We address this issue with SDC-Scissor, a framework that leverages Machine Learning to identify simulation-based tests that are unlikely to detect faults in the SDC software under test and skip them before their execution. Therefore, those tests can be characterized as “uninformative”, and running them generally means wasting precious computational resources. Shown that most of the automatically generated tests do not strongly contribute to establishing confidence in the quality and reliability of the SDC. However, previous results on testing SDCs using simulations have Simulation platforms facilitate the continuous development of complex systems such as self-driving cars (SDCs). Extensive experiments on KITTI, Waymo, our CrashD and SUN RGB-D show the high generalizability of our techniques to out-of-domain data, different models and sensors, namely LiDAR and ToF cameras, for both indoor and outdoor scenes. Towards this end, we propose and share open source CrashD: a synthetic dataset of realistic damaged and rare cars, with a variety of crash scenarios.
By augmenting normal samples with the deformations produced by these vector fields during training, we significantly improve robustness against differently shaped objects, such as damaged/deformed cars, even while training only on KITTI. The obtained vectors are transferrable, sample-independent and preserve shape smoothness and occlusions. Our approach constrains 3D points to slide along their sensor view rays while neither adding nor removing any of them. We achieve this with 3D-VField: a novel method that plausibly deforms objects via vectors learned in an adversarial fashion.
In this work, we substantially improve the generalization of 3D object detectors to out-of-domain data by taking into account deformed point clouds during training. As 3D object detection on point clouds relies on the geometrical relationships between the points, non-standard object shapes can hinder a method’s detection capability However, in safety-critical settings, robustness on out-ofdistribution and long-tail samples is fundamental to circumvent dangerous issues, such as the misdetection of damaged or rare cars.
0 kommentar(er)
