Three-dimensional shape sensing plays a critical role in applications ranging from soft robotics and interventional medical devices to emerging artificial perception systems. Optical fiber sensors are particularly attractive for these applications because of their small size, mechanical flexibility, and resistance to electromagnetic interference.
Most existing high-performance shape-sensing technologies rely on multicore fibers or multi-channel sensing architectures. While these approaches can achieve accurate shape reconstruction, they often require larger sensor footprints and more complex interrogation systems, limiting their use in highly constrained environments such as micro-robots and minimally invasive medical instruments.
Standard single-mode fibers, by contrast, offer an ideal platform for miniaturized sensing because of their compact size, mature manufacturing infrastructure, and excellent mechanical properties. Yet their single optical transmission channel has traditionally made it difficult to extract deformation information from multiple directions simultaneously. As a result, enabling high-resolution 3D shape reconstruction using a single standard optical fiber has remained a longstanding challenge.
Now, researchers from Northwest University have developed a new approach that enables a standard single-mode optical fiber to perceive its shape in three dimensions. By creating an array of misplaced orthogonal eccentric scatterers inside the fiber using femtosecond laser direct writing, the team introduced a fundamentally new deformation-sensing mechanism that encodes 3D shape information directly into the fiber’s scattering structure.
The work, entitled “Single-fiber three-dimensional shape sensing via femtosecond laser inscribed orthogonal eccentric scatterers,” was published in Light: Science & Applications . The study was led by Prof. Ruohui Wang and Associate Prof. Fengyi Chen, with doctoral student Pengtao Luo serving as the first author. The research also benefited from the guidance of Prof. Xueguang Qiao and Prof. Tuan Guo from Jinan University.
To address this challenge, the researchers developed a new three-dimensional deformation encoding mechanism based on Misplaced Orthogonal Eccentric Scatterers (MOESs). Using femtosecond laser direct writing, they precisely inscribed eccentric scattering structures within the cladding of a standard single-mode fiber. These scatterers were arranged in an orthogonally displaced pattern that functions as a spatial encoding architecture.
When the fiber bends, local deformation in different directions produces distinct changes in scattering intensity. By decoding these distributed scattering signatures, the system can retrieve directional bending information and reconstruct complex three-dimensional shapes.
Unlike conventional approaches that rely on multiple cores or multiple sensing channels, the MOES-based strategy embeds directional deformation information directly into the fiber structure itself. This allows a standard single-mode fiber to perform high-resolution 3D shape sensing while preserving its compact size and structural simplicity.
Rather than relying on multiple sensing channels, the researchers took a fundamentally different approach. Using femtosecond laser direct writing, they inscribed an array of microscopic scattering structures inside the cladding of a standard single-mode fiber. These scatterers were not placed randomly. Instead, they were arranged in a carefully designed orthogonal and spatially displaced pattern that allowed each scattering site to respond preferentially to bending from a specific direction.
Together, the scatterers form what the researchers describe as a spatiotemporal encoding architecture.That is to say, the fiber contains an embedded coordinate framework that converts three-dimensional deformation information into distributed scattering signals that can be read through a single optical channel.
The sensing mechanism is based on curvature-dependent modulation of Rayleigh scattering intensity. When the fiber bends, the optical mode inside the core shifts slightly, changing its overlap with nearby eccentric scatterers. As a result, the scattering intensity generated by individual structures varies according to the magnitude and direction of local bending. By tracking these intensity changes, the system can simultaneously retrieve curvature and orientation information along the fiber.
The orthogonally displaced arrangement of the scatterers provides the key to reconstructing complex three-dimensional geometries. Using the encoded scattering signatures, the researchers successfully reconstructed a range of representative shapes, including circular arcs, S-shaped curves, and crossed spatial configurations. The demonstrations achieved millimeter-scale spatial resolution while maintaining the simplicity and compact size of a standard single-mode optical fiber.
The MOES platform takes a different route. Instead of encoding information in wavelength, it encodes information in space. The position of each scatterer becomes part of the sensing code, meaning that sensor density is no longer fundamentally constrained by optical bandwidth. Instead, it is primarily determined by propagation loss and the spatial resolution of the interrogation system. This transition from wavelength encoding to spatial encoding could open the door to ultra-dense, highly miniaturized fiber sensing networks.
According to the researchers, the work represents the first demonstration of a three-dimensional deformation encoding mechanism based on misplaced orthogonal eccentric scatterers. By enabling a standard single-mode fiber to perceive and reconstruct its own three-dimensional shape, the technology overcomes longstanding trade-offs among sensor size, spatial resolution, and system complexity. Looking ahead, the team envisions applications in soft robotics, catheter navigation, artificial perception systems, and other intelligent platforms that require accurate spatial awareness in confined environments. More broadly, the concept may provide a foundation for future generations of highly integrated sensing systems capable of endowing machines with a form of proprioception similar to that found in biological organisms.
Light: Science & Applications
Single-fiber three-dimensional shape sensing via femtosecond laser inscribed orthogonal eccentric scatterers