PF-DOP hybrid path planning for safe and efficient navigation of unmanned vehicle systems

Abstract

Potential Field is one of the widely used path planning methods in robotics. This method assumes that a robotic system knows its current position and where the goal is. The location of obstacles is also known in advance for global path planning, or it can be detected from sensing in local path planning. Although the potential field method provides a convenient and simple way of generating paths, most previous approaches do not incorporate uncertainty of a platform position while it is moving. Unlike satellite-based navigation systems, a user positioning accuracy may significantly change in a local area if a positioning system basis on Ultra Wideband (UWB) or LTE/5G is used. Because the positioning accuracy varies while a vehicle navigates, an unmanned vehicle or a robot following a given path may experience dithering movement. In addition, this problem may lead to unnecessary oscillation and increase the probability of collision which is a critical treat for a safe navigation. To overcome the drawbacks, this paper proposes the PF-DOP hybrid path planning algorithms, which is a variant of potential field that takes into account a user positioning accuracy. A user positioning accuracy (UPA) in an interested navigation area can be conveniently represented by Dilution of Precision (DOP) field. From the mixture of the potential and DOP fields, the proposed approach generates a hybrid directional flow that is able to guide an unmanned vehicle to the safer and more efficient paths. The hybrid directional flow is made by mixing the two fields with weighting factors which is also used in traditional potential field method as a coefficient. The weighting factors will vary depending on the user locations, but as the positioning accuracy can be derived from the user location, the user positioning accuracy is also associated. The paper will describe the systemic procedure in generating the PF-DOP hybrid path and how to avoid local minimum by using this algorithm.