Validation must show, with enough confidence, that the risk remaining from hazardous scenarios not yet discovered satisfies the acceptance criteria, with a representative spread of the scenario space across all V&V activities. Methods include robustness validation, architecture and independence validation, randomised and in-the-loop tests, long-term vehicle and fleet tests, tests from field experience, corner and edge cases, comparison with existing systems, simulation over random scenario sequences, misuse tests with naive users, sensitivity and parameter analysis, real-world scenario exploration, probabilistic modelling and validation against ground truth. A rationale shows the selected methods can find hazardous scenarios, test distance and its split across track, simulation and road is justified, additional safety measures protect the public in open-road tests, and re-evaluation follows changes such as algorithm, ODD or policy changes.
The graph holds this control, the 0 it maps to, and the evidence behind each claim, over MCP and REST.