Dataset: Metrologically grounded evaluation of a zero-shot 6D pose estimation pipeline for adaptive robotic manipulation via motion capture technology

Content: This project contains the research data from the paper "Metrologically grounded evaluation of a zero-shot 6D object pose estimation pipeline for adaptive robotic manipulation via motion capture technology". The paper is yet to be published. Abstract: Realizing lot-size-one production requires robotic perception capable of rapid adaptation to novel components. To assess industrial readiness, perception pipelines must be evaluated under operating conditions. This paper presents a systematic evaluation of a zero-shot, training-free 6D object pose estimation pipeline leveraging CNOS and FoundationPose. The performance is benchmarked under operating conditions against an OptiTrack Motion Capture system and evaluated through a Taguchi-based experimental design. Given an Average Recall of 0.904, the results show strong performance of the evaluated perception pipeline and demonstrate the potential of motion capture technology as a metrologically grounded reference for object pose estimation benchmarking.