Recurrent curvature smoothing in path planning problems for wheeled robots

A path planning problem for a wheeled robot is considered. When robot follows a path in a manual mode, its positions are measured by a GNSS receiver and stored as a target path to be repeated in an automated mode. It is required to construct a trajectory that satisfies certain smoothness requirements and curvature constraints. Due to errors inherent in the data points, the shape of the curve approximating the desired path may turn out inappropriate, or even unacceptable, from the point of view of control. In the paper, a recurrent method for improving curvature of curves consisting of uniform cubic B-splines is proposed. The method can be applied to real-time processing of large-scale data under memory limitation.