Esmirna Anguiano
Research Mentor(s): Vasileios Tzoumas
Mentor Department: Aerospace
Authors: Esmirna Anguiano, Vasileios Tzoumas
Session: Session 7 (4:00pm – 4: 50pm)
Presentation Type: Poster 63
Abstract
This study employs multi-robot coordination to complete complex tasks such as tracking moving targets with unpredictable motions. During previous trials, we used the Robot Operating System (ROS) to run simulations, but it was slow and used a single centralized node for data processing. We will transition from ROS to ROS2 to enhance the robot-to-robot (r2r) simulated communication with decentralized data processing. Research into target-tracking is motivated by potential improvements and applications to autonomous navigation, security, and traffic control. Alternative algorithms have computational inefficiencies that increase with the number of robots. This project will expand on previous research on simultaneous system identification (SSI) and the Resource-Aware distributed Greedy (RAG) algorithm to allow for predictive modeling of moving targets and maximizing efficiency in r2r coordination. These additions will allow multi-robot networks to map and predict the movement of objects while optimizing resources. Research will begin with a literature review of distributed decision-making, system identification, and multi-agent control. Transitioning to the ROS2 library will improve simulations, environmental tuning, scalability, and ease real-world implementation with the Crazyflie drone system. A staged experimental methodology will be implemented, beginning with AirSim simulations and progressing to real-world scenarios. Real-world experiments will demonstrate RAG’s ability to distribute tasks and track moving objects. Intended results include an updated RAG algorithm that incorporates SSI to track moving objects while using minimal resources. The implications of this research can enhance autonomous vehicle navigation, drone surveillance, and even search and rescue missions. This research contributes to autonomous robotics by testing an algorithm that is both time and resource efficient.
