J Intell Robot Syst (2013) 70:361–371 DOI 10.1007/s10846-012-9696-3 Using Genetic Algorithms for Tasking Teams of Raven UAVs Marjorie Darrah · Edgar Fuller · Thilanka Munasinghe · Kristin Duling · Mridul Gautam · Mitchell Wathen Received: 9 May 2012 / Accepted: 9 July 2012 / Published online: 20 July 2012 © Springer Science+Business Media B.V. 2012 Abstract Control of multiple unmanned aerial ve- hicles is of importance given that so many have been deployed in the field. This work discusses how genetic algorithms (GA) have been applied to the cooperative tasking of the AeroViron- ment’s Raven unmanned aerial vehicle (UAV) engaged in an intelligence, reconnaissance, and surveillance (ISR) mission. Mission assumptions, development of the GA, the method used to test for convergence, and the outcome of preliminary testing are all discussed. Keywords UAVs · Genetic algorithms · Cooperative control · Tasking algorithms 1 Introduction Many researchers have completed investigations to determine the most efficient teaming arrange- M. Darrah (B ) · E. Fuller · T. Munasinghe · K. Duling Mathematics, West Virginia University, Morgantown, WV, USA e-mail: mdarrah@math.wvu.edu M. Gautam Mechanical and Aerospace Engineering, West Virginia University, Morgantown, WV, USA M. Wathen Army Research Laboratory, Adelphi, MD, USA ments for various cooperative control algorithms for a variety of candidate missions. Schumacher et al. used the Mixed Integer Linear Programming (MILP) approach to assign vehicles to station- ary ground targets for a Wide Area Search and Destroy (WASD) mission [1, 2]. The objective function to be minimized in this scenario was based on final mission completion time. The mis- sion assumptions in their work assumed a priori knowledge of the battle space and non-survivable vehicles. Darrah et al. [3] extended this MILP work to a Suppression of Enemy Air Defense (SEAD) mission with survivable vehicles prose- cuting pop-up threats. In their paper, a dynamic constraint builder was applied to update a model based on the current battle space situation when new threats arose. A re-plan was completed in real-time allowing new targets to be added and partially or fully prosecuted targets to be updated in the model. From the re-plan a new set of constraints was built and solved. Although the MILP approach provided an optimal solution and was backed by a formal derivation that proved optimality of the solution, it proved to be very computationally intensive and thus required more time to solve than available in a real-time system implementation. Shima et al. applied Genetic Algorithms (GA) to the task assignment problem and developed an encoding scheme for a feasible solution as a chromosome [4]. Their work in this area first