Improved path following for autonomous marine vehicles with low-cost heading/course sensors: comparative experiments

Abstract

By overcoming constraints of low-cost sensors equipped on the autonomous marine vehicle (AMV), the classic path following approaches based on heading autopilot and course autopilot are improved to enhance the following accuracy of AMV, respectively. For the heading autopilot, adaptive line-of-sight (ALOS) is proposed to compensate for the generalized sideslip including the actual sideslip angle and the magnetic compass bias. For the course autopilot, extended Kalman filter (EKF) is employed to fusion the position and yaw angular rate to obtain the more precise and stable course, so as to substitute the unreliable course feedback from the satellite navigation receiver in cases of low speed and non ideal positioning. The performance and characteristics of both improved schemes are summarized through comparative experiments conducted on a low-cost AMV. Compared with the classic path following method, the straight line path following accuracy of the heading autopilot with generalized sideslip compensation and the course autopilot with EKF feedback are improved by 78.3% and 85.8% respectively. The circular path following accuracy are improved by 61.8% and 66.7% respectively. Engineers can choose the path following methods flexibly based on the hardware configuration, working environment and path type of the AMV to improve the following accuracy.