Learning in-hand manipulation for a compliant underactuated gripper with interactive human supervision

2023 Fiscal Year Final Research Report

• Project/Area Number: 22K14221
• Research Category: Grant-in-Aid for Early-Career Scientists
• Allocation Type: Multi-year Fund
• Review Section: Basic Section 20020:Robotics and intelligent system-related
• Research Institution: Nagoya University
• Principal Investigator: Colan Jacinto 名古屋大学, 工学研究科, 研究員 (40936746)
• Project Period (FY): 2022-04-01 – 2024-03-31
• Keywords: compliant robotic hand / in-hand manipulation / reinforcement learning / sim-to-real

Outline of Final Research Achievements

Our project has focused on developing robotic grippers and hands for learning manipulation tasks. We proposed several innovative designs, including a cable-driven gripper with decoupled joints for effective grasping force estimation and a variation of anthropomorphic robotic fingers using linkage mechanisms to increase rigidity and pinch force. Additionally, we developed a three-finger underactuated robotic hand with compliant joints and cable-driven actuation, where each finger is actuated by a single motor managing both flexion and abduction movements. Significant progress has been made in enhancing the control mechanisms and interaction capabilities of this new hand gripper. We created a simulation environment to train in-hand manipulation of small objects, implementing and validating several reinforcement learning algorithms. Furthermore, we implemented Image-to-Image Contrastive Unpaired Translation (CUT) for realistic policy training inputs to facilitate sim-to-real transfer.

Free Research Field

ロボティクス、知能機械システム

Academic Significance and Societal Importance of the Research Achievements

Our research achievements enable safe and precise manipulation with robotic grippers and hands, promoting human-robot collaboration in several fields that require delicate manipulation such as assembly operations, wearable and field robotics, assistive devices, and prosthesis.