A-Eye: Driving with the Eyes of AI for Corner Case Generation

Abstract

In this study, we propose a novel approach to enrich the training data for automated driving by using a self-designed driving simulator and two human drivers to generate safety-critical corner cases in a short period of time, as already presented in [12]. Our results show that incorporating these corner cases during training improves the recognition of corner cases during testing, even though, they were recorded due to visual impairment. Using the corner case triggering pipeline developed in the previous work, we investigate the effectiveness of using expert models to overcome the domain gap due to different weather conditions and times of day, compared to a universal model from a development perspective. Our study reveals that expert models can provide significant benefits in terms of performance and efficiency, and can reduce the time and effort required for model training. Our results contribute to the progress of automated driving, providing a pathway for safer and more reliable autonomous vehicles on the road in the future.

Space, Time, and Interaction: A Taxonomy of Corner Cases in Trajectory Datasets for Automated Driving

“…In addition, it offers the possibility of creating safety-critical situations in the virtual world so as not to endanger road users. In [46], a human-in-theloop approach for generating safety-critical synthetic corner cases is presented. For this purpose, the authors have set up a test rig with two control units so that the same vehicle can be controlled by two human drivers in CARLA.…”

“…Interventions by the safety driver indicate poor situational awareness by the segmentation network and provide a safetycritical corner case. During the experiments in [46], several corner cases were recorded from which the trajectory data could subsequently be obtained. Two of them are now classified based on the categories defined in Sec.…”