17th World Congress, International Federation of Automatic Control, IFAC, Seoul, Güney Kore, 6 - 11 Temmuz 2008, cilt.17
In this paper protraction movement, namely stepping ahead, of a three joint robot leg is optimized for energy efficiency for any given pair of initial-final tip point positions. For the optimization a modified version of gradient descent based optimal control algorithm with Hamiltonian formulation is used. The objective function is modified in steps to jump over the infeasible and inefficient local optimums. The results of 79 optimizations are used to construct a radial basis function neural network (RBFNN) in order to interpolate between the optimized trajectories. The results are presented and discussed in the paper. Copyright © 2007 International Federation of Automatic Control All Rights Reserved.