Autonomous Vehicle Navigator Testing With Route Graph Constraints
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Testing autonomous vehicle control systems is challenging due to deviations in route generation by the navigator component, leading to mismatched behavior between the autonomous vehicle system under test (AV SUT) and the actual route driven by the capturing vehicle, which can result in corrupted test results or system crashes.
Innovation Solution
A testing utility generates navigator component inputs that cause the navigator component to produce routes matching the actual route driven by the capturing vehicle, using routing graph modification data to ensure consistency between the AV SUT's generated routes and the captured log data.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Extent of automation
If the navigator component generates routes based on its own logic during testing, then the AV SUT operates autonomously and produces realistic control commands, but the generated routes may deviate from the actual route driven by the capturing vehicle, causing mismatched behavior and corrupted test results
Solution Approach 1:
The testing utility pre-processes log data to generate routing metadata that describes the actual route driven by the capturing vehicle. This routing metadata is provided to the navigator component before testing begins, constraining the navigator to generate routes that match the captured route, thereby ensuring test reliability while maintaining autonomous operation
Solution Approach 2:
The routing metadata acts as an intermediary between the log data and the navigator component. It translates the captured route information into a format that the navigator can use to generate consistent routes, ensuring that the AV SUT's behavior matches the actual driving scenario without requiring direct human intervention during testing
2Reliability
If the navigator component is constrained to generate routes matching the captured route, then test result accuracy is improved, but the navigator's autonomous route generation capability is reduced
Solution Approach 1:
The system dynamically adjusts the navigator's operation mode based on testing requirements. During testing, the navigator operates in a constrained mode using routing metadata to ensure consistency. Outside of testing, the navigator can operate autonomously with full route generation flexibility, allowing it to adapt to different operational contexts
Data Source
AI summary
Examples described herein are directed to systems and methods for autonomous vehicle control system testing. A testing utility may receive log data describing a first location of a capturing vehicle at a first driven route time and a second location of the capturing vehicle at a second drive route time. The testing utility may, using at least the location data, determine a first routing graph modification to constrain a navigator component of an autonomous vehicle control system under test. The testing utility may provide the first routing graph modification to the navigator component to cause the navigator component to generate a testing route.


