Present position: Assistant Professor
|Thesis title:||Visual Servoing of Flexible Manipulators|
|Research area:||Control Systems|
Lightweight flexible manipulators are used today in a variety of applications, ranging from the traditional space robotics field to less known tasks, like exploration of hazardous environments or nuclear waste retrieval. Use of lightweight structures in industrial applications is also an interesting alternative to the traditional industrial robots: advances in dexterity and power consumption are expected. Reduced masses and extended workspaces however enhance the effects of the flexibility of the materials: in this scenario, an accurate model of the system and an efficient control technique are crucial. The first problem that has to be faced is thus the development of a dynamic model of the manipulator, that should be both accurate and computationally efficient. The model adopts a mixed Eulerian and Lagrangian formulation and exploits the chained structure of the equations of a serial manipulator. Moreover, in order to achieve the accuracy requirements, the dynamic effects of the motors at the joints, including gyroscopic terms, are fully taken into account. The flexibility of the materials makes also an accurate indirect estimation of the tip position extremely difficult, seriously compromising the accuracy of end effector positioning control loops based on standard mechanical sensors, such as encoders or strain gauges. The use of visual feedback can overcome this drawback, as visual information yields a direct measurement of the tip position, avoiding estimation errors due to a rough knowledge of the physical parameters. A visual servo controller for the tip trajectory tracking of a flexible manipulator has been conceived, inspired by the two-time scale control of flexible structures and by the well-known task space control. The visual information is used in the “slow” subsystem to achieve the tracking of a moving target, while strain gauge measurements are exploited in the “fast” subsystem to damp out the vibrations. Furthermore, if a high speed camera is available, the visual information yields the measurement both of the rigid position and of the displacement of the manipulator tip induced by the deformations. An improved visual servoing loop is thus presented, where the camera sensor is used both in the slow controller, to measure the tip position, and in the fast controller, together with the strain gauges, to determine the tip deformation. Finally, an experimental validation of the developed control techniques has been performed on a single-link flexible manipulator designed on purpose.