Conversational agents (CAs) have been successfully implemented to deliver digital health interventions. However, most of their applications focus on non-communicable diseases, e.g., diabetes. RehabCoach, a CA-based smartphone application designed to assist stroke survivors during at-home unsupervised rehabilitation, utilizes multiple CAs to deliver digital interventions for therapy adherence. Matthias, one of the CAs of RehabCoach, is a novel Large Language Model-based CA that leverages Retrieval Augmented Generation (RAG) to answer questions related to ReHandyBot, an upper limb rehabilitation device. In this work, we assessed the preciseness and conciseness of Matthias’s answers to 14 device-related questions. For each question, Matthias generated 6 answers, based on 3 different prompts and 2 different data sources, i.e., a conversational dataset containing dialogues about how to operate the device and the device manual of instructions. Results from a blinded evaluation performed by two device experts highlight that the responses of the CA generated using the conversational dataset were considered more concise (82,14%) and better overall (60,70%). This work shows the potential of using conversational data for device-assisted rehabilitation CAs deploying RAG.

The paper describes a motion planning and control software architecture developed for the automation of a surgical robot. The considered surgical robot is a dual-arm prototype developed with a redundant and modular mechanical structure, designed to be reconfigured for different surgical tasks, and with a hybrid parallel/serial kinematics. The motion planning solution proposed in the paper includes both an online collision-free path planner, based on the RRT-Connect algorithm, and a generator of predefined motion primitives. This solution allows the multi-arm robot to autonomously execute the complex motion patterns required for a suturing task. Since such motion patterns are specified in the Cartesian space, an efficient and univocal solution of the inverse kinematics of the robot, which is a challenging problem due to its hybrid structure, is another crucial issue addressed in the paper.

Introducing some form of autonomy in robotic surgery is being considered by the medical community to better exploit the potential of robots in the operating room. However, significant technological steps have to occur before even the smallest autonomous task is ready to be presented to the regulatory authorities. In this paper, we address the initial steps of this process, in particular the development of control concepts satisfying the basic safety requirements of robotic surgery, i.e., providing the robot with the necessary dexterity and a stable and smooth behavior of the surgical tool. Two specific situations are considered: the automatic adaptation to changing tissue stiffness and the transition from autonomous to teleoperated mode. These situations replicate real-life cases when the surgeon adapts the stiffness of her/his arm to penetrate tissues of different consistency and when, due to an unexpected event, the surgeon has to take over the control of the surgical robot. To address the first case, we propose a passivity-based interactive control architecture that allows us to implement stable time-varying interactive behaviors. For the second case, we present a two-layered bilateral control architecture that ensures a stable behavior during the transition between autonomy and teleoperation and, after the switch, limits the effect of initial mismatch between master and slave poses. The proposed solutions are validated in the realistic surgical scenario developed within the EU-funded I-SUR project, using a surgical robot prototype specifically designed for the autonomous execution of surgical tasks like the insertion of needles into the human body.

In this paper, a new hand exoskeleton device using a three-layered sliding spring mechanism is presented. In contrast to state of the art hand exoskeleton mechanisms (typically link, wire or pneumatically driven), the proposed mechanism is driven through large deformations of the compliant mechanism body. The mechanism can be made compact and lightweight by adequately positioning the compliant elements. In addition, the mechanism is designed to distribute 1-DOF actuated linear motion into three rotational motions of the finger joints, which translate into natural finger flexion/extension. The primary application of the proposed mechanism is to provide robotic support during physical therapy at the hospital (e.g. Continuous Passive Motion). However, thanks to its light and wearable structure, the proposed device could also be used at home as an assistive/therapeutic device to support activities of daily living. We introduce the mechanical structure of the three-layered sliding spring mechanism, present a prototype implementation as a hand exoskeleton device, and provide a preliminary evaluation.

Robot-assisted rehabilitation can complement conventional rehabilitation after stroke, by increasing the duration and intensity of therapy and providing precise and objective measurements of interaction dynamics and performance. Such information can be used to drive assist-as-needed control strategies or to complement clinical assessments by reconstructing the scores from robot data. This paper presents the ReHapticKnob, a new end-effector-based hand rehabilitation robot with unique sensing and actuation capabilities for therapy of grasping and forearm rotation tasks. A compact design with high stiffness and high-fidelity instrumentation is presented, allowing for precise assessment and dynamic interaction. The device has two degrees of freedom (DOF), allowing independent control of hand opening/closing and forearm rotation. Each degree of freedom is equipped with a brake to allow independent training of either DOF or to assess isometric force through the two six-axis force/torque sensors located beneath the exchangeable finger fixations. The design, safety features and performance evaluation of the device are discussed and preliminary results from a study on grasping performed with healthy subjects are presented.

The neural mechanisms of fine motor control and recovery, e.g. after a stroke, are not fully understood, nor how these are influenced by different types of motor therapies, leaving potential for optimization of current rehabilitation strategies. This paper presents the development and evaluation of a novel robotic tool for fMRI-based neuroscience studies allowing to investigate the neural mechanisms of dynamic precision grip and track therapy-induced neuroplasticity. In this proof of principle study we investigate the feasibility of high-fidelity haptic interaction with human motion using remote sensing and actuation. A cable-spring mechanism transmits force to the thumb and index finger in an unconstrained manner, actuated over a stiff cable transmission. Characterization of the prototype with a transmission length of two meters revealed good dynamic performance including a 16 Hz open loop force bandwidth and a maximal output force of 28 N. Combined with a remote and shielded conventional electromagnetic actuator, this device could be used to investigate the neural correlates of precision grasping as well as the effect of different hand function therapies on the neural correlates of motor recovery after stroke.

Haptic interfaces compatible with functional magnetic resonance imaging (fMRI) are finding increasing interest as a tool to explore the neural correlates of human motor control and related dysfunctions. To achieve safety and MR compatibility, such devices have mainly relied on unconventional actuation methods suffering from limited bandwidth and non-linearities. This has resulted in complex control and restricted their use in applications involving fine and dynamic interaction with the hand and fingers. To address these limitations, we propose a concept for a shielded high-performance actuation system to be located inside the MR room, evaluate the effectiveness of the shielding and perform detailed MR compatibility tests. A conventional electromagnetic actuator is located within a steel shield to prevent mutual disturbance with magnetic fields of the scanner, which, together with power and control hardware, is placed within a Faraday cage with only a fiber-optical USB link to the control room. Detailed compatibility tests show that disturbing dynamic electromagnetic fields generated by the actuation system are well below the detectable threshold of the scanner, and actuator performance is not degraded by the MR environment. In combination with a light and stiff cable or rod transmission, the presented actuator technology, providing high transparency and force bandwidth, paves the way for fMRI-based neuroscience studies, e.g., to investigate the fine motor control of hand and fingers.

Robot-assisted rehabilitation has only recently begun to be applied to improvement of hand function after stroke. In a preliminary study, involving 4 post-stroke subjects, more than 2 years following the stroke, we have been able to show that 8 weeks of robot-assisted training leads to changes in patterns of arm and finger muscle activation. The patterns were quantified in terms of synchronous muscle synergies which allowed for comparison with muscle activation patterns of healthy age-matched subjects. We found that the muscle synergies of the post-stroke subjects became more similar to those of the healthy subject group following training.

The systems which have been developed are safe, easily transportable and offer various training possibilities. Virtual-reality games have been implemented to increase motivation during the 8 weeks of training with post-stroke subjects. Clinical trials have shown that chronic stroke patients can perform the required tasks and that training improves their motor function.

This paper describes a two-degrees-of-freedom haptic interface to investigate the brain mechanisms of human motor control, which is capable of safely and gently interacting with human arm motion during functional magnetic resonance imaging (fMRI). A hydrostatic transmission separates the interface into a master and an MR compatible slave system, allowing the placement of all interfering components outside the electromagnetic shield of the MR room. The transmission mirrors force and motion of the master actuators on the slave system placed close to the MR scanner. The parallel architecture takes advantage of the linear MR compatible actuators and allows human subjects to perform reaching movements comfortably in the small workspace limited by the dimensions of the MR scanner and the biomechanics of the arm. The kinematic structure of the slave interface was optimized with respect to the available space and types of movements to be investigated. Materials were chosen based on their MR compatibility, their stiffness and weight. The interaction force with the subject is measured over two optical force sensors, located close to the output of the interface. Two shielded optoelectronic encoders measure the extension of the slave hydraulic pistons. Detailed tests demonstrated the fMRI compatibility even during movement of the interface

The strong impairment of motor functions in stroke survivors affects daily activities such as eating, manipulating objects or writing. Our goal is to induce long lasting improvements in such tasks by having patients perform systematic exercises using haptic interfaces. This paper describes a novel two-degrees-of-freedom interface which we have developed to help stroke patients gradually recover the ability to open and close the hand and manipulate knobs. Different solutions are studied and a design consisting of two parallelogram structures interacting with the fingers is proposed. The mechanical design offers the possibility to adapt the interface to various hand sizes and finger orientations, and to right or left-handed subjects. Design kinematics as well as actuation and system control are described. Several knobs are proposed to interact with patients, especially a cone mechanism to train a complete opening movement from a strongly contracted and closed hand to a large opened position. The interaction force with the subject is measured over four force sensors located close to the output of the interface. A preliminary study has been performed to evaluate the performances of the haptic interface