Ardua

The project aims to develop systems with heterogeneous robots (UGV, UAV, USV) operating in complex, large-scale, and dynamic or underground environments. Communication systems were developed using hybrid WiFi and LoRa protocols, with wireless mesh networks and UWB-integrated LoRa modules. Localization techniques include 2D/3D SLAM using LiDAR in caves and mines, semantic segmentation, and topo-semantic mapping via pose-graphs. Underground environments were simulated using scalable labyrinth generators with discrete topological features. Autonomous planning and navigation techniques were created for all robot types, including DWA-3D for UAVs and reactive navigation for USVs and UGVs. Advanced methods include high-speed drone navigation in tunnels and dynamic environments using Deep Reinforcement Learning. Demonstrators include a robust software architecture and the RAWFEE testbed for land, air, and water robots. Key applications were tested in real and simulated environments such as Somport Tunnel, Román Cave, and academic labs. A robotic boat was used for cave mapping, and UGVs and UAVs were deployed in tunnels and dynamic spaces. Overall, the project integrates communications, localization, and navigation for multi-robot systems in challenging settings.