This paper presents experimental results of formation control problems of omnidirectional mobile robots using distributed nonlinear model predictive control (NMPC). Two main objectives are (i) to maintain a desired flexible formation pattern and (ii) to follow a reference path. Both pose errors and formation errors are included into a local objective function, which is minimized at each update time. In the formation control, the curvilinear abscissa s has been used as a coupling term with neighboring robots. The strategy in such a way that the exchange of the most recent optimal state trajectory between coupled subsystems has been employed. The distinct features of NMPC are that constraints can be explicitly accommodated, as well as nonlinear and time-varying systems can be easily handled. Experiments with three omnidirectional mobile robots are presented to illustrate the validity of our proposed method.