|
With the rapid development of autonomous driving technology and the acceleration of automation across industries due to the pandemic, promoting vehicle automation has become a key development goal. In recent years, as autonomous vehicles have become increasingly common on domestic and international roads for testing and commercial operation, there has been a growing risk of collisions with vulnerable road users such as pedestrians and cyclists. Previous research has primarily focused on planning test routes or scenarios, with less emphasis on the interactions and reactions between autonomous vehicles and vulnerable road users. Additionally, existing autonomous vehicle testing scenarios often use pass/fail outcomes, making it difficult for the public to understand the capability levels of current autonomous systems. Therefore, this study consolidates testing scenarios from international standards such as ISO 22737 and UN R157 and incorporates common accident and violation patterns in Taiwan to develop autonomous vehicle testing scenarios. By integrating these scenarios with existing testing environments in Taiwan, and referencing autonomous vehicle scenario development methods, we aim to create testing scenarios and vehicle actions that involve interactions between autonomous vehicles and vulnerable road users. Subsequently, this study proposes three evaluation dimensions—"action difficulty," "vehicle progression," and "safety level"—to assess the performance levels of autonomous vehicle behavior. By inviting experts and scholars in vehicle engineering, traffic engineering, and traffic safety to participate in a survey, we aim to establish a performance-level table for autonomous vehicle behavior in corresponding test scenarios. This table will serve as a reference for evaluating the interaction performance between autonomous vehicles and vulnerable road users during future testing.
|