| Abstract | This paper presents a concept for the nonlinear control design of a 4-axis manipulator arm driven by pneumatic muscle actuators. This arm, named AirArm, was developed within a research project of Festo's Bio Learning Network 2008. The control design is based on a cascade structure with four levels. The upper level generates the reference trajectories for the joints of the arm, in consideration of biokinetic motion patterns. These provide the set points for the nonlinear control system based on exact I/O linearization, which calculates the torque set points of the four joints. In this context, the separate design steps for this control system will be described. Due to the strong coupling between all of the joints, classical local controllers like adaptive PIs or PIDs, are unable to achieve satisfactory closed-loop dynamics. In the next lower level, a mathematical model is applied to detect the set point pressure of each muscle, using the torque and the reference angles of the arm. The innermost loop includes local pressure adaptive PI controllers for the pneumatic muscles. The determination of the control parameters using a model-based nonlinear control design will be described. Experimental results show the high performance and dynamic of the 4-axis manipulator arm using the presented nonlinear control concept. |
|---|
