Neural-networks-based adaptive control of flexible robotic arms

This paper studies the use of RBF neural networks as key components of adaptive controllers aimed at controlling flexible robotic arms. The RBF networks in the proposed controllers are in charge of approximating the inverse dynamics of the arm. To this end, an efficient control law that exploits the approximation capabilities of RBF networks and the stabilizing properties of a servo loop is developed. A stability analysis for the proposed system is carried out and a series of results are discussed in detail. These results are encouraging as they serve to analyze the learning and adaptation behavior of RBF networks and also show the effectiveness of the proposed system.