Multi-agent systems (MAS) are ubiquitous in natural and artificial systems, they range from flocks of birds and fish school to swarm intelligence and the coordination of autonomous mobile robots. A basic research issue is to understand how locally interacting agents lead to collective behaviors of the overall systems. In this talk, we will present some recent results on the analysis of a basic class of MAS, where each agent moves in the plane with a constant speed and with its heading takes the average of those of its neighbor agents. We will also discuss how to use the idea of soft control to intervene the collective behaviors of such MAS without changing the existing local rule of interactions. Moreover, we will explain some key issues together with the mathematical tools in the analysis of the connectivity and synchronization of the dynamical networks of the underlining nonlinear dynamical systems.