Context-Sensitive Guidance, Navigation, and Control for Autonomous Vehicles in GNSS-Denied Underground Environments

Abstract

Autonomous vehicles operating in complex, GNSS-denied environments such as underground mines face challenges beyond standard navigation tasks. We present a context-sensitive Guidance, Navigation and Control (GNC) framework, first applied in an underground mining scenario, with the potential to generalize across othermission-critical domains.The system integrates multiple spatial reasoning layers, e.g., a high-level site layout map, administrator-defined maps of forbidden and preferred areas, a communication topology map indicating connectivity to fixed relay nodes, and a SLAM-based local map for real-time perception. These maps are generated using different information sources and are dynamically combined into context-sensitive cost maps based on the vehicle’s current objectives and system-level constraints.All processing is performed at the far edge, enabling real-time operation without reliance on external cloud infrastructure. This setup ensures resilience in communication-degraded environments, supporting fast, onboard decision-making. Moreover, as connectivity permits, the entire fleet is connected to a centralized map database, enabling consistent, shared situational awareness.A key component is a decision-making layer that governs when and how each map should be used based on available data’s uncertainty, reliability, and mission relevance. For instance, if SLAM-based local mapping becomes unreliable due to dust or featureless tunnel walls, the system may revert to a high-level site layout map combined with odometry and relay node signal strength to estimate position.Shared situational awareness enables vehicles to act not only on their own sensory inputs but also on the collective knowledge of the fleet. If a vehicle enters a low-connectivity zone, nearby vehicles with stronger communication links to the network can act as mobile relays, forwarding its data and updates to maintain mission coordination.This modular and adaptive approach to multi-map fusion for GNC enhances situational awareness and operational robustness, making it particularly suitable for mining, defense and security applications in subterranean or similarly constrained environments.