The paper considers a forward-backward system of parabolic PDEs arising in a Mean Field Game (MFG) model where every agent controls the drift of a trajectory subject to Brownian diffusion, trying to escape a given bounded domain $\Omega$ in minimal expected time. The important point is that agents are constrained by a bound on the drift depending on the density of the other agents at their location (the higher the density, the smaller the velocity). Existence for a finite time horizon $T$ is proven via a fixed point argument but, because of the diffusion, the model should be studied in infinite horizon as the total mass inside the domain decreases in time, but never reaches zero in finite time. Hence, estimates are needed to pass the solution to the limit as $T\to\infty$, and the asymptotic behavior of the solution which is obtained in this way is also studied. This passes through a combination of classical parabolic arguments together with specific computations for MFGs. Both the Fokker--Planck equation ruling the evolution of the density of agents and the Hamilton--Jacobi--Bellman equation on the value function display Dirichlet boundary conditions as a consequence of the fact that agents stop as soon as they reach $\partial\Omega$. The initial datum for the density is given (and its regularity is discussed so as to have the sharpest result), and the long-time limit of the value function is characterized as the solution of a stationary problem.