Clustering Method for Monte Carlo Model Predictive Control

摘要

Monte Carlo Model Predictive Control is a variant of a sampling-based model predictive control, which is suitable for extensively parallel processors. In this study, we focus on the multimodality of the objective function and propose a clustering method to cope with it. As an application of multimodality, we consider a navigation problem of a mobile robot that avoids some obstacles. Simulation results show that our method is effective when the avoiding path splits into pieces.