Modeling, Design and Control of Gliding Robotic Fish

MODELING, DESIGN AND CONTROL OF GLIDING ROBOTIC FISH By Feitian Zhang Autonomous underwater robots have been studied by researchers for the past half century. In particular, for the past two decades, due to the increasing demand for environmental sustainability, significant attention has been paid to aquatic environmental monitoring using autonomous underwater robots. In this dissertation, a new type of underwater robots, gliding robotic fish, is proposed for mobile sensing in versatile aquatic environments. Such a robot combines buoyancy-driven gliding and fin-actuated swimming, inspired by underwater gliders and robotic fish, to realize both energy-efficient locomotion and high maneuverability. Two prototypes, a preliminary miniature underwater glider and a fully functioning gliding robotic fish, are presented. The actuation system and the sensing system are introduced. Dynamic model of a gliding robotic fish is derived by integrating the dynamics of miniature underwater glider and the influence of an actively-controlled tail. Hydrodynamic model is established where hydrodynamic forces and moments are dependent on the angle of attack and the sideslip angle. Using the technique of computational fluid dynamics (CFD) water-tunnel simulation is carried out for evaluating the hydrodynamic coefficients. Scaling analysis is provided to shed light on the dimension design. Two operational modes of gliding robotic fish, steady gliding in the sagittal plane and tailenabled spiraling in the three-dimensional space, are discussed. Steady-state equations for both motions are derived and solved numerically. In particular, for spiral motion, recursive Newton’s method is adopted and the region of convergence for this method is numerically examined. The local asymptotic stability of the computed equilibria is established through checking the Jacobian matrix, and the basins of attraction are further numerically explored. Simulation and experiments are conducted to validate steady-state models and calculated equilibria for both motions. Tail-enabled feedback control strategies are studied in both sagittal-plane glide stabilization and three-dimensional heading maintenance. A passivity-based controller and a sliding mode controller are designed and tested in experiments for those two problems, respectively. In sagittalplane glide stabilization, a nonlinear observer is designed and implemented to estimate velocityrelated states. A three-dimensional curve tracking problem is also discussed and a two-degree-offreedom control scheme is proposed by integrating static inverse mapping and H∞ control technique. The differential geometric features, such as the torsion and curvature, are explored for planning the trajectory. Finally, the field tests with the lab-developed prototype of gliding robotic fish are conducted in the Kalamazoo River, Michigan and the Wintergreen Lake, Michigan for detecting oil spill and sampling harmful algal blooms, respectively. Both gliding and spiraling motions are tested in the experiments as well as the fish-like swimming. The field test results are presented to show the effectiveness of the designed robot in environmental monitoring tasks. Dedicated to my wife Mi Zhou and our son Andrew Anbang Zhang with all my love.