Systems and methods for autonomous swarm operations in subterranean and GPS-denied environments

An autonomous drone swarm system for subterranean and GPS-denied operations comprising:

• a hierarchical swarm architecture comprising:
• at least one command drone comprising a large language model processor configured to process natural language instructions and generate autonomous mission plans;
• and
• a plurality of subordinate drones configured to execute tasks delegated by the command drone;
• a multi-modal communication system establishing encrypted mesh network connectivity between drones using at least two communication modalities selected from radio frequency, optical beaconing, acoustic signaling, and visual communication channels, wherein the mesh network implements self-healing protocols that automatically adapt routing pathways when individual communication links become unavailable;
• a GPS-denied navigation system comprising:
• multi-sensor dead reckoning algorithms processing inertial measurement data, visual odometry, and LiDAR scan matching;
• a Unified Ground Frame with Surface-Referenced Z (UGF-SRZ) coordinate system that dynamically selects surface references from hierarchical data sources comprising satellite-derived digital surface models, fog-maintained surface mosaics, local drone-generated cursory mapping, and algorithmic surface estimation, and computes signed vertical coordinates where positive SRZ values indicate positions above the selected surface reference and negative SRZ values indicate positions below the surface reference, thereby enabling unified coordinate representation across surface and subterranean operational environments;
• and
• collaborative positioning protocols that share navigation data across the mesh network to enhance swarm-wide positioning accuracy;
• a distributed mapping and localization system specifically configured for subterranean operations comprising:
• edge processing algorithms at individual drones that generate keyframes, feature descriptors, and point cloud fragments optimized for GPS-denied environments;
• fog layer processing that performs multi-agent map fusion, loop closure detection, and pose graph optimization using data streams from multiple drones, wherein the fog layer integrates UGF-SRZ surface corrections with collaborative mapping to maintain coordinate consistency between surface and subterranean operations;
• and
• map correction distribution that provides updated environmental models and UGF-SRZ positioning corrections back to individual drones, enabling seamless navigation transitions between surface and underground environments.