Validation of ORB-SLAM3 and VINS-Mono with Low-Cost Sensor Setup in Outdoor Environment

In this paper, a comprehensive introduction of ORB-SLAM3 and VINS-Mono is presented. Both visual SLAM (vSLAM) solutions have been tested in real-time under real conditions. For their validation publicly available datasets collected by using high-quality synchronized sensors and a dataset created by the authors have been considered. The latter one contains high-quality reference data captured by an accurate IMU-RTKGNSS system as well as lower quality test data (acquired by our low-cost camera-IMU system) on which the above methods have been tested. The results of the performed evaluation well reflect both the advantages and disadvantages of both methods. This survey highlights also the problems, which arise due to the use of low-cost sensors. We also verified that the ORB-SLAM3 Mono-Inertial method cannot be used with unsynchronized sensors while VINS-Mono can handle the timeshifts between the sensor messages. According to our results, both methods provide the same level of accuracy in case of considering the above mentioned datasets.