Underactuation Design for Tendon-driven Hands via Optimization of Mechanically Realizable Manifolds in Posture and Torque Spaces (1905.11293v4)

Abstract: Grasp synergies represent a useful idea to reduce grasping complexity without compromising versatility. Synergies describe coordination patterns between joints, either in terms of position (joint angles) or effort (joint torques). In both of these cases, a grasp synergy can be represented as a low-dimensional manifold lying in the high-dimensional joint posture or torque space. In this paper, we use the term \textit{Mechanically Realizable Manifolds} to refer to the subset of such manifolds (in either posture or torque space) that can be achieved via mechanical coupling of the joints in underactuated hands. We present a method to optimize the design parameters of an underactuated hand in order to shape the Mechanically Realizable Manifolds to fit a pre-defined set of desired grasps. Our method guarantees that the resulting synergies can be physically implemented in an underactuated hand, and will enable the resulting hand to both reach the desired grasp postures and achieve quasistatic equilibrium while loading the grasps. We demonstrate this method on three concrete design examples motivated by a real use case, and evaluate and compare their performance in practice.

• The paper presents a dual-layer optimization framework that integrates torque and posture manifold optimization to achieve efficient underactuated hand design.
• It introduces Mechanically Realizable Manifolds that enable physical coupling of joint postures and torques for versatile grasping.
• Numerical simulations and prototypes validate that the design efficiently balances actuator limitations with multi-joint coordination for robust performance.

Overview of Underactuation Design for Tendon-driven Hands via Mechanically Realizable Manifolds Optimization

This essay discusses a paper focused on the optimization of underactuated tendon-driven robotic hands, specifically targeting the design of these systems through the articulation of two key concepts: Mechanically Realizable Manifolds in joint posture and torque spaces. The work investigates how low-dimensional manifolds can be physically realized in underactuated hand designs, effectively reducing complexity while maintaining versatility in robotic grasping.

Mechanically Realizable Manifolds

The paper introduces the notion of Mechanically Realizable Manifolds, which are characterized by representing possible configurations that can be achieved through mechanical coupling in underactuated hands. The manifolds encompass both joint posture and joint torque spaces, providing an integrated approach for configuring hand architecture to perform a range of pre-defined desired grasps.

Optimization Framework

To design a multifaceted underactuated hand, the authors propose a dual-layer optimization framework. This method balances the need for pre-contact posture attainment with the imperative of post-contact torque equilibrium. The approach delineates three optimization layers:

• Torque Manifold Optimization: Focused on post-contact equilibrium by minimizing the deviation of actual net torque from desired states.
• Inter-tendon Posture Manifold Optimization: Aiming to align the coordination of multiple tendons actuated by a common motor system.
• Intra-tendon Posture Manifold Optimization: Seeking to optimize the synergy of joints driven by a single tendon.

Numerical Results and Evaluation

The paper details three design examples for underactuated hands, particularly targeting applications in environments like the International Space Station. These examples underscore the hands' ability to balance a limited number of actuators against a broader range of required joints, emphasizing the approach's efficacy in achieving the desired grasps. Through a combination of detailed computational simulations and physical hand prototypes, the validated designs illustrate the optimized synergy and robustness of grasping different object types.

Implications and Future Directions

The introduction and application of Mechanically Realizable Manifolds in the optimization framework have significant implications for both the theoretical understanding and practical implementation of underactuated robotic hands. The framework not only ensures that resulting synergies are physically realizable but also enhances the hand's adaptability and functionality in complex, real-world tasks.

Future research directions may include expanding the design optimization to encompass a more comprehensive range of robotic configurations, possibly integrating adaptive synergies beyond predefined roles and potentially extending the approach to collaborative multi-robot systems.

Conclusion

This paper contributes to the design and optimization of underactuated tendon-driven hands by articulating and leveraging the synergy inherent in Mechanically Realizable Manifolds. Its dual-layer optimization framework represents a structured approach to achieving robustness and flexibility in robotic hand design, providing a foundation for ongoing advancements in robotic grasping technology.