This study examines the efficacy of gesture-based versus touch-based control interfaces in managing unanticipated scenarios in human-robot collaborative assembly (HRCA). As collaborative robots (cobots) become increasingly integrated into industrial environments, there is a need for control interfaces that enable workers to respond to unanticipated deviations from the assembly process. Such control interfaces must be intuitive to use, acceptable to the worker, and not result in an increased workload. We conducted a controlled experiment with 30 participants, comparing the performance of gesture-based against touch-based control interfaces in an assembly task, which included simulated disruptions. Workload was evaluated using the NASA Task Load Index (NASA-TLX), and acceptance via a Technology Acceptance Model (TAM), adapted for HRCA. The results demonstrate no statistically significant differences between the two measures. However, they indicate that touch control results in a lower workload and higher user acceptance, which may be attributed to users' familiarity with touch interfaces.

Human-Robot Collaboration (HRC) in dynamic assembly environments requires intuitive communication for seamless interactions. Eye gaze, a nonverbal cue, is critical for interpreting human intent and enhancing robot adaptability. This research introduces a three-tier framework categorizing eye gaze into fixation, scanning, and task-switching behaviors, aiming to improve real-time robot responsiveness. Preliminary evaluations with multiple human subjects demonstrated the potential of this classification to enhance task alignment and detect anomalies in human behavior. Future work will explore automated personalization of the algorithm for individuals and integrate multimodal data to refine adaptability in human-robot collaborative systems.

How best to attach exoskeletons to human limbs is an open and understudied problem. In the case of upperbody exoskeletons, cylindrical handles are commonly used attachments due to ease of use and cost effectiveness. However, handles require active grip strength from the user and may result in undesirable flexion synergy stimulation, thus limiting the robot’s effectiveness. This paper presents a new design, the Eminence Grip, for attaching an exoskeleton to the hand while avoiding the undesirable consequences of using a handle. The ergonomic design uses inverse impedance matching and does not require active effort from the user to remain interfaced with the exoskeleton. We compare the performance of the Eminence Grip to the handle design in a healthy subject target reaching experiment. The results show that the Eminence Grip achieves similar performance to a handle in terms of relative motion between the user and the exoskeleton while eliminating the requirement of grip force to transfer loads to/from the exoskeleton and avoiding stimulation of the flexion synergy. Taken together, the kinematic equivalence and improvement in ergonomics suggest that the Eminence Grip is a promising exoskeleton-hand attachment interface supporting further experiments with impaired populations.

Much effort has been placed into the development of robotic devices to support, rehabilitate, and interact with humans. Despite these advances, reliably modeling the neuromuscular changes in human motion resulting from a robotic intervention remains difficult. This paper proposes a method to uncover the relationship between robotic intervention and human response by combining surface electromyography (sEMG), the musculoskeletal modeling platform OpenSim, and artificial neural networks (ANNs). To demonstrate the method, a one degree of freedom (DOF) elbow flexion-extension motion is performed and analyzed. Preliminary results show that while the robot provides assistance to the subject, it also appears to produce other unexpected responses in the movement. Further investigation using the new method reveals the neuromuscular effect of an unintended resistance to the subject's motion applied by the robot as it enforces a speed slower than the subject selects. The characterization of the differences in expected and actual interaction is enabled by the method presented in this paper. Thus, the method uncovers previously obscured aspects of human robot interaction, and creates possibilities for new training modalities.