Wrist-driven passive grasping: interaction-based trajectory adaption with a compliant anthropomorphic hand.

Area of Science:

• Robotics
• Biomechanics
• Control Systems

Background:

• Human musculoskeletal systems exhibit complex passive dynamics crucial for adaptive grasping.
• Exploiting these passive dynamics in robotic systems enables nuanced environmental interactions.
• Existing robotic hands often struggle with dynamic adaptation due to soft structures and high degrees of freedom.

Purpose of the Study:

• To develop an adaptive control strategy for a passive anthropomorphic robot hand.
• To enable the robot hand to exploit its passive dynamics for improved object interaction.
• To address the challenge of generating bespoke wrist trajectories for dynamic environments.

Main Methods:

• Developed a passive anthropomorphic robot hand with complex passive dynamics.
• Proposed a novel approach to adapt existing wrist trajectories based on environmental interaction analysis and classification.
• Mapped object-hand interactions back to hand control for trajectory generation parameterized by object size and task.

Main Results:

• Achieved significant improvements in grasping success rates by adapting to environmental changes.
• Demonstrated up to an 86% improvement in grasping success for object size variations up to ±50%.
• Successfully parameterized trajectory generation based on object size and task requirements.

Conclusions:

• The proposed adaptive approach effectively utilizes the passive dynamics of the robot hand.
• This method enhances robotic grasping capabilities in unstructured and changing environments.
• The system offers a promising solution for robust and adaptive robotic manipulation.