We consider the observation problem for a particular class of bidimensional systems with scalar output which requires the construction of an embedding in higher dimension. We propose a new approach that does not require any coordinates transformation. This approach is based on the design of parallel estimators in the same dimension than the original system. Each estimator uses the knowledge of the first two derivatives of the output, and the further derivatives are used to discriminate at any time among the different estimators. We give three examples showing the applicability of this approach with measurement noise. Biological systems used in batch bioprocess models are of particular motivation for this work.