[en] For a particle of mass μ moves on a 2D surface $f\left(\mathbf{x}\right)=0$ embedded in 3D Euclidean space of coordinates x, there is an open and controversial problem whether the Dirac's canonical quantization scheme for the constrained motion allows for the geometric potential that has been experimentally confirmed. We note that the Dirac's scheme hypothesizes that the symmetries indicated by classical brackets among positions x and momenta p and Hamiltonian H_c remain in quantum mechanics, i.e., the following Dirac brackets ${[\mathbf{x},H_c]}_D$ and ${[\mathbf{p},H_c]}_D$ holds true after quantization, in addition to the fundamental ones ${[\mathbf{x},\mathbf{x}]}_D$, ${[\mathbf{x},\mathbf{p}]}_D$ and ${[\mathbf{p},\mathbf{p}]}_D$. This set of hypotheses implies that the Hamiltonian operator is simultaneously determined during the quantization. The quantum mechanical relations corresponding to the classical mechanical ones $\mathbf{p}/\mu ={[\mathbf{x},H_c]}_D$ directly give the geometric momenta. The time t derivative of the momenta $\dot{\mathbf{p}}={[\mathbf{p},H_c]}_D$ in classical mechanics is in fact the generalized centripetal force law for particle on the 2D surface, which in quantum mechanics permits both the geometric momenta and the geometric potential.