Active End-Effector Pose Selection for Tactile Object Recognition\n through Monte Carlo Tree Search

2017/02/28 by Mabel M. Zhang, Nikolay Atanasov, Zhang, Mabel M. +3 · 1 citation
Computer Science · Engineering · #Artificial Intelligence in Games #FOS: Computer and information sciences #Human Pose and Action Recognition #Reinforcement Learning in Robotics #Robot Manipulation and Learning #Robotics (cs.RO)

paper · pdf · doi:10.48550/arxiv.1703.00095

openalex publication_date 2017/02/28 · openalex created_date 2022/10/04 · openalex updated_date 2026/07/28

Abstract

This paper considers the problem of active object recognition using touch\nonly. The focus is on adaptively selecting a sequence of wrist poses that\nachieves accurate recognition by enclosure grasps. It seeks to minimize the\nnumber of touches and maximize recognition confidence. The actions are\nformulated as wrist poses relative to each other, making the algorithm\nindependent of absolute workspace coordinates. The optimal sequence is\napproximated by Monte Carlo tree search. We demonstrate results in a physics\nengine and on a real robot. In the physics engine, most object instances were\nrecognized in at most 16 grasps. On a real robot, our method recognized objects\nin 2--9 grasps and outperformed a greedy baseline.\n

Citations

• Hand movements: A window into haptic object recognition

Cited by

• Active Tactile Exploration for Rigid Body Pose and Shape Estimation

Related

• Object Pose Estimation through Dexterous Touch
• E-TRoll: Tactile Sensing and Classification via A Simple Robotic Gripper for Extended Rolling Manipulations
• Pose-free object classification from surface contact features in sequences of Robotic grasps
• Active Visuo-Tactile Point Cloud Registration for Accurate Pose Estimation of Objects in an Unknown Workspace
• PoseFusion: Robust Object-in-Hand Pose Estimation with SelectLSTM