A Direct-drive Haptic Thimble for Fine Tactile Feedback in Virtual and Robotic Manipulation

A Direct-drive Haptic Thimble for Fine Tactile Feedback in Virtual and Robotic Manipulation Daniele Leonardis, Federica Serra, Marcello Palagi, Simone Marcheschi, Cristian Camardella Institute of Mechanical Intelligence, Sant’Anna School of Advanced Studies, Pisa, Italy Abstract—In this work, we present a direct-drive approach applied to an haptic thimble based on soft actuated belts. Building on a previous gearmotor-based design, the method leverages direct transmission to enhance linearity and to extend the output bandwidth of the device, thereby enriching the rendering of dynamic tactile feedback. While the approach reduces the maximum achievable force, it enables wide-bandwidth interaction in a thin and lightweight form factor. We validate the device in a virtual reality manipulation experiment, where physical simulation requires careful pick-and-place of virtual cubes. We also propose a proof-of-concept implementation in a teleoperation setup, where the thimble is coupled with a robotic hand equipped with sensitive pressure-based fingertip sensors. Results in VR highlight the effectiveness of the feedback in reducing the average virtual indentation and improving repeatability of the fine manipulation task. Index Terms—Tactile, direct-drive, haptic, manipulation, virtual reality, teleoperation I. INTRODUCTION Tactile feedback at the fingertips is essential for dexterous interaction in both virtual environments and teleoperation [1], however, design of fingertip haptic devices remains challenging due to mechanical constraints and the need to render a wide variety of tactile cues [2]. Focusing on just the rendering of normal forces and indentation, design solutions must balance different competing requirements: the intensity of the output force, the dynamic bandwidth of the signal, and the overall signal-to-noise ratio. Conventional solutions based on gearmotors address this trade-off by providing relatively high output forces with limited added mass [3]. Yet, the mechanical transmission introduces friction, backlash, and vibrations from gear teeth, which significantly reduce the output bandwidth and contaminate the rendered signals with noise. Such limitations are critical because fingerpad mechanoreceptors are highly sensitive to dynamic and transient events [4]. High-frequency transients and fast variations are particularly informative, contributing to texture discrimination, perception of contact events and the stability of grasp [5], [6]. In this work, we build upon the approach proposed in [7], which employed gearmotors to tension a thin and compliant belt interface for rendering normal indentation and lateral stretch. We re-design the device implementing a direct-drive actuation system (Figure 1) aiming at increasing the linearity and dynamic response of the rendered signals. Fig. 1. a) Section view of the direct-drive haptic thimble b) Picture of the device worn at the fingertip c) Characterization of the current-to-force characteristic measured as normal force at the fingerpad d) Output bandwidth of the device II. METHODS The proposed fingertip haptic device employs two rotary electromagnetic motors (Minebea PNN13RB08RD, 8mm diameter), operating at 3.7 V. Each motor’s output shaft is coupled to a flexible band that encircles the fingertip, as shown in Figure 1 (a) and (b). Thanks to the direct-drive implementation, the output torque and in turn the linear force applied by the motors to the belt can be controlled in feedforward by adjusting the current supplied to the motor coils. The thimble is fabricated using stereolithographic 3D printing (SLA) with soft resin (Liqcreate Elastomer X), which enabled a one-piece, structured design of the thimble. The structure has cavities to allow passage of the belt directly towards the fingerpad, without contact and friction with the thimble. A. Device Characterization We characterized the device at the bench using a miniature force sensor (Optoforce N10) in place of the user’s finger. The current-to-force characteristic is measured in quasi-static conditions applying slow voltage ramps to the motors. THe 2025 I-RIM Conference October 17-19, Rome, Italy ISBN: 9788894580570 10.5281/zenodo.17629890 247 Fig. 2. a) The pick-and-place task implemented in VR and used for evaluation of the haptic device. b) Results of the VR manipulation experiment. output bandwidth is measured using a logaritmic chirp signal with fixed amplitude, ranging from 1 to 250 Hz. B. Virtual Reality Experiment Five participants (3 male and 2 female, aged 24–39) used the thimble device to pick and place a virtual cube (Figure 2(a)) between two target positions. Grasping had to be modulated to avoid both slippage and breaking of the cube. Hand tracking was performed using a Leapmotion device. Physical simulation was run at 100Hz, streaming references over Wi-Fi (UDP protocol) from the host PC to the haptic devices. Each participant completed a 20 trials session under two conditions: HV (haptic + visual feedback) and V (visual feedback only). A virtual spring was implemented in the physical simulation to couple the tracked position of the fingers with a spherical collider interacting in the physical simulation (blue spheres in Figure 2 (a)). The indentation between the physical position of the fingertips and the spherical collider was used to proportionally drive the force of the actuated belt, and measured for data analysis. C. Teleoperation Proof-of-Concept As a further implementation example, the proposed haptic thimbles have been integrated in a teleoperation scenario, implementing a Franka Emika Panda robot, and an improved version of the CORA robotic hand [8]. For this setup, dedicated fingerpad sensors, highlighted in Figure 3, were used. They implement a pressure sensor (Bosch BMP280) embedded in an airtight silicone rubber dome, enabling the detection of fine contact transients on a wide area of the fingerpad, with a relative accuracy of 0.1 mbar. The teleoperation task involved the grasping and insertion of a plastic water bottle in slots having different position and orientation. III. RESULTS AND CONCLUSIONS Results of the fingertip characterization shown in Figure 1(c) and (d) highlight the linearity of the direct-drive belt transmission, and the wide output bandwidth of the device, ranging up to 120 Hz. Consistency of the feedback was shown in the virtual reality manipulation experiment (Figure 2(b) ), where the addition of haptic feedback significantly reduced Fig. 3. The robotic teleoperation setup, implementing dedicated, sensitive fingerpad sensors at the robotic side to measure fine interaction signals rendered by the haptic thimbles (paired T-Test, Indentation mean HV = 5.2 mm, V = 7.1 mm, p < 0.05) the average virtual indentation and the variance between repetitions (paired T-Test, Indentation variance HV = 3.1 mm, V = 4.7 mm, p < 0.01). Overall, the paper presented a viable and effective solution for informative, dynamic tactile feedback rendered by a thin and wearable fingertip device, compatible with vision-based tracking systems. The implementation in the teleoperation setup represents a proofof-concept of the approach to introduce tactile feedback in a robotic manipulation task. The teleoperation task could be accomplished with coherent tactile feedback; further investigations with experimental comparisons in different feedback conditions will be carried on as next steps. IV. ACKNOWLEDGEMENTS This work was supported by Social and hUman ceNtered XR (SUN) project - Horizon Europe Research & Innovation Programme Grant agreement N. 101092612. REFERENCES [1] C. Pacchierotti and D. Prattichizzo, “Cutaneous/tactile haptic feedback in robotic teleoperation: Motivation, survey, and perspectives,” IEEE Transactions on Robotics, vol. 40, pp. 978–998, 2023. [2] A. Frisoli and D. Leonardis, “Wearable haptics for virtual reality and beyond,” Nature Reviews Electrical Engineering, vol. 1, no. 10, pp. 666– 679, 2024. [3] S. B. Schorr and A. M. Okamura, “Three-dimensional skin deformation as force substitution: Wearable device design and performance during haptic exploration of virtual environments,” IEEE transactions on haptics, vol. 10, no. 3, pp. 418–430, 2017. [4] S. J. Bensmaıa and M. Hollins, “Complex tactile waveform discrimination,” The Journal of the Acoustical Society of America, vol. 108, no. 3, pp. 1236–1245, 2000. [5] M. Wiertlewski, J. Lozada, and V. Hayward, “The spatial spectrum of tangential skin displacement can encode tactual texture,” IEEE Transactions on Robotics, vol. 27, no. 3, pp. 461–472, 2011. [6] F. Clemente, M. D’Alonzo, M. Controzzi, B. B. Edin, and C. Cipriani, “Non-invasive, temporally discrete feedback of object contact and release improves grasp control of closed-loop myoelectric transradial prostheses,” IEEE Transactions on Neural Systems and Rehabilitation Engineering, vol. 24, no. 12, pp. 1314–1322, 2015. [7] K. Minamizawa, S. Fukamachi, H. Kajimoto, N. Kawakami, and S. Tachi, “Gravity grabber: wearable haptic display to present virtual mass sensation,” in ACM SIGGRAPH 2007 emerging technologies, 2007, pp. 8–es. [8] D. Leonardis and A. Frisoli, “Cora hand: a 3d printed robotic hand designed for robustness and compliance,” Meccanica, vol. 55, no. 8, pp. 1623–1638, 2020. 248