For traditional leader-follower multi-agent control systems, a three-tier leader-follower topology structure is proposed. It is assumed that agents on the same level can obtain information about their neighbors, and agents directly connected across tiers can acquire state information through the restraint matrix. Based on this assumption, a hierarchical design of the consensus controller is conducted. Herein, the second-tier leaders are able to estimate the velocity information of the first-tier leaders through positional data, and the position and velocity information of the second-tier leaders can be fully acquired by the third-tier followers. By employing the Lyapunov theory, the consistency of each tier's agents is assessed and the range of parameters is determined. Lastly, the correctness and feasibility of the results are verified through simulation experiments.