In the paper, the high-order adaptive algorithm to controlling the multi-agent system is proposed. Control of the pitch/yaw angles of quadrotors is taken as an example. It is shown that realistic noisy measurement errors of sensors do not affect drastically the overall system performance with the proposed algorithm. The real-word 2DOF laboratory quadrotor testbed was used for demonstrating the trajectories convergence. Consensus achievement in the multi-agent system consisting of four quadrotor testbeds in the noisy sensor case is evaluated by the simulations. It is shown that the proposed technique allows using decentralized cooperative control for the examined multi-agent system. Realistic gyroscope properties were taken into account while studying the noisy measurement case. The series of numerical simulations were performed to validate of proposed multi-agent system performance for the various algorithm parameters.