Human initiated grasp space exploration algorithm for an underactuated robot gripper using variational autoencoder

Clement Rolinat (1, 2, 3) , Mathieu Grossard (1, 3, 2) , Saifeddine Aloui (4, 2, 5) , Christelle Godin (4, 2, 5)

1 LIST (CEA) - Laboratoire d'Intégration des Systèmes et des Technologies
2 CEA - Commissariat à l'énergie atomique et aux énergies alternatives
3 Université Paris-Saclay
4 CEA-LETI - Commissariat à l'énergie atomique et aux énergies alternatives - Laboratoire d'Electronique et de Technologie de l'Information
5 UGA - Université Grenoble Alpes

Grasp planning and most specifically the grasp space exploration is still an open issue in robotics. This article presents an efficient procedure for exploring the grasp space of a multifingered adaptive gripper for generating reliable grasps given a known object pose. This procedure relies on a limited dataset of manually specified expert grasps, and use a mixed analytic and data-driven approach based on the use of a grasp quality metric and variational autoencoders. The performances of this method are assessed by generating grasps in simulation for three different objects. On this grasp planning task, this method reaches a grasp success rate of 99.91% on 7000 trials.

Mots clés: multifingered gripper; grasp space exploration; variational autoencoder; grasp quality metric

Domaines: Automatique / Robotique; Réseau de neurones [cs.NE]

Clement Rolinat, Mathieu Grossard, Saifeddine Aloui, Christelle Godin. Human initiated grasp space exploration algorithm for an underactuated robot gripper using variational autoencoder. ICRA 2021 - International Conference on Robotics and Automation, May 2021, Xi'an, China. pp.2598-2604, ⟨10.1109/ICRA48506.2021.9561765⟩. ⟨cea-03387866⟩