Aiming at the problem of underwater cleaning robot positioning in deep-sea aquaculture cages; this paper proposes a multisensory navigation system. Considering the characteristic that the cage is a structured working environment; this paper proposes to identify targets attached to side-net or center column through visual algorithms; extract 5 coplanar corner points; and use orthogonal iterative algorithm to estimate pose of robot; which is used to correct the output of inertial navigation system periodically. The simulation results show that the proposed method is real-time and relatively accurate; which meets the requirements of the underwater cleaning robot cleaning operation.
With the development of science and technology; people pay more and more attention to the exploration of marine resources. Aquaculture is one of the fastest growing food industries in the world; whose production exceeds 50% of the world fish consumption. At present; aquaculture mostly adopts the form of deep-sea cages. In order to improve the efficiency; regular cleaning and inspection of the cages are essential. Without cleaning; the pollutants on the net hole not only affect the water exchange; but also pollute the water. The underwater cleaning robot can clean and inspect fish-cages at a set depth repeatedly; thus improving cleaning efficiency and decreasing the effect of environment (Livanos; Zervakis; Chalkiadakis; Moirogiorgou and Papandroulakis; 2018). Therefore; how to locate and navigate the underwater cleaning robot in the aquaculture cage is one of the key issues.
In underwater environment; the most commonly used navigation method is inertial navigation system (INS); providing information for the control of robots (Xu; Tang; Han and Xie; 2019). INS estimate the pose of robot by integration of accelerations and angular velocities; which is measured by inertial measurement unit (IMU) (Wu; Ta and Xiao; 2019). Because it does not rely on any external information during the update process; INS gets more and more attention; especially the strapdown inertial navigation system (SINS); which has the advantages of low price and small size. Due to the error of IMU; the measurement of INS presents an error which increases gradually overtime; making it necessary to be corrected by other navigation system; such as GPS and sonar (Telle and De Barros; 2013). GPS is real-time and accuracy. However; it can not be used in deep sea because its signal doesn't propagate in the water. Sonar is widely used among ROV; while it is limited by its high cost and vulnerable signal due to the high density of fish (Rundtop and Frank; 2016).
