Visual-inertial-actuator odometry for multirotor UAVs with rotor drag and external disturbance

Accurate knowledge of an unmanned aerial vehicle’s (UAV) state and external forces is crucial for motion control. One method to obtain this information is Visual-Inertial-Actuator Odometry (VIAO) which combines commonly available vision and inertial sensors with actuation data which consists of a model of the vehicle dynamics and input measurements. In this paper, we propose VIAO system based on the existing VIMO (Visual Inertial Model-Based Odometry) method. Our approach includes a dynamical model for a multirotor UAV with rotor drag. In addition, a disturbance observer is added for constant force estimation. The observer is integrated to the optimization-based method by adding a residual term to the cost functional. The proposed system can differentiate between the constant or slowly time-varying component of the external force and accelerometer bias. We evaluate the performance using a benchmark dataset and show an average of \(21\%\) $21\%$ improvement in position estimation accuracy.