This paper presents a novel method for reconstructing curves relying on tangential data which are provided by embedded sensors. The reconstruction process is based on the knowledge of the distribution of the sensors along the curve, represented by a ribbon, and on the associated tangential orientation measurements without any information about their positioning in space, so that this problem is not an envelope problem. We first show how we can obtain these data from sensors and the prototypes we have created. Then we provide methods for planar curves, then for spatial curves and we analyze results with physical sense and convergence in order to validate these methods. Finally, we show some results from both simulated data and real data.