Driving safely is a complex task. It involves a broad range of skills invoked in a vast number of potential scenarios. Assessing a set of behavioral competencies offers a valuable directional indication of automated driving system-dedicated vehicle (ADS-DV) safety performance. Behavioral competencies provide a starting point for additional assessment and contribute to a manufacturer’s case for safety.
This best practice provides an approach to specify testable ADS behavior by:
- Clarifying a lexicon surrounding ADS behaviors
- Enumerating an elemental set of behaviors
- Demonstrating how to derive metrics to evaluate behavioral competence
To evaluate as many dynamic driving task (DDT) subtasks as possible, ADS developers decompose the DDT into a generalized set of behaviors. Subsequently, developers use system engineering techniques to ensure that this decomposition maps to an elemental set of behavioral competencies. In other words, the ADS may not always invoke a single behavior, but a set of behaviors covering a predictable part of the driving task. Although real-world conditions involve complex interactions among numerous systems in various situations, competency across a broad set of behaviors provides evidence of, and increases confidence in, baseline safety performance. Similarly, well-established system engineering techniques inform the validation of the driving capabilities by prescribing test criteria and integrating the results with developmental testing and overall safety case evaluation.
Providing a framework for manufacturers to link behavioral competencies to key scenarios in a given ODD provides evidence for ADS safety performance that manufacturers can use to build their case for safety. It should help to make individual test cases more extensible and therefore increase confidence in ADS safety performance.
