ROBOTIC NAVIGATION AND MAPPING IN GPS-DENIED ENVIRONMENTS WITH 3D LIDAR AND INERTIAL NAVIGATION UTILIZING A SENSOR FUSION ALGORITHM

Abstract

Global Navigation Satellite Systems (GNSS) do not currently offer viable solutions for autonomous robotic navigation in a tactical scenario, as GNSS networks are susceptible to enemy jamming and loss of coverage within buildings. Several methods for interior navigation and mapping fuse data inputs from inertial measurement units (IMU) and light detection and ranging (lidar) sensors to generate more robust simultaneous localization and mapping (SLAM) solutions. However, these methods rely on large point-cloud data sets to achieve SLAM that increases processing requirements. This work seeks to find a novel solution for decreasing point-cloud processing requirements for SLAM by combining IMU and lidar sensor inputs. Utilizing a strap-down navigation algorithm with zero-velocity updates, a fusion algorithm generates an estimated transform between lidar scans that is then provided as the input to an iterative closest point registration (ICP) algorithm to generate a SLAM solution. It was found that the required number of point-clouds for generating SLAM solutions was reduced by at least five times while still maintaining functionality through multiple rotations and translations over several meters. Future work recommendations include expansion of the fusion algorithm onto autonomous platforms and generating more efficient process flows to further reduce SLAM processing requirements.