Alessandro Altobelli

ThimbleSense: A Fingertip-Wearable Tactile Sensor for Grasp Analysis

Accurate measurement of contact forces between hand and grasped objects is crucial to study sensorimotor control during grasp and manipulation. In this work, we introduce ThimbleSense, a prototype of individual-digit wearable force/torque sensor based on the principle of intrinsic tactile sensing. By exploiting the integration of this approach with an active marker-based motion capture system, the proposed device simultaneously measures absolute position and orientation of the fingertip, which in turn yields measurements of contacts and force components expressed in a global reference frame. The main advantage of this approach with respect to more conventional solutions is its versatility. Specifically, ThimbleSense can be used to study grasping and manipulation of a wide variety of objects, while still retaining complete force/torque measurements. Nevertheless, validation of the proposed device is a necessary step before it can be used for experimental purposes. In this work, we present the results of a series of experiments designed to validate the accuracy of ThimbleSense measurements and evaluate the effects of distortion of tactile afferent inputs caused by the devices rigid shells on grasp forces.

Three-digit grasp haptic device with variable contact stiffness for rehabilitation and human grasping studies

This paper describes an haptic system designed to vary the stiffness of three contact points in an independent and controllable fashion, by suitably regulating the inner pressure of three pneumatic tactile displays. At the same time, the contact forces exerted by the user are measured by six degree-of-freedom force sensors placed under each finger. This device might be profitably used in hand rehabilitation and human grasping studies. We report and discuss preliminary results on device validation as well as some illustrative measurement examples.

Reflex control of the Pisa/IIT SoftHand during object slippage

In this work, to guarantee the Pisa/IIT SoftHands grasp robustness against slippage, three reflex control modes, namely Current, Pose and Impedance, are implemented and experimentally evaluated. Towards this objective, ThimbleSense fingertip sensors are designed and integrated into the thumb and middle fingers of the SoftHand for real-time detection and control of the slippage. Current reflex regulates the restoring grasp forces of the hand by modulating the motors current profile according to an update law. Pose and Impedance reflex modes instead replicate this behaviour by implementing an impedance control scheme. The difference between the two latter is that the stiffness gain in Impedance reflex mode is being varied in addition to the hand pose, as a function of the slippage on the fingertips. Experimental setup also includes a seven degrees-of-freedom robotic arm to realize consistent trajectories (e.g. lifting) among three control modes for the sake of comparison. Different test objects are considered to evaluate the efficacy of the proposed reflex modes in our experimental setup. Results suggest that task-appropriate restoring forces can be achieved using Impedance reflex due to its capability in demonstrating instantaneous and rather smooth reflexive behaviour during slippage. Preliminary experiments on five healthy human subjects provide evidence on the similarity of the control concepts exploited by the humans and the one realized by the Impedance reflex, highlighting its potential in prosthetic applications.

An instrumented manipulandum for human grasping studies

This work presents a novel haptic device to study human grasp, which integrates different technological solutions thus enabling, for the first time, to achieve: (i) a complete grasp characterization in terms of contact forces and moments; (ii) an estimation of contact point location for varying-orientation contact surfaces; (iii) a compensation of force/torque offsets and estimation of the mass and center of mass of the device, for different orientations and configurations in the workspace; (iv) different stiffness properties for the contact points, i.e. rigid, compliant non-deformable and compliant deformable, thus allowing to study the effects of cutaneous cues in multi-finger grasps. In addition, given the modularity of the architecture and the simple mechanism to attach/detach the contact modules, this structure can be easily modified in order to analyze different multi-finger grasp configurations. The effectiveness of this device was experimentally demonstrated and applications to neuroscientific studies and state of the art of devices for similar investigations are discussed in depth within the text.

Human Hand Motor Control Studies

This chapter aims to give a brief overview on the complexity that is typical of hand motor control studies. Starting from biomechanical hand models to recent theories on motor control in grasping tasks, in this dissertation the important factors which affect grasp proprieties are dealt. The mechanical structure of human hand is extremely complex and difficult to model; its rigid internal framework is made by 27 bones that are moved by 18 intrinsic muscles and 18 extrinsic muscles coupled by a network of tendons. To have a simple hand model, at least 23–24 DoFs are needed: 4 DoFs for each finger, 5 for the thumb, 1 for the radioulnar joint, and 2 at the wrist. In a more detailed model, the number of DoFs increases just taking into account the hand’s capability to create a palmar arch when it closes. A complete biomechanical model includes 36 muscles coupled to the bones by a complex tendons network; moreover, several biomechanical constraints have to be included in the model. Joint limits or finger dimensions are clear examples of constraints which can affect the interaction of the hand with the world, and additional constraints arise from the coupling of tendons and muscles. Some muscles span several phalanges, making it difficult to move only one joint independently; for example, the flexor digitorum superficialis (FDS) and extensor digitorum communis (EDC) muscles are divided on each finger; therefore, a contraction of these muscles engages several hand joints. Understanding how humans exploit biomechanics and sensory feedback of hand in everyday tasks is a challenging topic that still is not completely understood. Several studies and theories, focused on kinematic and grasping tasks have been developed. In the next section, I will give an introduction on the most recent studies which focus on important aspects of manipulation: (i) hand control in pre-grasp phase, (ii) grasp force distributions, (iii) muscle activations, and (iv) impedance control.

Haptic Devices for Studies on Human Grasp and Rehabilitation

The investigation of the strategies of human motor control in grasping task represents a relevant topic in neuroscience with applications in robotics. Such an investigation requires the development and the exploitation of sensing tools and devices, which are able to record all the necessary information, and for this purpose, new custom devices are developed and exploited. The ambitious goal of this work is twofold: (1) to advance the state of the art on human strategies in manipulation tasks and provide tools to assess rehabilitation procedure and (2) to investigate human strategies for impedance control that can be used for human robot interaction and control of myoelectric prosthesis. Although the goal complexity requires many efforts, this book achieved tangible and original contributions that are suitable for robotic/prosthetic and human motor control studies.

Sensorized Object Approach

In this section, I presented two instrumented objects designed to be grasped with three-digit finger posture. Four different force/torque sensors are fixed in a profitably configuration to allow measures of contact forces exerted by each finger and the external wrench. Changing the stiffness at each contact independently is possible by two different haptic solutions. Experimental results show the validity and utility of proposed devices to investigate human grasp proprieties. As evidenced in the previous chapter, these systems constrain the hand posture throughout the grasp phase but allow to measure all: (i) the contact positions and (ii) the force/torque components at each contact. In the next subsections, the devices are described and validated in a more detailed way.

Wearable Approach: ThimbleSense, a Fingertip-Wearable Tactile Sensor for Grasp Analysis

In this chapter, I tested and validated the “ThimbleSense” system, a new wearable individual-digit force/torque sensor developed and presented by Battaglia et al (Thimblesense: an individual-digit wearable tactile sensor for experimental grasp studies, 2014 IEEE International Conference on Robotics and Automation (ICRA), 2728–2735 (2014) [1]). This system aims to integrate the grasp analysis achievable with sensorized objects presented in the Chap. 2 where the position of the contact surfaces is fixed. ThimbleSense allows to obtain measurement of contact forces between hand and grasped objects without constrains on the hand postures. The main advantage of this approach with respect to more conventional solutions is the possibility of being versatile without losing accuracy: instead of building many sensorized objects for different experiments, it is possible to employ ThimbleSense to study grasps of a variety of objects, while still retaining the complete force/torque measurements. Unfortunately, ThimbleSense rigid shells are interposed between the fingertip and grasped object. This inevitably modifies the physiological mechanical deformation that would otherwise occur at the bare fingerpad in direct contact with objects. An experiment shows that excessive grip forces are attenuated with training as the subjects familiarize him/herself with the ThimbleSense. This effect evidences that sensorized object and wearable object are both necessary to investigate different aspects of human grasp. In this work, I briefly introduce the concept and the implementation of individual-digit wearable force/torque sensors; later, I present some experiments to validate the device (Battaglia et al, Thimblesense: A fingertip wearable tactile sensor for grasp analysis (2015) [2]). In particular, my contributions in this work are: (i) to define a procedure to handle F/T sensor offsets and to estimate the inertial parameters of the device in static conditions and (ii) to validate the measures of the device with some experiments. Results evidenced that internal forces estimated with the ThimbleSense are inside the null space of the grasp matrix.

Effect of Homogenous Object Stiffness on Tri-Digit Proprieties

In the previous chapter, I evidenced a relation between finger stiffness and EMG signals of principal finger muscles; in this study, I monitored the same muscle activity throughout the grasp of a instrumented manipulandum with different stiffnesses at contact points. To investigate the effect of the stiffness at contact point on the grasping force distribution, I profitably used the tools and method presented in the previous chapters. In effect, grasping of compliant objects presents additional uncertainties and Winges et al. (Winges et al, J Neurophysiol, 101(5), 2447–2458, 2009 [1]) showed that during a grasp, when one or two contact points are compliant, the activation patterns of finger muscles are different with respect to the case where the contact points are rigid. Besides analyzing the grip forces, to fully understand the control of hand grasping by the CNS, it is important to study how the hand stiffness is regulated during a grasp: stiffening behavior is commonly realized to stabilize movement or to fix posture in isometric tasks (Humphrey and Reed, Adv Neurol, 39, 347–372, 1983, [2]) and recent findings suggest that to some extent, grip stiffness is independent from grip force (Hoppner et al, Plos one 8(12), e80889, 2013, [3]). This study (Godfrey et al, Effect of homogenous object stiffness on tri-digit grasp properties, EMBC, 2015, [4]) aims to investigate the relation between object compliance and grasping stiffness of the hand. To achieve this goal, 11 subjects perform a grasp experiment exploiting a modified version of the manipulandum (see Sect. 2.2) with three different contact modules. Each module is characterized by a certain level of stiffness: rigid, high, medium, or low stiffness. The experiment consisted of four blocks of trials, corresponding to the four different levels of stiffness; in each trial, the subject grasped and lifted the manipulandum 25 times, while the EMG was recorded from the flexor digitorum superficialis (FDS) and extensor digitorum communis (EDC). These two muscles are the main finger antagonist pair and thus can be used to monitor the EMG activity resulting in the production of grasp force as well as overall hand stiffness; this assumption is in agreement with the capability of the human control system to increase hand stiffness exploiting the co-contraction of antagonist muscles (Smith, Can J Physiol Pharmacol, 59(7), 733–747, 1981, [5]).

Effect of Homogenous Object Stiffness on Tri-digit Grasp Properties

This paper presents experimental findings on how humans modulate their muscle activity while grasping objects of varying levels of compliance. We hypothesize that one of the key abilities that allows humans to successfully cope with uncertainties while grasping compliant objects is the ability to modulate muscle activity to control both grasp force and stiffness in a way that is coherent with the task. To that end, subjects were recruited to perform a grasp and lift task with a tripod-grasp device with contact surfaces of variable compliance. Subjects performed the task under four different compliance conditions while surface EMG from the main finger flexor and extensor muscles was recorded along with force and torque data at the contact points. Significant increases in the extensor muscle (the antagonist in the task) and co-contraction levels were found with increasing compliance at the contact points. These results suggest that the motor system may employ a strategy of increasing co-contraction, and thereby stiffness, to counteract the decreased stability in grasping compliant objects. Future experiments will examine the extent to which this phenomenon is also related to specific task features, such as precision versus power grasp and object weight.