Hyunkyu Park

This paper introduces a novel pneumatic origami module that embeds a proprioceptive closed-loop shape control through seamless sensor-hardware integration. Each module incorporates a spring along the stretchable airflow passage in conjunction with an individual solenoid valve, enabling intrinsic proprioceptive sensing via inductance measurement while maintaining pneumatic and mechanical reliability. The compact module, measuring 40 × 80 × 35 mm and weighing 55 g, features an adjustable height range of 29 to 71 mm under -25 to 25 kPa pressure. Experimental validation confirmed the notable performance of the module, with a consistent inflation time under various assembly combinations, a maximum extension force of 80.37 N, a low sensing hysteresis error of 7.58 %, and sufficient durability to withstand 5,000 activation cycles. The resulting module assembly enables closed-loop shape control in both individual and synchronized multi-module operations, eliminating the need for external sensor attachments and demonstrating a highly responsive, precise, and adaptive shape-morphing process. This compliant, controllable, and straightforward module can serve as an unprecedented foundation for shape-programmable embodied intelligence in emerging autonomous robotic systems.

Shape transformation by folding showcases intricate geometrical change over dimension, that has long provided the embodied intelligence of autonomous systems capable of adapting to challenging environments and displaying functional versatilities. Hinge-face material assembly interfaced by shape-morphing mechanisms produced the associated means. However, the fixed hinge structure limits the accessible modes of folding configurations despite the existing capability of rectification in plant. Here we introduce a programming strategy of the two-dimensional fold of a robotic sheet into an unbounded set of hinge configurations in the field post-deployment, which is referred to as being field-programmable, driven by a densely distributed electro-thermo-responsive system design. An interconnected set of metallic resistors, incorporated into the thermo-responsive polymer film, performs the dual functionalities of a heater and thermoreceptor, selectively recruited to configure folds through electronic modulation of its electrical power distribution. Electronic layouts, computational algorithms, and closed-loop control schemes present an intuitive means of blending user intent in situ, yielding a servoed, swift, and robust fold-programming process. The system is intrinsically driven by embeddable electronics to enable autonomous system engineering, as potentiated by multi-purpose applications in grasping and locomotion.

The demand for tactile sensors in robotics, virtual reality, and health care highlights the need for high performance and customizability. Despite advances in vision-based technologies, tactile sensing remains crucial for precise interaction and subtle pressure detection. In this work, we present a design and fabrication method of customizable tactile sensors based on thermoformed three-dimensional electronics. This approach enables ultrawide modulus tunability (10 pascals to 1 megapascal) and superior mechanical properties, including negligible hysteresis and high creep resistance. These features allow the sensor to detect a broad spectrum of pressures, from acoustic waves to body weight, with high performance. The proposed sensors have high sensitivity (up to 5884 per kilopascal), high linearity ( R ² = 0.999), low hysteresis (<0.5%), and fast response (0.1 milliseconds). We demonstrate applications in human-computer interaction and health care, showcasing their potential in various fields. This platform provides a scalable solution for fabricating versatile, high-performance tactile sensors.

Electrical impedance tomographic (EIT) tactile sensing holds great promise for whole-body coverage of contact-rich robotic systems, offering extensive flexibility in sensor geometry. However, low spatial resolution restricts its practical use, despite the existing deep-learning-based reconstruction methods. This study introduces EIT-GNN, a graph-structured data-driven EIT reconstruction framework that achieves super-resolution in large-area tactile perception on unbounded form factors of robots. EIT-GNN represents the arbitrary sensor shape into mesh connections, then employs a two-fold architecture of transformer encoder and graph convolutional neural network to best manage such the geometrical prior knowledge, resulting in the accurate, generalized, and parameter-efficient reconstruction procedure. As a proof-of-concept, we demonstrate its application using large-area face-shaped sensor hardware, which represents one of the most complex geometries in human/humanoid anatomy. An extensive set of experiments, including simulation study, ablation analysis, single-touch indentation test, and latent feature analysis, confirm its superiority over alternative models. The beneficial features of the approach are demonstrated through its application in active tactile-servo control of humanoid head motion, paving the new way for integrating tactile sensors with intricate designs into robotic systems.

Dexterous robotic grasping gains great benefits from tactile sensation, but delicate object exploration by tactile information is challenged by difficulty in rich and efficient data production. In this letter, we propose a tactile-based in-hand object perception approach, empowered by a sim-to-real approach toward a data-efficient learning process. A high-fidelity tactile input, measured by a pair of vision-based tactile sensors, was represented as a point cloud facilitating dual functionality of object classification and the associated pose estimation. For the classification, we constructed PoinTacNet , a variation of PointNet to fit into tactile data processing. A reliable simulation methodology on tactile input was employed for the pretraining of the model, transferred to the fine-tuning process facing real tactile data. Taking inspiration from human behaviors, a re-grasping strategy was imparted by means of conditional accumulation of class likelihood distribution. The result of the framework facilitates a high object classification accuracy of 83.89 $\%$ on the ten objects from McMaster-Carr's CAD models, which is significantly improved by the re-grasping. In addition, a set of benchmarks displays the computational efficiency in the sim-to-real transfer. In line with the successful classification, the posture of in-hand objects is estimated using point cloud registration algorithms, achieving an average angular and translational RMSE of 5.03 $^\circ$ and 2.41 mm, respectively. The proposed approach has the potential to enable robots to attain human-like haptic exploration skills for perceiving unstructured environments.

This study proposes a bidirectional double-spring pneumatic artificial muscle (PAM) with inductive self-sensing for wearable robotic devices. The actuator structure, composed of commercially available latex and springs, enables a low-cost and consistent manufacturing process. Springs, which serve as the skeleton of the actuator, were used to increase the stability and stiffness of the structure, as well as to embed a self-sensing ability using the inductance of the spring. The actuator design parameters were selected through simulations to prevent buckling while achieving a notable actuation performance. The bidirectional force characteristics of the actuator were evaluated through quasi-static experiments, and the inductance-based length-sensing performance was validated during both contraction and elongation. The need for self-sensing in bidirectional actuation was emphasized through closed-loop length control. The proposed actuator can potentially enable the development of lightweight and efficient wearable robotic devices with improved sensing and actuation capabilities.

Electrical resistance tomography (ERT) can be used to create large-scale soft tactile sensors that are flexible and robust. Good performance requires a fast and accurate mapping from the sensor's sequential voltage measurements to the distribution of force across its surface. However, particularly with multiple contacts, this task is challenging for both previously developed approaches: physics-based modeling and end-to-end data-driven learning. Some promising results were recently achieved using sim-to-real transfer learning, but estimating multiple contact locations and accurate contact forces remains difficult because simulations tend to be less accurate with a high number of contact locations and/or high force. This paper introduces a modular hybrid method that combines simulation data synthesized from an electromechanical finite element model with real measurements collected from a new ERT-based tactile sensor. We use about 290,000 simulated and 90,000 real measurements to train two deep neural networks: the first (Transfer-Net) captures the inevitable gap between simulation and reality, and the second (Recon-Net) reconstructs contact forces from voltage measurements. The number of contacts, contact locations, force magnitudes, and contact diameters are evaluated for a manually collected multi-contact dataset of 150 measurements. Our modular pipeline's results outperform predictions by both a physics-based model and end-to-end learning.

Electrical impedance tomography (EIT) based tactile sensor offers significant benefits on practical deployment because of its sparse electrode allocation, including durability, large-area scalability, and low fabrication cost, but the degradation of a tactile spatial resolution has remained challenging. This article describes a deep neural network based EIT reconstruction framework, the EIT neural network (EIT-NN), alleviating this tradeoff between tactile sensing performance and hardware simplicity. EIT-NN learns a computationally efficient, nonlinear reconstruction attribute, achieving high-resolution tactile sensation and well-generalized reconstruction capability to address arbitrary complex touch modalities. We train EIT-NN by presenting a sim-to-real dataset synthesis strategy for computationally efficient generalizability. Furthermore, we propose a spatial sensitivity aware mean-squared error loss function, which uses an intrinsic spatial sensitivity of the sensor to guarantee a well-posed EIT operation. We validate an outperformance of EIT-NN against conventional EIT sensing methods by conducting a simulation study, a single-touch indentation test, and a two-point discrimination test. The results show improved spatial resolution, sensitivity, and localization accuracy. The beneficial features of the generalized sensing of EIT-NN were demonstrated by examining touch modality discrimination performance.

A pneumatic artificial muscle (PAM) has been widely adopted in a wide range of wearable applications, but the bulkiness and the slow response of the actuator have remained a challenge. In this work, we developed a flat fabric PAM (ffPAM), achieving high compactness, rapid response, and low hysteresis error. Our ffPAM exerts a maximum force of 118.0 N ± 0.1 N at 172 kPa, an average maximum contraction ratio of 23.84% ± 0.06%, and a maximum hysteresis error of 6.2%. Using the thin fabric as a base material, the actuator's total weight is 34.8 g. Based on dynamic response experiments, the ffPAM was verified to have a response time of 0.0318 s ± 0.0012 s for contraction, and the bandwidth of more than 4 Hz, regarding the contraction length with respect to input pressure. Moreover, a maximum contraction length was maintained during 10 000 cycles. Additionally, we embedded a fabric-based soft contractile sensor for the measurement of the contraction. We validated that the capacitance of the embedded capacitive contractile sensor has a linear relationship with the contraction length at constant pressures. To emphasize the compactness of the ffPAM in practical wearable applications, the actuator was placed between the legs, and its compactness was compared with that of previous PAMs.

Data-driven approach has been recently investigated for identifying human joint angles by soft sensors because it is possible to circumvent of difficulty of the modeling. However, this approach commonly requires high computational burden from the voluminous amount of data and the time-series oriented network architecture. Moreover, the nature of the soft sensors make the problem worse due to the inherent material nonlinearity and hysteresis. In this study, we developed a novel sensing sleeve design to minimize the nonlinearity and hysteresis of the soft sensors and used a computationally efficient calibration method based on transfer learning. Through attaching the soft sensors to a link rather than directly to a joint, the effects of external disturbances during joint motions were minimized. To calibrate the sensor to the body motions, transfer learning was used with the results from the open-source musculoskeletal simulation software (OpenSim) and the sensor data. Tracking accuracy of ankle motions using the proposed method significantly improved measurement accuracy and reduced calibration time for five healthy subjects. The proposed method could be used to calibrate the soft sensors with a shorter training time with more than 1000 times faster computation speed and while maintaining tracking accuracy with a lower amount of data.