This paper deals with the development of an extended Kalman filter(EKF) algorithm for the localization of underwater mining vehicles with sensor time delay. To develop the EKF algorithm, we use a kinematic model including the inner/outer track slips and the slip angle for the vehicle. For the experimental verification, a scale down tracked vehicle is run in air on a soil bin containing cohesive soil of Bentonitewater mixture. The measurements include the inner and outer wheel speeds from encoders, the heading angle from a compass sensor and a fiber optic rate gyro, and x and y coordinate position values from a vision system. The vision sensor replaces the LBL(Long Base Line) sonar system used in the real underwater positioning situations. Artificial noise signals mimicking the real LBL noise signal are added to the vision sensor information. Also, to simulate the LBL sensor time delay, the vision output data are delayed. Experimental results show the effectiveness of the EKF algorithm in rejecting the sensor measurements noise and sensor time delay effect.


Deep underwater mining vehicles operate on extremely soft cohesive soil to excavate valuable metals such as maganise and send those metals to surface ships through piping systems. Therefore, the vehicles can experience large slips on its tracks. Also, the underwater piping systems attached on the vehicle induce large disturbance forces and additional slip to the vehicle. The vehicle should have the capability to follow a desired mining trajectory. For this purpose, it is very important to know the current position of the vehicle. Sonar systems such as LBL(Long Base Line) can be used to estimate the position. However, the output signals of most sonar systems usually have measurement errors which can range from 0.1 m to several meters, and long time delay as large as several seconds.

This content is only available via PDF.
You can access this article if you purchase or spend a download.