In this note we propose a controller that solves the problem of coordination of two (or more) robots, under a master-slave scheme, in the case when only position measurements are available. The controller consists of a feedback control law, and two non-linear observers. It is shown that the controller yields ultimate uniformly boundedness of the closed loop errors, a relation between this bound and the gains on the controller is established. Simulation results on two two-link robot systems show the predicted convergence performance.