The measurement and generation of mechanical deformations is a key functionality in health monitoring, human-machine interaction, and soft robotics. However, current methods typically rely on small and hard transducers, which result in poor performance, cumbersome implementation, and incompatibility with the human body. An alternative approach consists of imparting functionality within long and soft fibers. However, the multi-material assemblies, which are required for advanced functionalities, are challenging to realize in fibers. Additionally, the 1-dimensional geometry of fibers, which stands in stark contrast to traditional 0-dimensional devices, requires novel schemes to extract meaningful information and induce targeted movements. In this Thesis, four distinct innovations to fiber-based transducers are made, which are composed of soft materials and fabricated by the thermal drawing technique. The first three works are each aimed at one of the three conceptual elements of a sensor: the structural design, the constituting materials, and the signal processing. In the fourth work, actuation is coupled with sensing to develop robotic fibers. (i) Considering the design element of a sensor first, fibers with an asymmetric elastomer architecture are developed, featuring several electrodes separated by an air gap. Compressive loads on the meter-long fibers result in the selective contacting of electrodes within the deformed fiber structure, triggering pronounced electrical signals at distinct pressure levels. This developed sensing concept enables the facile functionalization of large surfaces, demonstrated by a fiber-augmented gymnastic mat for the monitoring of human body position, posture, and motion. (ii) Focusing on the materials next, a novel polymer nanocomposite with carefully engineered rheological, mechanical, and electrical properties is introduced. In particular, controlled changes to the nanoparticle network and the intertwined electrical conduction of the material are explored - induced either by viscous flow during thermal drawing fiber fabrication or elastic deformation in applications. The nanocomposite represents a platform upon which a family of fiber devices can be developed. (iii) As the third pillar of sensing, signal processing is addressed by employing electrical reflectometry within microstructured elastomeric fibers that integrate tens of liquid metal conductors. Reflected electrical waves, triggered by multimodal deformations of the fiber structure, are investigated experimentally and theoretically. This unprecedented physical mechanism enables multiplexed measurements of the mode, magnitude, and position of multiple simultaneous pressing and stretching events. By integrating a single fiber with a single interface port into a larger fabric, the technique can be used to create an electronic textile that can decipher convoluted mechanical stimulation. (iv) Finally, both actuation and sensing schemes are combined, resulting in robotic fibers with closed-loop control. Thanks to three integrated tendons pulling on the distal fiber end, bending with two degrees of freedom is achieved. Envisioned as steerable catheters, the fibers can autonomously avoid obstacles to minimize tissue damage, scan the environment to execute the best navigation path, and accurately deliver mechanical tools, fluids, as well as optical and electrical stimulation.