Skip to Main Content
In recent years, the concept of modular and reconfigurable robotics emerged as a means for flexible and versatile automation. This concept allows for the execution of many complex tasks that cannot be performed by fixed-configuration manipulators. Nevertheless, reconfigurable robots introduce a challenging level of complexity to the problem of design of controllers that can handle a wide range of robot configurations with reliable performance. This paper addresses the position control of modular and reconfigurable robots. We develop a practical intelligent-control architecture that can be easily used in the presence of dynamic parameter uncertainty and unmodeled disturbances. The architecture requires no a priori knowledge of the system-dynamics parameters. Adaptive control is provided using fuzzy gain tuning of proportional-integral-derivative parameters in the presence of external disturbances. The architecture also provides learning control using feedforward neural networks. Moreover, the architecture has the capability of updating the adaptive control under reconfigurability. Experiments on a modular robot test bed are reported to validate the effectiveness of the control methodology.