非ホロノミック拘束系の最適軌道計画に関する研究 : 第1報, 関数近似解析法の提案

抄録

We have proposed a new optimum trajectory planning method for a system with nonholonomic constraints due to passive joints based on a function approximation method. The trajectory is approximated by a linear combination of Hermite functions and Fourier series. The combination variables are solved by a nonlinear programming method that considers the constraints in the same function space as that of the trajectory. We have examined the validity and features of this method by applying it to the optimum trajectory planning of a steady giant swing motion. It is found that this method has efficient convergence characteristics in regards to iteration calculation and can provide accurate solutions depending on the order number of the approximating functions. Optimum giant swings with no input torque can be obtained for various motion periods.