Abstract

This project was completed at the West Virginia Univeristy robotics REU program in 2019 as part of a cohort of 10 peers.

Robotic swarms, where many simple agents interact to achieve a shared goal, are robust to failure and can be produced at low cost, making them desirable for tasks such as search-and-rescue or exploration. Human-swarm interaction (HSI) can measurably increase a swarm’s performance, as a human operator may have intuition or knowledge unavailable to the swarm, and thus could reactively shape the swarm’s behavior as needed. Despite the potential benefits of HSI, few labs have the capability to run human-in-the-loop tests on large (greater than 10) groups of robots. In this project, we developed the Adaptable Platform for Interactive Swarm robotics (APIS), which lowered the barrier for entry to HSI research without being constrained to a fixed ruleset or interface. APIS’s architecture emphasizes flexibility of algorithms and interface methods, and maintains a low cost to improve researchers’ access to swarm platforms. Our results demonstrated the ability for a human operator to guide a 50-strong swarm exhibiting foraging behaviors by introducing environmental information through intuitive gestures.

Technologies Used

• ROS, Linux
• C++
• Motion capture, depth cameras
• Embedded development
• TCP/IP networking

Under Construction!