Novel Synthetic Data Tool for Data-Driven Cardboard Box Localization

TL;DR

The paper tackles the need for labeled data in 3D cardboard box localization for industrial bin-picking. It introduces a Blender-based synthetic data tool with a parametric cardboard box model and a structured-light scanning pipeline to generate labeled 3D point clouds and ground-truth transforms. A 6D pose estimator trained on data from the novel generator outperforms a baseline synthetic generator, achieving substantial reductions in translation and rotation errors on validation and real test sets, demonstrating improved generalization. The approach reduces reliance on real labeled data and offers a practical path toward robust, data-driven bin localization in factory environments, with sample synthetic data made publicly available.

Abstract

Application of neural networks in industrial settings, such as automated factories with bin-picking solutions requires costly production of large labeled data-sets. This paper presents an automatic data generation tool with a procedural model of a cardboard box. We briefly demonstrate the capabilities of the system, its various parameters and empirically prove the usefulness of the generated synthetic data by training a simple neural network. We make sample synthetic data generated by the tool publicly available.

Figures (3)

• Figure 1: Box creation process, operations are exaggerated for visual clarity.
• Figure 2: Illustration of box parameters and the resulting rendered image.
• Figure 3: Sample from the baseline generator and network predictions.