Deployment of an Aerial Multi-agent System for Automated Task Execution in Large-scale Underground Mining Environments

TL;DR

This work tackles the challenge of autonomous inspection in large-scale underground mines by coupling a central, auction-based task allocator with fully autonomous, modular aerial agents. The methodology hinges on automatically synthesizing task-specific behavior trees via a back-chaining approach from a library of primitive actions, and on a risk-aware, path-planning stack that combines D$^*_+$ planning with NMPC-APF tracking. Field validation with three UAVs in an active LKAB mine demonstrates reactive task allocation, infrastructure-free execution, and operator-driven task addition via an intuitive GUI, backed by a mobile Wi-Fi mesh. The results show effective seven-task completion across 200+ meters of tunnels within 311 seconds, highlighting improved safety and efficiency for routine inspections, gas monitoring, and mapping in subterranean environments. The framework thus enables scalable, robust aerial automation in hazardous mining contexts and can generalize to broader underground automation tasks.

Abstract

In this article, we present a framework for deploying an aerial multi-agent system in large-scale subterranean environments with minimal infrastructure for supporting multi-agent operations. The multi-agent objective is to optimally and reactively allocate and execute inspection tasks in a mine, which are entered by a mine operator on-the-fly. The assignment of currently available tasks to the team of agents is accomplished through an auction-based system, where the agents bid for available tasks, which are used by a central auctioneer to optimally assigns tasks to agents. A mobile Wi-Fi mesh supports inter-agent communication and bi-directional communication between the agents and the task allocator, while the task execution is performed completely infrastructure-free. Given a task to be accomplished, a reliable and modular agent behavior is synthesized by generating behavior trees from a pool of agent capabilities, using a back-chaining approach. The auction system in the proposed framework is reactive and supports addition of new operator-specified tasks on-the-go, at any point through a user-friendly operator interface. The framework has been validated in a real underground mining environment using three aerial agents, with several inspection locations spread in an environment of almost 200 meters. The proposed framework can be utilized for missions involving rapid inspection, gas detection, distributed sensing and mapping etc. in a subterranean environment. The proposed framework and its field deployment contributes towards furthering reliable automation in large-scale subterranean environments to offload both routine and dangerous tasks from human operators to autonomous aerial robots.

Figures (20)

• Figure 1: The evaluation environment at a depth of over 500m in an active production area of the mine located in Kiruna, Sweden. (a) and (b) show the tunnel environment, with the aerial vehicles on the ground before a mission is initiated, (c) illustrates one of the many challenges for autonomy in a mining environment, flying in smoke-filled, or dusty, environments. Finally, in (d), a snapshot of an empty tunnel is presented to show the scale of the operating environment.
• Figure 2:
• Figure 3:
• Figure 4:
• Figure 6: Local autonomy overview. Describes how the overall architecture operates. Highlighting the key components and that the coordination layer is responsible, based only on the estimated task execution costs $c_{k, j}$, to allocate tasks $x^*$ from the pool of available tasks $\mathfrak{T}$. Fig. \ref{['fig:fullSys']} illustrates how every agent operates locally in greater detail.