Modeling and control of underwater pan/tilt camera tracking system.

摘要

It is always a challenging problem to clearly observe in situ the daily activities of a moving creature in a darkness underwater environment. A master-slave coordination system with mobile illuminator and camera was proposed and studied to improve the clearance of video image. A dynamic model of Pan/Tilt platform and a coordinated geometry model which represents the space geometrical relationship between Master and Slave were firstly developed and verified. Hydrodynamic forces were modeled using computational fluid dynamic (CFD) through directly incorporating the hydrodynamic coefficients into the effects of forces. The experiment based-on selection of turbulence models among k-epsilon, k-o, shear stress transport (SST), Reynolds stress model (RSM) and large eddy simulations (LES), was accomplished comparing the experimental data to the results of numerical simulation. To control underwater platforms with a manipulator is challengeable due to the complex external forces in an underwater environment Linear proportional-derivative (LPD) and nonlinear PD (NPD), optimal control and variable structure control (VSC) were employed to the tracking control of this underwater observation system, emphasizing on sliding mode control (SMC). The main contributions can be concluded as followings: 1) A hybrid ROV/AUV hybrid scheme was successfully developed, camera platform (ROV), light platform (AUV). Based on the simulation results, this hybrid architecture can comprehensively utilize the merits of ROV and AUV. This system becomes more agile, implementable for the observation of a coast area in Lake Michigan. 2) A Master-Slave coordinate architecture was developed to solve the underwater illumination problem in order to overcome the back-scattering of water/contaminant particles. The performance of illumination can be improved by a mobile light source. This slave (light platform) can precisely follow a desirable moving subject under the commandments from master (camera platform). 3) A highly-accurate hydrodynamic model has been formulated through numerical modeling instead of employing the traditional methods. From the visual simulation of flow fields, numerical modeling can predict the hydrodynamic forces very well. 4) A simple, implementable adjusting method for boundary layer in SMC was developed and implemented. From the simulation results, variable BL can decrease the tracking error, and smooth the control inputs.