This article introduces a general description of a multi-robot system (MRS) architecture aimed to provide guidance for a group of humans. At a first approach, only the strategy and architecture framework are described, further than the ethological and human factors involved in human-robot interaction. The paper encompasses a special strategy to conduct and crowd multiple people. Such strategy includes a methodology to localize multi-human; a MRS architecture design; and a control for people trajectory and a robots motion planner. A key-problem in the implementation of the system is that, there is no signal of any type for accomplishing guidance. In addition, some experimental and simulation results are presented, which exhibits the effectiveness of the proposed architecture, and the social force model adapted in a manner to simulate behavior of groups of people.