LinkerBot Robot Hands Put Finger Coordination in Focus

The post describes a compilation of LinkerBot hand work across several object types. A small part asks for a different contact arrangement from a larger object that sits across the palm. Food handling adds another reason to pay attention to where the fingers press.

I am most interested in the change between grips. Fingers that close around an object still have to reposition or release it without losing control of the next step. The compilation is a useful prompt to inspect that sequence in longer recordings.

LinkerBot's current product page lists 6 active and 5 passive degrees of freedom for the O6, with a maximum five-finger grip force of 70 N. It lists tactile sensing as an option. These details describe that product configuration.

The number of independently driven motions differs from the total number of moving joints. A passive joint can move through mechanical coupling or contact without having its own motor command. That distinction helps explain why joint counts alone are a poor way to compare hands.

A maximum grip-force specification also leaves several questions open. How much force reaches each contact? What happens as the fingers move? How does performance change during repeated use? The specification does not answer those questions for the posted tasks.

A clip can show a completed motion while leaving the control method outside the frame. An operator, a stored sequence or a learned controller may produce similar visible movements over a short interval. The source post does not establish the method for every segment.

The same care applies to sensors. An optional tactile configuration on a product page does not confirm that a particular recording used touch feedback. That requires a description of the hardware fitted to the hand and the signals used by its controller.

For a repeatability check, I would keep the full sequence from the initial approach to the final release. Move the object between trials, record every attempt and report drops, regrasping and operator intervention.

Those observations would help separate a hand's range of motion from its reliability on a task. The current compilation gives readers examples to examine. It provides no shared trial count or task-level success rate across the clips.