In order to search and rescue injured during earthquake, we proposed a method for multi-robots motion planning and distributed control in this paper. At first, we have created two probabilistic search models to considering the search area and the characteristics of sensors, which we used to search the injured targets. And after finding the targets, they are assigned to the mobile robots on the land to afford emergency rescue. In order to reach to the targets, a path planning method based on map matching is proposed. There are three parts here. Firstly, to obtain the global and local map: continuous ground images are first collected using the UAV’s vision system, and subsequently, a global map of the ground environment is created by processing the collected images. The local map of the ground environment is obtained using the 2D laser radar sensor of the leader (UGA). Established the coordinate conversion relationship between UAV and UGV, unknown values during map matching are determined via the least square method. Secondly, our robots moved by group (leader-follower). The leader’s path was planned globally and locally. The other multi-robots moved along the path planned by the leader. Thirdly, in order to plan and coordinate the robots in the group, the finite state machine is used to describe the logical level of control system for each robot in the group. After that, at the tactical level of the control system, the movement control law of formation maintaining mode and formation switching mode is designed.