A controller design method based on functionality

Robots are expected to expand their range of activities to human environment. Robots in human environment need redundancy for environmental adaptation. Furthermore, they have to automatically modify their controllers in response to varying conditions of the environment. Therefore, the authors have proposed a method to design a hyper-degrees-of-freedom (DOF) control system efficiently. The method decouples a large control system into small independent components called "function." Motion of the entire control system is expressed as superposition of multiple functions. Combination of some functions realizes many patterns of motion. Hence, various motions are realized with much smaller efforts on controller design. Additionally, the controller design is explicit since a controller and a function correspond directly. This paper expands the method to multi-DOF robots in 3-D space, since the conventional method was limited to a multirobot system in 1-D space. A new problem of interference among function-based systems occurs along with the expansion. A disturbance observer is applied on each actuator to eliminate the interference. Procedures of controller design under varying conditions are also shown. The proposed method is applied to a grasping manipulator with 18 DOF. Its experimental results show the validity of the method.