Game-Theoretic and Genetic-Based Approach for Cooperative Mission-Oriented Swarms of Drones

Abstract

The artificial intelligence applied to a drone has enabled a drone-swarm to operate autonomously as a group and unlocked many new potential applications that deal with more sophisticated tasks. In this paper, we present a game theory mechanism and nature-inspired algorithm that enable a fully autonomous drone-swarm to perform cooperative mission-oriented operations to some distinct targets. These operations require a small-team formation for each target with the potential overlap team member between teams and multiple task assignment and operations scheduling to ensure the mission success in a timely manner. The drone-swarm is modeled and simulated as a multi-agents system. A fully autonomous drone is represented as an intelligent agent with a certain dynamic risk tolerance level. An agent can decide based on the current risk tolerance level to participate in the auction-based team formation for a specific target while the genetic algorithm approach is used for the task assignment and operations scheduling. A multi-agent system simulator, which can be used to visualize, evaluate, and analyze the proposed team formation, task assignment, and operation schedule; is built using Netlogo, a multi-agent programmable modeling environment. A case study and its simulation results are provided to demonstrate the potential use of the proposed approach.