Core Concepts
Robotic Total Stations provide more stable and reliable measurements compared to GNSS systems, making them valuable for SLAM benchmark development.
Abstract
I. Introduction
Various datasets exist to evaluate SLAM algorithms.
Current datasets lack comprehensive data-collection protocols.
The RTS-GT dataset aims to provide six-DOF ground truth trajectories using Robotic Total Stations (RTSs).
Comparison with GNSS-based setups shows RTSs are 22 times more stable in various environments.
II. Related Work
Majority of datasets use GNSS-Aided INS for reference trajectory generation.
Challenges with GNSS systems in urban environments.
Previous datasets have used single RTS for reference trajectories.
III. Hardware
Two robotic platforms used: Clearpath Warthog and SuperDroid HD2 UGV.
Setup includes Trimble S7 RTSs and Reach RS+ and Trimble R10-2 GNSS receivers.
IV. Data Collection
Dataset gathered over 17 months in diverse conditions.
Deployments in campus, tunnels, and forest environments.
V. Discussion and Challenges
Precision and Reproducibility:
RTS system exhibits median precision of 4.5 mm, while GNSS system has a median precision of 118.1 mm.
Challenges Encountered:
Leveling RTSs on snow-covered ground.
Obstacles disrupting measurements between prisms and RTSs.
Dust on lenses affecting tracking mode of the RTS.
Stats
"Our results demonstrate that RTSs provide measurements that are 22 times more stable than GNSS in various environmental settings."