Teacher-Informed Tactile Sensing Upgrade for Child–Robot Physical Interaction: A Comparison of Velostat and FBG Sensors
Social robots designed to engage with children require additional co-designed features that reflect authentic interactions. Physical human robot interaction (pHRI) in this application requires tactile sensors that are reliable and safe for natural children’s touch. This work reports on the design and assessment of two tactile sensing systems—one using Velostat and the other based on optical sensing—developed in line with teacher feedback to enhance the integration of the CASTOR robot into educational activities. Both sensors were characterized under controlled forces representative of human touch. The Velostat sensor exhibited a nonlinear force response, higher spatial resolution, and lower hysteresis, while the optical sensor demonstrated a highly linear response and greater sensitivity. The sensing systems were subsequently evaluated with 5 participants (25.73 ± 8.32 years) in a hand-shaking pHRI task. A heatmap-based analysis revealed that both sensors effectively captured the temporal and spatial distribution of contact forces, while also highlighting qualitative differences in localized forces among participants, with a maximum exerted force of 1.30 N. Overall, the results indicate that the Velostat-based sensor is advantageous for the educational scenario, as it provides detailed contact localization, exhibits low hysteresis, is more robust, and is replicable.
Characterization and functional evaluation of a fiber-optic instrumented impedance-controlled upper-limb exoskeleton
The use of active exoskeletons for musculoskeletal disorder (MSD) prevention is increasing in industrial environments. Wearable sensing technologies enable closed-loop control of these devices. However, magnetic disturbances and sensor fixation issues can compromise measurement reliability and affect exoskeleton behavior. Polymer optical fiber (POF) sensors offer intrinsic immunity to electromagnetic interference and mechanical compliance, making them a promising alternative for wearable kinematic sensing. Nevertheless, their integration within the control loop of active exoskeletons remains largely unexplored. This study investigates the impact of POF sensing modality integrated within the impedance control loop of an upper-limb exoskeleton, focusing on control performance and system dynamics. A two-stage experimental protocol was conducted under an approved ethical protocol. First, a test-bench evaluation showed that POF sensing modality achieves comparable rise time and mean absolute error to IMU sensing modality, though it resulted in a 0.565 s increase in settling time. Second, participants performed standardized lifting tasks under assisted (Exo-POF and Exo-IMU) and unassisted (No-Assist) conditions. Kinematic and electromyographic (EMG) data showed a 16–23% reduction in muscle activation while preserving range of motion. These findings demonstrate that POF sensing modality significantly influences system dynamics and human–robot interaction. Subsequently, the proof of concept suggests the feasibility of the proposed approach during lifting, handling, and lowering tasks, opening the way for future validation under representative industrial environments.