Comparison of Localization Algorithms between Reduced-Scale and Real-Sized Vehicles Using Visual and Inertial Sensors

TL;DR

This study compares visual and inertial localization algorithms on reduced-scale vehicles versus real-sized vehicles, showing minimal differences in rotational accuracy and highlighting OpenVINS's superior performance.

Contribution

It evaluates the impact of vehicle scaling on localization accuracy using ROS2-compatible algorithms and provides insights into their applicability for testing autonomous vehicle systems.

Findings

OpenVINS has the lowest average localization error among tested algorithms.

Minor differences in translational motion estimation between scaled and real vehicles.

No significant differences in rotational motion estimation accuracy.

Abstract

Physically reduced-scale vehicles are emerging to accelerate the development of advanced automated driving functions. In this paper, we investigate the effects of scaling on self-localization accuracy with visual and visual-inertial algorithms using cameras and an inertial measurement unit (IMU). For this purpose, ROS2-compatible visual and visual-inertial algorithms are selected, and datasets are chosen as a baseline for real-sized vehicles. A test drive is conducted to record data of reduced-scale vehicles. We compare the selected localization algorithms, OpenVINS, VINS-Fusion, and RTAB-Map, in terms of their pose accuracy against the ground-truth and against data from real-sized vehicles. When comparing the implementation of the selected localization algorithms to real-sized vehicles, OpenVINS has the lowest average localization error. Although all selected localization algorithms have overlapping error ranges, OpenVINS also performs best when applied to a reduced-scale vehicle. When reduced-scale vehicles were compared to real-sized vehicles, minor differences were found in translational vehicle motion estimation accuracy. However, no significant differences were found when comparing the estimation accuracy of rotational vehicle motion, allowing RSVRs to be used as testing platforms for self-localization algorithms.