Collaborative Mission Planning & Autonomous Control Technology (CoMPACT) System Employing Swarms of UAVs
Jovan D. Bošković, Nathan B. Knoebel, Nima Moshtagh, Jayesh Amin, Gregory Larson · 2009
In this paper an autonomous hierarchical architecture for controlling swarms of unmanned aerial vehicles (UAV) is investigated. Its purpose is to carry out complex missions involving Intelligence, Surveillance, and Reconnaissance (ISR), Suppression of Enemy Aerial Defenses (SEAD), Destruction of Enemy Aerial Defenses (DEAD), Close Air Support (CAS), and other functions in an environment characterized by large uncertainty and vehicle loss or failures. The architecture is referred to as the Collaborative Mission Planning & Autonomous Control Technology (CoMPACT) and consists of six coordinated hierarchical layers encompassing tasks and objectives from the mission planning stage to the UAV task execution level. The main advantage of the CoMPACT system is that it e ectively combines mission planning using evolutionary algorithms, hybrid automata-based task execution at each hierarchical layer, and biologically-inspired emergent swarm behaviors. Judicious combination and coordination of these capabilities results in a system that is scalable to architectures involving a very large number of autonomous agents in which mission-relevant objectives are accomplished in highly uncertain dynamic environments through real-time mission re-planning, dynamic agent-to-task assignment, and reactive motion planning. Through simulation and experiment, the CoMPACT system is shown to be robust to intermittent loss of communications and communication delays, and to dynamic changes in the environment such as moving targets, unknown targets and obstacles, pop-up threats, vehicular failures and loss of agents. The CoMPACT system has been extensively tested on a Matlab simulation of a realistic mission scenario involving ISR, SEAD and DEAD functions, and simple models of UAV dynamics operating in a dynamic threat and target-rich environment.