Platform Development for the Implementation and Testing of New Swarming Strategies

Abstract

Swarm robotics--the use of multiple autonomous robots in coordination to accomplish a task--is useful for mapping, light package transport, and search and rescue operations, among other applications. Researchers and industry professionals have developed robotic swarm mechanisms to accomplish these tasks. Some of those mechanisms or “strategies” have been tested on hardware; however, the technical requirements involved in fielding a drone swarm can be prohibitive to physical testing. Team SWARM-AI has developed a platform that provides a starting point for testing new swarming strategies. This platform allows the user to select vehicles of their choosing- either air, land, or water based, or some combination thereof- as well as define their own swarming method. Using a novel decentralized approach to ground control software, this platform provides a user interface and a system of computational “units” to coordinate drone swarms with a centralized, decentralized, or combination architecture. Additionally, the platform propagates user input from the master unit to the rest of the swarm and allows each unit to request sensor data from other units. The user is free to edit the processes by which each drone interacts with the environment and the rest of the swarm, giving them freedom to test their swarming strategy. The software system is then tested with a swarm of quadcopters using Software in the Loop (SITL) testing.

Notes

Gemstone Team SWARM-AI

Rights