American Institute of Aeronautics and Astronautics 1 A Novel Cooperative Path Planning for Multiple Aerial Platforms Genshe Chen * , Dan Shen † , Jose B. Cruz, Jr. ‡ , Chiman Kwan § Stephanie P. Riddle ** , Steven Cox †† , and Carey Matthews ‡‡ Cooperative path planning (CPP) is one of the core steps to effectively exploit the capabilities of multi-level cooperative control of multiple aerial platforms. The main purpose of CPP is to develop a set of algorithms such that platforms in a given scenario could cooperatively find a desired path to reach certain destinations. CPP specifies way-points for platforms to reach certain destinations while meeting certain requirements, including minimizing en route threats, meeting time constraints, keeping mutual spacing, and decreasing fuel consumption. In this paper, we propose a novel path planning method based on Pareto optimization and a foraging algorithm. Our graph cut based Pareto solution serves as a reference trajectory for the foraging algorithm, which further dynamically refines the reference path. We have implemented and evaluated our proposed CPP algorithm. In addition, extensive experiments and comparative studies have been carried out. In particular, we compared our algorithm with the Pareto-Voronoi approach and the original Voronoi algorithm. I. Introduction n recent years a significant shift of focus has occurred in the field of autonomous unmanned platforms such as Unmanned Air Vehicles (UAV) or Unmanned Ground Vehicles (UGV) as researchers have begun to investigate problems involving multiple units rather than a single unit. Cooperative Path Planning is the guidance of a group of agents - in our case, a team of UAVs - from a source to a destination, while avoiding all encountered obstacles or threats. Traditionally, there are four main path planning algorithms in the literature [1]. The first one is called skeleton or highway approach, in which all the feasible paths are retraced to a roadmap. Path planning is reduced to a graph problem. The well known skeleton approaches are Voronoi diagrams [5, 6], and a visibility graph [7]. This approach is static and offline. It uses discrete methods for a problem that is fundamentally continuous. The second is cell decomposition. In this approach, the planning space is reduced to a fixed grid or a set of cells, which is usually modeled by an un-directed graph. The famous Dijkstra’s algorithm and its variant [2-4] fit into this catalog. There are two disadvantages to this approach. The grid graph can become very large and consequently the algorithms can be very time-consuming. Also, the resulting paths may involve numerous line segments, and some of which may have excessive turn angles. The third one is called the potential field approach, in which a scalar function is constructed from the information of obstacles (threats) and contractions (targets). The path is determined by performing steepest gradient descent on the potential function. This approach is simple and natural. But the potential function can be very complex when many obstacles exist. Another disadvantage is that the agents or UAVs will be trapped by a local minimum. The foraging or swarm algorithm [8] is an improved potential field approach. In this swarm algorithm, the requests of all UAVs to stay in a group while maintaining collision avoidance are also considered in the potential function. The last approach is mathematical programming, in which all the obstacles are modeled by a set of inequalities. Path planning is converted to a mathematical optimization problem, to which the * Senior Research Scientist, Intelligent Automation, Inc., AIAA Senior Member. E-mail: gchen@i-a-i.com. † Ph. D. Student, Electrical and Computer Engineering, The Ohio State University, Columbus, OH, 43210. ‡ Distinguished Professor of Engineering, The Ohio State University, Columbus, OH, 43210. § Vice President, Intelligent Automation, Inc., 15400 Calhoun Dr., Suite 400, Rockville, MD 20855. ** NAVAIR Science and Technology Air 4.0X1 †† NAVAIR Science and Technology Air 4.0X1 ‡‡ NAVAIR Science and Technology Air 4.0X1 I Infotech@Aerospace 26 - 29 September 2005, Arlington, Virginia AIAA 2005-6948 Copyright © 2005 by the American Institute of Aeronautics and Astronautics, Inc. All rights reserved. Downloaded by UNIVERSITY OF MARYLAND on October 20, 2015 | http://arc.aiaa.org | DOI: 10.2514/6.2005-6948