Anyone using simulation-only training for grasp policies with zero real data?

[ronald.clark]

Genuine beginner question - A lot of the manipulation shown in production demos still leans heavily on teleoperation, particularly for anything involving fine force control or novel objects - autonomous grasp success rates on genuinely unstructured, previously-unseen clutter are still well below what teleoperation can achieve. Tendon-driven fingers let you move the heavier actuators back into the palm or forearm, keeping the fingers themselves light and fast, but they introduce cable routing, tensioning, and long-term wear problems that direct-actuated fingers don't have.

[greta78]

+1 to this. Worth adding: Object occlusion by the robot's own hand during the final approach to a grasp is a common and annoying perception problem - the closer the hand gets to a good grasp position, the more it blocks the camera's view of exactly what it's about to grab. Underactuated hands (fewer actuators than joints, using mechanical coupling to shape the grasp) are a reasonable engineering compromise for robust power grasps on a budget, but they generally can't do fine in-hand manipulation the way a fully actuated hand can.

[nschmidt]

Counterpoint: Cable routing through a wrist joint with multiple degrees of freedom is a genuinely tricky mechanical design problem - tendons and wiring both need enough slack to avoid binding through the full range of motion without tangling or fraying over thousands of cycles.

[olga24]

@nschmidt Here's what I know on this: Compliant wrists that absorb impact during a bad approach or misjudged contact reduce mechanical stress on the whole arm, which matters a lot for long-term reliability even though it's a less visible feature than the hand itself. Grasp planning for deformable or non-rigid objects (bags, cables, cloth) remains one of the genuinely unsolved problems in manipulation - rigid-body grasp models simply don't capture how the object will behave once contact starts.

[rtorres]

Here's what I know on this: Wet, oily, or otherwise low-friction objects are still a genuine edge case for most current hands, since tactile sensing and grasp-force controllers are typically tuned and validated on dry, higher-friction test objects. Imitation learning from human demonstration video (without robot teleoperation data) is an appealing way to scale up training data cheaply, but it runs into the embodiment gap - human hand kinematics and force profiles don't map directly onto a robot hand's very different mechanism.

[joseph.robinson]

@rtorres Small correction on one detail: Underactuated hands (fewer actuators than joints, using mechanical coupling to shape the grasp) are a reasonable engineering compromise for robust power grasps on a budget, but they generally can't do fine in-hand manipulation the way a fully actuated hand can.

[scott21]

@joseph.robinson I dealt with almost this exact situation. Payload-to-hand-weight ratio varies a lot across current dexterous hands, and it's a meaningful tradeoff - more DoF and finer sensing generally means more actuators and mass in the hand itself, which eats into the arm's usable payload budget. Totally unrelated but has anyone else noticed how fast component costs are dropping this year.

[benjaminsanchez]

@scott21 Not to derail, but this reminds me of something adjacent: Figure's fourth-generation Dexterous Hand (on Figure 02/03) reportedly offers 16 degrees of freedom per hand with sensors integrated into each finger, aimed at fine force control tasks like handling small electronic components without crushing them.

[ramirez77]

@benjaminsanchez Pretty much this. One thing to add: Tendon-driven fingers let you move the heavier actuators back into the palm or forearm, keeping the fingers themselves light and fast, but they introduce cable routing, tensioning, and long-term wear problems that direct-actuated fingers don't have. Grasp planning for deformable or non-rigid objects (bags, cables, cloth) remains one of the genuinely unsolved problems in manipulation - rigid-body grasp models simply don't capture how the object will behave once contact starts.