Dynamic Swarming Autonomous Underwater Vehicles through Behavioral Rules Using Declarative Programming

This article proposes a framework for dynamic swarming of Autonomous Underwater Vehicles (AUVs) using declarative programming to enhance underwater exploration and acoustic sensing. By integrating swarm intelligence, bio-inspired behaviors, and declarative rules (Prolog), the system enables adaptive reconfiguration of sensor arrays to optimize the detection and localization of underwater sound sources. The study addresses the challenge of dynamically alternating between vertical and horizontal linear array configurations (critical for estimating depth/range and azimuth, respectively) while maintaining swarm cohesion and collision avoidance. The open-source LRAUV model is employed for hydrodynamic simulations, with trajectories integrated into acoustic simulations combining modal propagation (low frequencies) and ray tracing (higher frequencies). Prolog's declarative paradigm allows high-level specification of formation control rules, collision avoidance constraints, and mission objectives, decoupling mission intent from low-level execution. This approach enables intuitive addition of constraints (e.g., safety radius, separation distances). We validate the framework in a ROS2/Gazebo simulation environment, in which five AUVs execute a circular trajectory, alternating between horizontal and vertical formations. Simulations demonstrate the swarm's ability to dynamically reconfigure formations while respecting non-holonomic motion constraints. In addition, an AUV mule autonomously navigates to collect data from swarm members and proves the scalability of the system for data collection tasks. The results underscore the robustness of the framework to keep formation stability, avoid collisions, and adapt to specific mission requirements.