Abstract
This paper is concerned with the planning problem of smooth global path for mobile robots with kinematic constraints. A new approach is proposed that is based on a modified particle swarm optimization (MPSO) combined with the (3)-splines. Some preliminaries on (3)-splines are first introduced to interpolate an arbitrary sequence of points with assigned kinematic parameters associated with the motion and control of mobile robots. Then, the MPSO algorithm with adaptive random fluctuations (ARFs) is proposed to deal with the frequently encountered local convergence issue in the planning of the global smooth path. In the MPSO algorithm, the evolutionary state is classified averagely at each iteration according to the evaluated evolutionary factor, where the velocity updating dynamics switches among various modes according to the evolutionary state and the ARFs are enforced on the global/local best particles in each mode of the current iteration. The proposed MPSO is verified to outperform five other well-known PSO algorithms via comprehensive simulation experiments on a collection of benchmark functions. Finally, the new approach by combining the MPSO algorithm with the (3)-splines is exploited to handle a double-layer smooth global path planning problem with several kinematic constraints.