and a generation unit that generates the test case upon occurrence of the incident on the basis of the target information.