A real-time three-dimensional (3D) object sensing and reconstruction scheme is presented that can be applied on any arbitrary corporeal shape. Operation is demonstrated on several calibrated objects. The system uses curvature sensors based upon in-line fiber Bragg gratings encapsulated in a low-temperature curing synthetic silicone. New methods to quantitatively evaluate the performance of a 3D object-sensing scheme are developed and appraised. It is shown that the sensing scheme yields a volumetric error of 1% to 9%, depending on the object. © 2012 Optical Society of America OCIS codes: 060.2370, 060.3735, 120.4820, 110.6880, 280.4788. One of the main goals driving forward the research and development of fiber-optic sensing systems is their application in the field of structural health monitoring [1,2] and shape sensing [3,4]. There are several applications for shape-sensing, such as spatial awareness and control of robots in hazardous conditions (e.g., deep-water semisubmersible drilling platform and subsea production systems) [5,6] or profiling minimally invasive (keyhole) surgery [7], to name a few.There are a large number of shape sensing systems that have fiber Bragg gratings (FBG) as the sensing elements that monitor the strain experienced by the sensors that are adhered to the object and use a shape determination algorithm based upon a strain mapping technique [1,3]. Other sensing systems include fiber-optic loop sensors based on bend-induced loss; this technique is used in the commercially available system called ShapeTape [2] with others based upon distributed sensing, such as intrinsic Rayleigh backscattering employing optical frequency domain reflectometry [8]. Alternative approaches to shape sensing include employing camera systems using complex, shape-sensing recognition algorithms [9].The three-dimensional (3D) shape-sensing system demonstrated in this Letter is based upon fiber bidirectional sensing elements that have the ability to distinguish both positive and negative curvature variation upon a twodimensional (2D) plane. The element consists of two FBGs that are temperature self-compensating due to the fact that the difference in wavelength is measured and calibrated to curvature (concave and convex) (see Fig. 1). The curvature information deduced from all sensing nodes within the scheme is processed by bespoke algorithms to construct a 3D representation of the object in question. As we will show, our method offers advantages in comparison to previously existing approaches, the most important of which is the real-time 3D shape reconstruction through tailor-made algorithms. The volumetric error of the sensing system was found for shapes with known volumes (i.e., arcs, ellipsoidal cylinders, and spheres) and ranged from 1% to 9% depending upon the shape being used. In addition, we present an efficient and, to the extend of our knowledge, entirely new approach to the evaluation of the performance of a 3D object shape sensing system, obtaining a figure-of-merit that should be a useful tool for...