Computational Time Reduction in Stream Field Navigation Through Dynamic Path Search and Selection in Kinematic Environments

Abstract

Path planning algorithms based on physical models have been in development to allow navigating an environment in a way that mimics nature. Contrary to combinatorial and sampling-based algorithms, which represent the field by a set of nodes then use searching techniques to find a path, the artificial potential field method solves this problem by attracting the robot to the target and repelling it from the obstacles While the Stream Field Navigation (SFN) algorithm solves the problem by simulating a fluid field. This research proposes a dynamic path search and selection schema to allow the SFN algorithm to deal with moving obstacles. This research shows how the algorithm deals with a kinematic (variable) environment using the discretized path calculations. The dynamic search and selection algorithm is also capable of updating the robot's path when the goal is re-located. The developed approach is shown to greatly reduce the computational time required to obtain a path compared to re-simulating the field when a change is introduced to the environment.