Abstract:
© 2017, Springer International Publishing AG. Successful interactive collaboration with a human demands mobile robots to have an advanced level of autonomy, which basic requirements include social interaction, real time path planning and navigation in dynamic environment. For mobile robot path planning, potential function based methods provide classical yet powerful solutions. They are characterized with reactive local obstacle avoidance and implementation simplicity, but suffer from navigation function local minima. In this paper we propose a modification of our original spline-based path planning algorithm, which consists of two levels of planning. At the first level, Voronoi-based approach provides a number sub-optimal paths in different homotopic groups. At the second, these paths are optimized in an iterative manner with regard to selected criteria weights. A new safety criterion is integrated into both levels of path planning to guarantee path safety, while further optimization of a safe path relatively to other criteria is secondary. The modified algorithm was implemented in Matlab environment and demonstrated significant advantages over the original algorithm.
