This paper presents a cable-driven actuation module design that can be easily integrated into soft wearable robots to enhance their functional versatility. The actuation module is easily mountable to various wearable systems through a simple locking mechanism and customized electronics. Furthermore, the symmetrical design of the actuation module allows for multiple orientation options, significantly expanding its range of applications. Through the use of finite element analysis (FEA), we reduce the weight of the actuation module by selecting lightweight materials, without compromising its reliability and durability. The modular and lightweight characteristics of this actuator make it suited for integration into various wearable usages. We applied the modular actuator to the soft wearable lower-limb device and showed the effectiveness through metabolic cost and electromyography(EMG) signal measurements.

Woosung Koh, Janghan Yoon, MinHyung Lee, Youngjin Song, Jaegwan Cho, Jaehyun Kang, Taehyeon Kim, Se-Young Yun, Youngjae Yu, Bongshin Lee. Proceedings of the 2025 Conference of the Nations of the Americas Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 1: Long Papers). 2025.

Self-Taught Reasoners (STaR), synonymously known as Rejection sampling Fine-Tuning (RFT), is an integral part of the training pipeline of self-improving reasoning Language Models (LMs). The self-improving mechanism often employs random observation (data) sampling. However, this results in trained observation imbalance; inefficiently over-training on solved examples while under-training on challenging ones. In response, we introduce Adaptive STaR (AdaSTaR), a novel algorithm that rectifies this by integrating two adaptive sampling principles: (1) Adaptive Sampling for Diversity: promoting balanced training across observations, and (2) Adaptive Sampling for Curriculum: dynamically adjusting data difficulty to match the model's evolving strength. Across six benchmarks, AdaSTaR achieves best test accuracy in all instances (6/6) and reduces training FLOPs by an average of 58.6\% against an extensive list of baselines. These improvements in performance and efficiency generalize to different pre-trained LMs and larger models, paving the way for more efficient and effective self-improving LMs.

Hip assistance with cable-driven devices has been proven to help decrease the metabolic cost of gait. However, most existing devices use heavy actuating modules or provide assistance in only one direction, limiting the effectiveness. Cable-driven devices are also difficult to accurately estimate the hip position using only motor encoders, therefore utilizing various auxiliary sensors. This paper introduces a 1.5 kg cable-driven soft wearable hip assist device that can provide assistance in both flexion and extension, using a velocity-dependent delayed output feedback controller (v-DOFC). The device is designed with the consideration of ergonomics and pressure distribution of wearable parts, to increase the anchoring performance and comfort. The controller uses time-delayed feedback proportional to the velocity output state, allowing control without requiring accurate position estimation. Additionally, directional weighting is used to provide different assistance forces for extension and flexion to match different optimal assistance values. Experimental results show that the device can reduce metabolic cost by 13.8 % compared to walking without the device. The soft wearable hip assist device can be applied to help the elderly with weaker muscles to walk longer distances.

To provide an alternative to conventional large-scale fitness equipment, we previously developed a soft passive wearable device for upper limb resistance exercises that utilized elastic exercise bands. However, the user was required to manually adjust the level of strength. In this paper, we introduce a novel wearable fitness device for upper limb exercise that constitutes cable-driven actuation to control the resistance profiles. Our proposed device allows for the generation of isotonic force trajectories, similar to those produced during dumbbell lifting, utilizing custom cable-driven actuators. The cable path of the device was determined based on the results of user testing aimed at optimizing muscle stimulation levels. In addition, linear resonant actuators were installed at customized haptic handles and upper arm modules to enhance proprioceptive sensitivity and exercise efficacy. The immersive “exer-tainment” user display interface was also redesigned to increase user motivation. The efficacy of our device was assessed by comparing the surface electromyography (sEMG) activities of upper limb muscles during chest press and triceps extension exercises performed with our device versus those using traditional weight training machines and rubber bands. It was found that our device could effectively strengthen during arm exercise, such as triceps extension.

An increase in health awareness has fueled the development of fitness equipment or devices nowadays. Most conventional fitness devices have had some issues in space limitation and the high cost of equipment. With the advance in wearable robotics, we proposed a soft passive fitness wearable device for upper limb resistance exercises such as chest press, frontal raise, and chest fly. Users can customize the exercise intensity by adjusting the length of the elastic bands embedded in the wearable device. Moreover, the exer-tainment (Exercise-entertainment) user display interface was designed to motivate the user. Movements of users were estimated using inertial measurement units (IMUs), and haptic feedback was provided through the vibro-stimulation. Furthermore, the effectiveness of the proposed device was evaluated with the Borg scales representing the rating of perceived exertion (RPE) and measuring the surface electromyography (sEMG) of the three muscles located one on the shoulder and two on the chest. Both the Borg 6–20 and CR 10 scales were increased, and the normalized sEMG activities of the upper limb muscles with the activated device had more than double in magnitude compared to that with a bare condition; therefore, the proposed device has a potential effectiveness as for resistance exercise. Overall, this research devotes preliminary evidence on the benefits of the device in promoting the user to work out and contributing to the exercise effects.

This paper presents an adjustable knee joint for a wearable robot for the elderly, to provide physical gait assistance. In order to compensate for the translational and rotational movements of the knee in the sagittal plane as well as aligning the frontal plane, the proposed knee joint is implemented with a rolling joint, two aligning passive joints, four-bar linkage, and a sliding mechanism. Because the sliding mechanism, four-bar linkage, and rolling joint exist between the actuator and the joint mechanism, the rotation angle of the actuator and the flexion angle of the rolling joint do not change equally due to the kinematic characteristics. In order to define the flexion angle change of the rolling joint with respect to the change to the actuator angle in the flexion/extension motion of knee, the design parameters and flexion model are proposed in this paper. The proposed artificial knee joint effectively delivers the torque required for assistance while adapting to the joint motion of the wearer. This paper describes the mechanical design of this knee mechanism and its implementation on a wearable robot and in preliminary experiments. The performance of the proposed mechanism was verified by simulations and experiments. Finally, preliminary experiments were performed to demonstrate the possibility of reducing the metabolic cost of using a knee assist.

This paper presents an adjustable knee mechanism for walking assistance devices for the elderly to provide physical gait assistance. The adjustable knee mechanism can assist in flexion/extension motions of the knee joint and compensate for the transitional movements of the knee in the sagittal plane as well as aligning the frontal plane. In order to compensate for the center of rotation, the proposed adjustable knee mechanism is implemented by connecting several rolling cams with unique and precisely calculated contact geometries. The key idea of the proposed mechanism is to realize linear motion and accurate torque transmission by a pulley method using the rolling joint in which the distance of the center between the rolling cams is changed. The proposed articulated joint can effectively deliver the torque required for assistance while adapting to the joint motion of the wearer. This paper describes the mechanical design of this knee mechanism and its implementation on a wearable robot and in preliminary experiments. The performance of the proposed mechanism was verified by simulations and experiments.

This paper proposes a new concept called “a flexible exoskeleton” and presents the design of a wearable walking assistance device to provide physical gait assistance for the elderly. To overcome some limitations of previous wearable walking assistance devices, 3 novel mechanisms are proposed: 1) flexible thigh frames that withstand the designated load direction, 2) an adjustable flexible hip brace surrounding the wearer's body, and 3) slim and backdrivable actuators that minimize the maximum height and resistance. The proposed walking assistance device has reduced metabolic energy consumption by 13.2±4% compared with a free condition.

Crowdfunding, which helps getting financial sources for niche market entrepreneurs, lowers the entry barrier to start a social and for-profit project. However, it is widely believed that overwhelmingly successful projects, called "blockbuster" projects, would have a significant impact on the overall crowdfunding projects. In theoretical perspectives, there have been two countervailing effects of blockbuster, cannibalization effects and spill-over effects. Therefore, this motivates our research to investigate the competitions among projects within same categories. We expect to fill in the gaps of prior literatures which concluded that the blockbuster project has the same effects on all the other projects within same categories. As a major contribution of this paper, we obtain market structure insights from the creator generated contents based on a text-mining approach and combine it with semantic network analysis to assess the degree to which the latent competition causes cannibalize or spill-over effects inside the same category. Our study uses unique data set, compiled by Python based self-developed crawler. The dataset includes 148,398 projects and 2,915,821 records of daily pledged amount. Our analysis of the concurrent impact of blockbuster project shows the much larger spillover effects to the inside cluster relative to the outside clusters. However, this spillover effect to the outside cluster significantly increase as the magnitude to the inside cluster in the long run perspective. The results of this paper suggest practical marketing implications to entrepreneurs and platform.

We propose a control framework for dynamic bipedal locomotion with compliant joints. A novel 3D dynamic walking is achieved by utilizing natural dynamics of the system. It is done by 1) driving robot joints directly with the posture-based state machine and 2) controlling tendon-driven compliant actuators. To enlarge gait's basin attraction for stable walking, we also adaptively plan step-to-step motion and compensate stance/swing motion. Final joint input is described by a superposition of state machine control torques and compensation torques of balancers. Various walking styles are easily generated by composing straight and turning gait-primitives and such walking is effectively able to adapt on various environments. Our proposed method is applied to a torque controlled robot platform, Roboray. Experimental results show that gaits are able to traverse inclined and rough terrains with bounded variations, and the result gaits are human-like comparing the conventional knee bent walkers.

This paper presents a framework to generate dynamic walking for biped robots. A set of self-stable gait primitives is first constructed. It is done by 1) representing parametric gait primitives, 2) utilizing state-dependent torque control, and 3) doing numerical optimization that takes into account the complex multi-body dynamics with frictional contact forces. Dynamic walking to follow the arbitrary path including a curve is then generated online via sequentially composing primitive motions. Results show that dynamic gaits are humanlike and efficient compared to the conventional knee bent walkers. Our proposed method is applied to a torque-controlled, human-sized biped robot platform, `Roboray' which is cable-driven partially for joint compliance. Following a discussion on robot design and control, experimental results are also reported.

We propose a balancing control framework for a torque-controlled biped robot, Roboray. Roboray has two 6 DOF legs and torque sensors are integrated at all the leg joints. It has a new cable-driven joint module as a pitch joint drive, which is highly back-drivable and elastic. Using these hardware characteristics, we propose a new balancing control algorithm. This algorithm is the combination of gravity compensation, virtual gravity control and damping control. A friction compensation technique is also introduced in order to eliminate the nonlinearity of damping and to improve the performance of torque tracking. Our proposed method is applied to a simple inverted pendulum system and Roboray. Experimental results show that these two system keep their balance when they are pushed slightly.