Traffic Participants in the Loop: A Mixed Reality-Based Interaction Testbed for the Verification and Validation of Autonomous Vehicles

Abstract

In order to verify and validate autonomous vehicles, testbeds integrating the whole system from perception to actuation are necessary. Above all, this applies to the assessment of the autonomous vehicle's performance in the presence of vulnerable road users' behavior: While experiments for the interaction between autonomous vehicles and pedestrians in critical traffic scenarios are hardly conductable in reality, virtual experiments often suffer from plausibility. In order to solve this issue, we present a mixed reality testbed for the verification and validation of autonomous vehicles faced with realistic road user behavior in critical, worst case traffic scenarios. We achieve this by registering an immersed pedestrian and the automated driving function within a common environment model, providing challenging traffic scenarios. The testbed is applicable within different integration levels of the automated driving function and enables a high level of behavioral realism. The testbed is evaluated qualitatively and discussed within a concrete use case.