Abstract:

 Multiple short curves are joined to a long curve in path planning, which usually could only realize C¹ continuity and could cause the second order derivative to be discontinuous in the joint of adjacent short curves. To resolve this problem, the cubic B-spline curve with the feature of C2 continuity was used to implement path planning carried out by a two-stage particle swarm optimization. Control vertexes of the B-spline curve were determined at the first stage of two-stage particle swarm optimization, and then the optimal path was searched for at the second stage. To ensure the particles effectiveness, the particles initialization was improved according to the characteristic of the Bspline curve. The length of the path and collision detection reflected on the arc length factor and collision factor respectively in the fitness function. The global optimum was gradually approached with the loop by the particles. Experimental results indicated that the improved initialized particles were closer to the actual path and that the C2 continuous path could be achieved by two-stage particle swarm optimization.

Key words: particle swarm optimization, path planning, navigation;collision avoidance, mobile agents