High-quality path planning for autonomous mobile robots with η3-splines and parallel genetic algorithms

In this paper we apply the composite eta3 splines to collision-free curvature-derivative continuous shorter path planning of wheeled mobile robots, modelled as unicycle, within known static environments. The path planning problem is formulated as a bi-objective optimization problem of searching a sequence of N ordered intermediate configurations between start and goal configurations over the group of all possible configurations that avoid collisions with obstacles while minimize the path length. We employ Pareto-based evolutionary multi-objective optimization using island parallel genetic algorithm (IPGA) with migration to tackle this problem. Extensive simulations demonstrate that the generated paths are statistically both shorter and smoother, up to third order geometric continuity, when compared to path planning by cubic spiral. In particular, the effect of N that represents path complexity on path planning performance of the evolutionary path planner based on IPGA is extensively tested on environments consisting of similar number of obstacles but different configurations.