Automatic data collection for object detection and grasp-position estimation with mobile robots and invisible markers

摘要

Deep convolutional networks have dominated advances in object detection and grasp-position estimation using computer vision. The data-collection process for these networks is, however, time-consuming and expensive. We propose an automatic data-collection method for object detection and grasp-position estimation using mobile robots and invisible markers. Our method offers clear advantages over manual data annotation and synthetic data generation in terms of time consumption, cost, consistency, and similarity to real-world data. We compared data generated with our method against synthetically generated data to show how it can affect the robustness of the deep learning model when inferred under real-world conditions. We also conducted a comparison between our method and manual data-collection and synthetic-data-generation methods and demonstrated how our method could be used for data collection of asymmetric objects and key-point estimation tasks.