May the Force Be With Us: Enhancing Real-World Human-Human Interactions With Bidirectional Haptic Feedback Via Wearable Robots

Joint actions are fundamental to human life as they enable individuals to interact and collaborate to achieve goals that often exceed their individual capabilities. Many daily tasks rely on physical interaction and span a wide spectrum, from handing a glass to a friend or moving a table together to complex tasks like teaching a child to walk or assisting a patient through motor rehabilitation. In these contexts, haptic communication--the exchange of forces and motion through shared objects or direct physical contact--plays a central role in enabling effective interaction, whether to foster coordination, convey intent, or provide guidance. Recently, the physical Human-Robot-Human Interaction paradigm (pHRHI) has leveraged dual robotics systems to simulate physical contact between individuals by rendering virtual viscoelastic connections at their end-effector levels during joint tasks, thus enabling the investigation of the role of haptic feedback in human-human interactions. Several studies have demonstrated that dyads physically coupled through robots outperform individuals working alone, and that the improvements are modulated by two factors: the relative skill level of the individuals and the mechanical properties of the virtual viscoelastic connection, specifically the stiffness. However, these studies have been conducted in simplified, structured laboratory settings using handheld end-effector robots with limited degrees of freedom and during basic tracking or reaching tasks. As a result, their ecological validity and generalizability to more complex, real-world joint actions, particularly those involving multi-joint coordination, remain limited, and the potential of haptic feedback to enhance coordination remains underexplored. This dissertation aims to overcome these limitations and extend the pHRHI framework to real-world contexts by introducing and testing key enabling technologies: \textit{i)} an open-source software platform for managing multimodal, human-centred experiments in ecological settings, and \textit{ii)} an upper-limb wearable exoskeleton capable of delivering joint-level haptic feedback. These technologies were deployed and validated in the specific context of violin ensemble performance. Moreover, the framework was further applied to the domain of neurorehabilitation, where bidirectional haptic feedback rendered by lower-limb wearable exoskeletons was explored to support the interaction between physical therapists and individuals with incomplete spinal cord injury. The first part of the dissertation provides a comprehensive overview of the sensorimotor principles of human-human interaction and the role of haptic feedback therein, representing the theoretical basis for the work. Specifically, it highlights the key findings and open questions from the literature on pHRHI and highlights the need for more ecologically valid settings. One of the core technical contributions of the dissertation is the development of the Central Control Software (CCS), an open-source platform for synchronizing real-time data streams from multiple hardware and software modules, including wearable sensors and robotic devices. Designed for extensibility and usability, the CCS enables researchers to run complex experimental protocols beyond tightly controlled laboratory settings. The software demonstrated high real-time performance (<1% error in packet reception time), reliability (<1% of lost packets) and efficiency (<1% of RAM usage and <22% CPU usage), across different operational periods (1, 5, 10, 20 minutes)--demonstrating its stability over time--and with varying number of connected modules (4 modules tested separately or simultaneously)--demonstrating its scalability with increasing workload. Moreover, usability tests with 20 users (10 experts, 10 non-experts) revealed an excellent usability score (90/100) and comparable performance metrics across expertise, suggesting a remarkable ease of use. The CCS was subsequently used in experiments to validate a novel upper-limb exoskeleton designed to haptically assist complex upper-limb tasks, such as violin playing. The exoskeleton operates in both transparent mode, minimizing interaction torques to follow the user’s intended motion with minimal interference, and haptic mode, rendering viscoelastic torques proportional to the deviation from a reference trajectory. Validation was performed with three expert violinists in a single-user, unidirectional scenario, where musicians received haptic feedback based on a self-recorded baseline trajectory, and confirmed that the exoskeleton does not hinder free movements and preserves the natural expressivity required by the task, while rendering accurate and meaningful feedback, with torque tracking errors between 0.5 and 1.2 Nm, therefore below the just noticeable threshold for haptic stimuli. Building on this, a study involving 20 violin duos (10 amateur, 10 professional) evaluated the effects of robot-mediated bidirectional haptic feedback on coordination during ensemble performance, which traditionally relies on auditory and visual cues only. Musicians performed under four feedback conditions--auditory, auditory-visual, auditory-haptic, and auditory-visual-haptic--and were evaluated in terms of spatiotemporal coordination of the bow motion and upper limb joint angles. Results showed that haptic feedback significantly enhanced coordination, even outperforming the most familiar auditory-visual condition, and that the auditory-visual-haptic condition yielded the best overall synchrony, underscoring the value of multisensory integration in complex sensorimotor tasks. The final part of the dissertation extends the framework to neurorehabilitation. Eight individuals with incomplete spinal cord injury were coupled with physical therapists during a gait training session via lower-limb exoskeletons configured with asymmetric virtual stiffness to account for differences in motor abilities. Preliminary results suggest that asymmetric haptic coupling improves interpersonal coordination and patient's engagement, reduces therapist's effort compared to symmetric stiffness configurations, and facilitates the exchange of haptic information and dynamic adoption of roles. This asymmetric approach may foster shared control and engagement, potentially improving the therapeutic efficacy. In summary, this dissertation contributes to both the technological and experimental advancement of the pHRHI paradigm. By developing an open-source software for multimodal human-centred experiments, validating a novel exoskeleton for bidirectional haptic communication, and experimentally testing the paradigm in different real-world applications, it demonstrates how robot-mediated bidirectional haptic feedback, rendered directly at the joint level via wearable robots, can enhance coordination and performance in human-human interactions. These findings pave the way for future applications in human-robot and human-human collaborative performance, education, rehabilitation, and beyond.