Position, Velocity and Orientation Estimation of Minesto’s Crossflow Underwater Kite

Abstract

Crossflow underwater kites have shown promising potential to generate green energy, with Minesto’s Deep Green on the front line. The power output is optimized when the kite is following a figure-eight trajectory, where the motion is controlled by a rudder on the kite. This work focuses on providing the control system with improved input about position, velocity and orientation of the kite, with the use of inertial sensors and knowledge of depth, in order to steer the rudder optimally. The sensor signals were processed and filtered in order to handle problems of sensor and process noise. An algorithm was designed that combined the different sensors to predict the pose and motion of the kite. This was done by first approximate the noise of the sensors, which were used as input, into an extended Kalman filter for orientation estimation, together with inertial measurements from a gyroscope and an accelerometer. After an initial guess on position, based mainly on depth of kite, a steady-state Kalman filter was applied in order to improve the position estimate and also obtain velocity. The result show that the sensor fusion performed has potential in predicting the movement of the kite. However, the limited access to data prevents us from drawing too big conclusions. Even if there are some challenges regarding bias drift and robustness of the algorithm, it can be shown that the proposed algorithm produces realistic output when it comes to physical constraints due to the tether length but also in terms of periodicity of the orientation.