Autonomous Fly Actuated Rescuer

AFAR is the Autonomous Fly-Actuated Rescuer—a search-and-rescue drone that sits at the intersection of biology, artificial intelligence, and robotics. The drone aims to help rescuers search dangerous or difficult-to-access environments by autonomously navigating obstacles while following a structured search plan. AFAR is powered by a combination of a male Drosophila (fruit fly) brain and Muse Spark. This architecture enables AFAR to balance the high-level planning of a large language model with the speedy decision making of a lightweight neural net. Our approach separates two challenges: responding quickly to nearby obstacles and planning where to search next. By using the brain of a ubiquitous flying animal, AFAR uses a nature-made neural network tuned for navigation while still being guided by a sophisticated large language model to successfully search for missing people. This division allows the navigation controller to handle immediate obstacles while Muse Spark maintains the broader search strategy. The repository includes Gazebo simulation runs demonstrating AFAR’s navigation across forests, collapsed buildings, and caves. AFAR’s sensory reception and navigation algorithm is analogous to an actual animal's. The drone’s camera and LiDAR provide sensory input, which the system processes to identify geometric shapes and potential obstacles from the images and point clouds they sense. This information is provided to the Drosophila neural network housed on the drone, which acts as the subcortical, low-level thought controller, processing the geometric information to quickly navigate around obstacles in the environment. Muse Spark is the cortical, high-level thought controller, processing the image data more deliberately to create a structured search plan for AFAR as it attempts to find missing people in risky environments.