In this paper, we apply a recently proposed distributed estimation algorithm to a simple multi-agent flocking model with collision avoidance. 3D MATLAB simulations are performed with randomly generated agents' initial positions and velocities. The collision avoidance is realized through a repelling force between agents moderated by an alignment measure. The distributed estimation algorithm estimates certain global feature indicator of the multi-agent systems through consensus of agents' local estimations. The agents exchange information about their local estimations at discrete time instants, which are controlled by an event-triggered mechanism. The algorithm's effectiveness is illustrated through the example of centroid estimation, and simulations show that all agents' local estimations converge to the true position of the group centroid.