Evaluation of LiDAR Inertial Odometry method with 3D LiDAR-based Sensor Pack

As the 4th industrial revolution emerges at the forefront of South Africa's national strategy, research areas like mapping and localization find importance in more fields than just robotics. The mining industry is well-positioned to be a potential beneficiary of these technological changes. By nature, mining settings could be labeled with a similar status to indoor areas as they are both GPS-denied type environments. Mapping and localization algorithms using Simultaneous Localization and Mapping (SLAM) are proven to function in similar conditions. These SLAM-based algorithms are highly effective at mapping, yet they can be susceptible to registration, motion distortion, and drift issues if provided with no external odometry. Also, using mobile robots may not always be possible in these environment types for practical reasons. Employing a device with a different form factor, such as a mapping sensor pack, could be an option. This study evaluates a Lidar Inertial Odometry solution integrated on a LiDAR-based sensor pack developed for mapping and localization applications. For the chosen LiDAR Inertial Odometry (LIO) solution the Root Mean Squared Error was computed. This was found to be greater than the Root Mean Squared Error computed by the LiDAR sensor pack's Eth-ICP Mapper implementation. However, the LIO solution produces pose estimates at a higher rate, which is beneficial for localization continuity and mapping.