Gross, Michaël (Arts et Metiers Institute of Technologies, LISPEN), OLABI, Adel (Arts et Metiers Institute of Technologies, LISPEN), Bearee, Richard (Arts et Metiers Institute of Technologies, LISPEN), Touchard, Mathieu (Alten)

Building Dataset for Parallel-Jaws Grasping Problems with Practicals Issues from Simulation to Reality

Scheduled for presentation during the Regular Session "Robot Hand Control" (WeAT5), Wednesday, July 16, 2025, 10:20−10:40, Room 109

Joint 10th IFAC Symposium on Mechatronic Systems and 14th Symposium on Robotics, July 15-18, 2025, Paris, France

This information is tentative and subject to change. Compiled on July 16, 2025

Abstract

This article presents a practical building of a synthetic dataset for parallel-jaws grasping problem. The creation of this dataset uses two-step pipeline, considering model-based sampling of bests grasping poses for over 2000 geometrically diverse object meshes from EGAD dataset, classified by grasping difficulty level, and experiments on real robot for performances assessment. In addition, authors propose a fair comparison of two physically based robotic simulators, CoppeliaSim and Isaac Sim, and provide some guidelines to reduce simulation to reality gap. The resulting dataset is made freely available at https://github.com/MichaelGs1/Grasping-Dataset.