Dexterous grip control with a four-finger planar manipulator
This书 explains how to control multiple fingers to manipulate a grasped object with precision, using a plan that keeps forces balanced and slips avoided.
This work introduces a closed-loop approach where the object is treated as a virtual finger, allowing finger positions and forces to be computed from a task’s target object motion and interaction forces. It describes phases for planning targets, computing finger commands from a mechanical model, and real-time adjustments to keep the task on track. The study centers on a simple yet versatile planar system called the NYU Four Finger Manipulator, with four independently controlled fingers, force sensors, and position encoders to support force/position control in experiments.
- How a grip transformation maps finger forces to object forces and torques, including strategies to ensure slipping is avoided.
- Methods for choosing a point of concurrence for finger forces to minimize friction dependence and maximize stability.
- Practical insights from implementing the approach on a planar four-finger setup, including task scenarios like moving a held object and following a wall with a tool.
- Discussion of sensor and actuator requirements, as well as the balance between theoretical models and real-time control.
Ideal for readers of robotics and manipulation research who want a concrete, experiment-backed view of multi-finger force/position control and its application to planar hands.