This thesis deals with the problem of controlling a biomimetic fish-like underwater robot developed in the 7th framework projects FILOSE and ARROWS. The idea of these projects is to build faster, more agile, and less energy consuming submarines than the state of the art. Mimicking fish-like propulsion and flow sensing is part of these projects. Control algorithms of robots can be improved using knowledge about the flow to react to disturbances, change control schemes, and use features of the flow for energy saving. A prototype of the robot is endowed with five pressure sensors at its head to sense the flow and is actuated by a single motor driving its compliant tail. The swimming speed and orientation are determined by the actuation amplitude, frequency, and o set angle of the driving motor. All experiments are conducted in a flow tank creating a laminar flow of 0.2 to 0.4 bl/s (body length per second). The subtask solved here, is following a trajectory in the horizontal plane, consisting of swimming speed as well as lateral movement control while minimizing energy consumption. Either camera data giving the absolute position of the robot in the flow tank or pressure signals giving the relative speed and orientation are used as feedback signals. As a contribution to the work of other members of the team, the resonance frequency of the compliant tail is determined using the motor current. It is found to be 3.2 Hz and to be independent of the flow speed. Further, it is shown that the robot needs least energy to achieve a certain swimming speed using an actuation frequency of 2 Hz and controlling the actuation amplitude. This frequency is used further for actuation. In the first step, camera feedback controllers are implemented. The lateral movement of the robot as well as the dependency of swimming speed and actuation amplitude are characterized first. The derived and approximated relations are linear. Consequently, PID controllers are used to control the lateral position, orientation angle, and swimming speed. The actuation amplitude is used to control the swimming speed and the actuation o set angle is used to control the lateral movement as well as the orientation angle. The highest steady state error for the lateral position control is 13 mm corresponding to 15% of the robot's width and for the forward movement control the steady state error is 5 mm corresponding to 5% of the robots width. Comparable experiments for trajectory following with and without knowledge about the flow are conducted and show lower current consumption and better trajectory following for the navigation strategy using information about the flow. In a second step, the integrated pressure sensors are characterized and used as feedback system. A two point lateral movement direction control is implemented. The designed controllers using camera feedback meet the requirements confirming the decision for a linear control scheme. Further it is shown, that knowledge about the flow speed helps to improve controller performance and energy consumption. The pressure feedback controllers are inferior to the camera feedback controllers resulting from the poor signal to noise ratio of the pressure sensors. Furthermore, is not possible to implement an orientation and stagnation point control using the current pressure sensors. Nonetheless, it is shown that the principle of determining the orientation of the robot with respect to a flow using pressure sensors works and that these controllers can be implemented using an appropriate pressure sensor system. In the future, a basic step towards better controllers could be a more accurate determination of the flow speed leading to a better characterization of the robot. The camera based controllers can be improved by adopting the PID parameters depending on the flow speed and implementing a tracking control in addition to the implemented fixed point control. The pressure feedback control can be improved by the use of more accurate pressure sensors and the implementation of new control algorithms using the new sensor system. Besides this report, a paper with the title 'Flow-aided path following of an underwater robot' has been written and submitted to the IEEE International Conference on Robotics and Automation (ICRA) 2013.

EMK-spezifische Daten:

Lagerort Dokument: Archiv EMK, Kontakt über Sekretariate

Bibliotheks-Siegel: 17/24 EMKM1801

Art der Arbeit: Masterarbeit

Beginn Datum: 10-05-2012

Ende Datum: 09-11-2012