Achieving precise manipulation with multi-fingered robotic hands

Determine effective approaches for achieving precise manipulation—defined as fingertip-scale, finely controlled motions and contacts that yield accurate object poses and delicate interactions—using multi-fingered robotic hands.

Background

The paper distinguishes power manipulation—where multi-fingered hands excel due to increased contact stability—from precise manipulation, which requires fingertip-scale control and accurate object poses. In practice, parallel grippers are more commonly used for precision tasks, underscoring a gap in capabilities for multi-fingered hands.

Within this context, the authors explicitly note that achieving precise manipulation with multi-fingered hands remains an open problem for the community, motivating their co-design approach that optimizes both control strategies and fingertip geometry to address this challenge.

References

How to achieve precise manipulation, especially with multi-fingered hands, is still an open problem for the community.

— From Power to Precision: Learning Fine-grained Dexterity for Multi-fingered Robotic Hands  (2511.13710 - Ye et al., 17 Nov 2025) in Section 2: Related Work, Power to Precise Manipulation

It is an open question how to extend such rapid online learning techniques to handle grasping or finger gaiting motions.

— Rapid Learning of Dexterous In-Hand Pen Writing through Real-Time Jacobian Estimation  (2609.11775 - Stewart et al., 10 Sep 2026) in Section 8, Discussion, subsection “Limitations.”

Generalization to hands with multiple fingers remains an open question.

— Uni-PNP: Unified Prehensile and Non-prehensile Policy Learning for Generalizable Object Pose Reaching  (2610.21735 - Zhu et al., 15 Oct 2026) in Section CONCLUSIONS AND LIMITATIONS