Successful Integration of Multispectral Imaging and Autonomous Robotic System at KGHM Underground Mine
Advancing Mining Automation Through Data Collection and System Integration
In November 2024, the first integration campaign has been successfully accomplished at the KGHM underground mine, bringing together experts from GSB, LTU, KGHM, and CUP. These tests allowed to evaluate the compatibility and operational efficiency of the developed technologies withing the project. The primary objective was to integrate GSB’s multispectral camera with LTU’s robotic autonomy stack and assess how well different sensor technologies function together in the mining environment. This initial campaign was entirely focusing on data collection and system validation, before transitioning to autonomous operations in future tests.
Over two days, the team carried out rigorous testing of a multispectral imaging system mounted on the LTU robot. The robot was equipped with a sensor package that includes 3D LiDAR for high-resolution mapping, RGB-D camera for visual scene reconstruction, an inertial measurement unit (IMU) for spatial orientation, and onboard computing unit. The integration process aimed to ensure seamless communication, operation and proper placement of these sensors for optimal performance in envisioned exploration scenarios.
The campaign involved multiple manual data collection steps to evaluate the effectiveness of all the technologies. Initially, LTU’s autonomy stack and GSB’s multispectral camera have been mounted on the robotic platform and the operation of all the systems has been successfully evaluated. As a next step LiDAR SLAM has been evaluated in the selected mining area to create baseline reference data for the digital twin. The multispectral imaging has been evaluated under direct human control in selected mining faces relevant to the deep mining use-case. As an outcome of the integration campaign partners collected combined data from hyperspectral camera and robotic system for detecting ore and mineral deposits. The campaign also allowed to collect rich data containing underground tunnels with multi-branched junctions, and dead-end drifts. These tests provided essential insights into sensors’ performance, PERSEPHONE robot requirements, and the overall integration of imaging for mineral detection and perception for navigation technologies.
This initiative represents a crucial step in advancing mining automation. Prior to this, underground mineral analysis relied on manual exploration and drilling campaigns, which were time-consuming and prone to errors. The successful integration and validation of multispectral imaging, alongside LiDAR and other sensing technologies, now pave the way for future mine automation. With the compatibility and operational effectiveness of these sensors now confirmed, the next phase of the project will focus on autonomous mineral exploration campaigns in deep and abandoned mines.
