A semi-optimal path generation scheme for a frame articulated steering-type vehicle
Yossawee Weerakamhaeng, Takashi Tsubouchi, Masamitsu Kurisu, Shigeru Sarata · Advanced Robotics · 2006
A wheel loader is one of the main machines that is used in various applications. The present authors have been interested in autonomous control of its locomotion and navigation. An inevitable step in achieving this is path planning. The authors choose a technique of designing a steering method for car-like vehicles by proposing a semi-optimal path generation scheme using a canonical path skeleton consisting of three line segments and two clotohid pairs. Taking such a path skeleton, for any given pair of a locally desired initial configuration and a locally desired destination configuration, path planning is equivalent to finding the suitable values of parameters which characterize this skeleton. Due to the lower number of constraints than the number of parameters, ascertaining the values of parameters comes from the optimizations of the authors' formulated non-linear programmings with equality constraints. In addition, the realizable path must be subject to the upper-bounded curvature and curvature derivative. Adopting a genetic algorithm as the search tool in one of the optimizations makes it possible to allow the use of only a single canonical skeleton for the desired path planning.