Path Optimization and Multi-level Path Planning for the Steam Field Navigation Algorithm

The computational problem of path planning is considered as a core functionality for robotic systems. Physics-based path planning algorithms are developed to generate a path for a robotic agent between two points using a mathematical model as the main building block. In these physical algorithms, the field is not represented as a grid of cells or a set of nodes where a search technique is used to find the path, such as the case for combinatorial and sampling-based algorithms. The Stream Field Navigation (SFN) algorithm is a physical algorithm that is developed to solve the path planning problem by simulating a fluid field. One of the properties of paths generated by the SFN algorithm is that they may take a longer path than required due to the nature of the underlying physical model. To overcome this phenomena, this research proposes a method to optimize the path generated by the SFN algorithm. The proposed method consists of three algorithms that can be successively applied to a path with the aim of making it shorter. The results for the developed path optimization algorithms show an average of 14.5% reduction in path length compared to the original path generated by the SFN algorithm. Such contribution has a favorable effect on energy expenditure by systems using the SFN algorithm due to the direct relation between path length and system energy consumption. This research also proposes a method to allow the SFN algorithm to operate in multi-level environments. This addition enhances the standard SFN algorithm as it is designed for single-plane path planning. The results for the developed multi-level path planning approach show a comparison of the effectiveness of the SFN algorithm with other algorithms in these multi-level room-based environments where the SFN algorithm is shown to produce paths in less computational time and the paths are of higher quality.