Identification and validation of the dynamic model of a tendon-driven anthropomorphic finger

TL;DR

The paper addresses the lack of a complete identification framework for the dynamic behavior of tendon-driven fingers by formulating a general model that couples rigid-body dynamics, moment-arm nonlinearity, joint viscoelasticity, and tendon friction. It proposes a sequential identification-then-validation pipeline and a modular 18-module test-bed, and demonstrates its viability on a Dexmart-inspired 3D-printed finger, identifying kinematic/inertia parameters, the coupling matrix, viscoelastic torques, tendon friction, and fingertip forces. Key findings include nonlinear moment-arm behavior, accurate tendon-excursion validation, quantification of MCP viscoelasticity with elastic and friction components, and finite-tension tendon-friction curves predicting about 60% transmission efficiency at moderate loads. The approach lays the groundwork for consistent musculoskeletal hand modeling—potentially extended to cadaver hands—to enable accurate inverse-dynamics verification and improved dexterous control of robotic hands.

Abstract

This study addresses the absence of an identification framework to quantify a comprehensive dynamic model of human and anthropomorphic tendon-driven fingers, which is necessary to investigate the physiological properties of human fingers and improve the control of robotic hands. First, a generalized dynamic model was formulated, which takes into account the inherent properties of such a mechanical system. This includes rigid-body dynamics, coupling matrix, joint viscoelasticity, and tendon friction. Then, we propose a methodology comprising a series of experiments, for step-wise identification and validation of this dynamic model. Moreover, an experimental setup was designed and constructed that features actuation modules and peripheral sensors to facilitate the identification process. To verify the proposed methodology, a 3D-printed robotic finger based on the index finger design of the Dexmart hand was developed, and the proposed experiments were executed to identify and validate its dynamic model. This study could be extended to explore the identification of cadaver hands, aiming for a consistent dataset from a single cadaver specimen to improve the development of musculoskeletal hand models.

Figures (9)

• Figure 1: The identification and validation methodology overview: identification (ID) and validation (VD) experiments in sequence targeting different parameters. The tendon-driven finger is depicted in grey, in which the solid and hollow circles mean the free-to-move and fixed joints, respectively. The parameters to be identified or validated in each experiment are formulated in blue rows. The necessary parameters to identify the target variables in the model are listed in the green row. The measurements in each experiment are highlighted in red. The mathematical equations of validation scenarios are summarized in the yellow row.
• Figure 2: The overview of the identification test-bed and robotic finger. The external motor with torque sensor is shown in (b). The external 3-dimensional force sensor with the thimble is depicted in (c) for measuring the fingertip force. Fig. (d) illustrates the finger design based on the Dexmart hand with joint shafts and encoders. A rubber band is integrated as a stiffness component at the MCP joint. One of the eighteen actuation modules with a customized pulley mechanism and tendon force sensor is shown in (e).
• Figure 3: The illustration of the finger with the tendon configuration and associated notation. The kinematic and estimated mass parameters are depicted in (b).
• Figure 4: The moment arm curves from ID.1 experiment. The curves of all four active tendons at associated joints are visualized. The positive and negative moment arm values refer to the directions of flexion and extension, respectively.
• Figure 5: The results of the VD.1 experiment. The joint trajectories of MCP, PIP, and DIP are visualised in (a). All four active tendon excursions are estimated online (dashed lines) and compared to the measurements (solid lines). The errors between the estimation and measurement of tendon length are depicted in (c).