Enhancing Body-Mounted LiDAR SLAM using an IMU-based Pedestrian Dead Reckoning (PDR) Model
Hamza Sadruddin, Ahmed Mahmoud, Mohamed Atia
Abstract
Simultaneous localization and mapping for human body-mounted platforms have been recently the focus of navigation research to support a wide range of applications such as rescue, first-responders, mining, and defense. For vehicular platforms, wheel odometry has been used to enhance the accuracy of SLAM. However, wheel odometry is not available in body-mounted platforms. Using raw inertial measurement unit (IMU) as odometry is not accurate enough to support SLAM due to the large and rapid drifts caused by IMU data integration. To address this challenge, we propose a sensor fusion scheme for body-mounted SLAM that integrates the IMU-based Pedestrian Dead Reckoning (PDR) model with a low-cost lightweight 2D LiDAR sensor. In the proposed fusion, the PDR model is used as a replacement for wheel odometry in vehicular platforms. A system prototype consisting of a helmet-mount IMU from Xsens and RPLIDAR A1 2D LiDAR sensor has been developed and used for field data collection. The developed PDR model was integrated into the Cartographer SLAM engine and compared with Hector SLAM. Our experiments demonstrated that the integration of PDR has enhanced the SLAM accuracy and contributed in bridging featureless portions of the environment leading to an overall average improvement of 71.47%.