Most image super-resolution (SR) methods are developed on synthetic low-resolution (LR) and high-resolution (HR) image pairs that are constructed by a predetermined operation, e.g., bicubic downsampling. As existing methods typically learn an inverse mapping of the specific function, they produce blurry results when applied to real-world images whose exact formulation is different and unknown. Therefore, several methods attempt to synthesize much more diverse LR samples or learn a realistic downsampling model. However, due to restrictive assumptions on the downsampling process, they are still biased and less generalizable. This study proposes a novel method to simulate an unknown downsampling process without imposing restrictive prior knowledge. We propose a generalizable low-frequency loss (LFL) in the adversarial training framework to imitate the distribution of target LR images without using any paired examples. Furthermore, we design an adaptive data loss (ADL) for the downsampler, which can be adaptively learned and updated from the data during the training loops. Extensive experiments validate that our downsampling model can facilitate existing SR methods to perform more accurate reconstructions on various synthetic and real-world examples than the conventional approaches.

Designing softness into robots holds great potential for augmenting robotic compliance in dynamic, unstructured environments. However, despite the body's softness, existing models mostly carry inherent hardness in their driving parts, such as pressure-regulating components and rigid circuit boards. This compliance gap can frequently interfere with the robot motion and makes soft robotic design dependent on rigid assembly of each robot component. We present a skin-like electronic system that enables a class of wirelessly activated fully soft robots whose driving part can be softly, compactly, and reversibly assembled. The proposed system consists of two-part electronic skins (e-skins) that are designed to perform wireless communication of the robot control signal, namely, "wireless inter-skin communication," for untethered, reversible assembly of driving capability. The physical design of each e-skin features minimized inherent hardness in terms of thickness (<1 millimeter), weight (~0.8 gram), and fragmented circuit configuration. The developed e-skin pair can be softly integrated into separate soft body frames (robot and human), wirelessly interact with each other, and then activate and control the robot. The e-skin-integrated robotic design is highly compact and shows that the embedded e-skin can equally share the fine soft motions of the robot frame. Our results also highlight the effectiveness of the wireless inter-skin communication in providing universality for robotic actuation based on reversible assembly.