Hierarchical estimation of uncertainties in multi-rotors:A yaw-motion study using test bench

• Full estimation in multi-rotors: parameters, disturbances, and actuator faults. • Novel network of global, robust observers able to deal with nonlinearity and high number of uncertainties. • Proven theoretical results with observability analysis. • Experiments using real test bench in quantitative comparison with two other methods. We design observers to estimate yaw uncertainties including unknown parameters, input disturbances, and actuator faults affecting a multi-rotor. The considered experimental test bench, subject to unmeasured external disturbances, consists of a drone body actuated by two independently controlled motors-propellers, each subject to disturbances and potential actuator faults. Following an observability analysis, we design a nonlinear observer to estimate the constant drag coefficient, disturbance, and angular velocity of each actuator from the measured motor current. We derive sufficient conditions and construct a Kalman-like observer to estimate the loss-of-effectiveness coefficients and the disturbances acting on the yaw motion of the drone body, with the yaw rate as the measurement and the estimates from the actuators’ observers as inputs. Realistic simulations and experiments on the test bench and comparisons with existing estimation methods illustrate the effectiveness and practical feasibility of our approach.