The rise of deep learning and cloud computing have changed the design paradigms of swarm robotics, giving a higher level of embodiment to robotic cyber-physical systems. This paper shows how real-world scenarios can be developed in an integrated platform made of autopilot, robotic operating system and 3D simulator. We focus on two key swarm intelligence mechanisms: stigmergy, for indirect coordination via virtual pheromones, and flocking, for decentralized and flexible movement. Developing a series of pilot experiments on real-world scenarios, and a platform built on PX4, Gazebo, and ROS2, the embodiment of such mechanism is shown. Early results highlight the effectiveness of this approach in promoting fault tolerance, adaptability, and efficient exploration in scenarios such as wildfire detection, posts-earthquake recovery, and underwater monitoring
Embodying swarm coordination in robotic cyber-physical systems
Mario G. C. A. Cimino;Pierfrancesco Foglia;Salvatore Arancio Febbo;Cosimo A. Prete
In corso di stampa
Abstract
The rise of deep learning and cloud computing have changed the design paradigms of swarm robotics, giving a higher level of embodiment to robotic cyber-physical systems. This paper shows how real-world scenarios can be developed in an integrated platform made of autopilot, robotic operating system and 3D simulator. We focus on two key swarm intelligence mechanisms: stigmergy, for indirect coordination via virtual pheromones, and flocking, for decentralized and flexible movement. Developing a series of pilot experiments on real-world scenarios, and a platform built on PX4, Gazebo, and ROS2, the embodiment of such mechanism is shown. Early results highlight the effectiveness of this approach in promoting fault tolerance, adaptability, and efficient exploration in scenarios such as wildfire detection, posts-earthquake recovery, and underwater monitoringI documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
