Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

An Earth Rover dataset recorded at the ICRA@40 party

Qi Zhang, Zhihao Lin, Arnoud Visser

TL;DR

This work addresses the need for open-world, real-world navigation benchmarks for Earth Rover-like platforms by proposing a Rotterdam dataset collected with the FrodoBot Earth Rover Zero during the ICRA@40 event. It combines multiple sequences with front/rear cameras, IMU, GPS, and 4G connectivity, capturing at $3$ Hz and later upsampling to $6$ Hz, with depth maps produced from monocular RGB via LiteMono. Baseline demonstrations pair YOLOX-X object detection and LiteMono depth estimates with ORB-SLAM3 trajectories, revealing robustness limitations in waterfront scenes with repetitive textures and moving objects. The authors advocate extending to Scene Action Maps and cross-embodiment navigation to improve real-world SLAM and planning, arguing that the dataset will be valuable for training and evaluating autonomous navigation under open-world conditions, including public and waterfront contexts.

Abstract

The ICRA conference is celebrating its $40^{th}$ anniversary in Rotterdam in September 2024, with as highlight the Happy Birthday ICRA Party at the iconic Holland America Line Cruise Terminal. One month later the IROS conference will take place, which will include the Earth Rover Challenge. In this challenge open-world autonomous navigation models are studied truly open-world settings. As part of the Earth Rover Challenge several real-world navigation sets in several cities world-wide, like Auckland, Australia and Wuhan, China. The only dataset recorded in the Netherlands is the small village Oudewater. The proposal is to record a dataset with the robot used in the Earth Rover Challenge in Rotterdam, in front of the Holland America Line Cruise Terminal, before the festivities of the Happy Birthday ICRA Party start. See: https://github.com/SlamMate/vSLAM-on-FrodoBots-2K

An Earth Rover dataset recorded at the ICRA@40 party

TL;DR

This work addresses the need for open-world, real-world navigation benchmarks for Earth Rover-like platforms by proposing a Rotterdam dataset collected with the FrodoBot Earth Rover Zero during the ICRA@40 event. It combines multiple sequences with front/rear cameras, IMU, GPS, and 4G connectivity, capturing at Hz and later upsampling to Hz, with depth maps produced from monocular RGB via LiteMono. Baseline demonstrations pair YOLOX-X object detection and LiteMono depth estimates with ORB-SLAM3 trajectories, revealing robustness limitations in waterfront scenes with repetitive textures and moving objects. The authors advocate extending to Scene Action Maps and cross-embodiment navigation to improve real-world SLAM and planning, arguing that the dataset will be valuable for training and evaluating autonomous navigation under open-world conditions, including public and waterfront contexts.

Abstract

The ICRA conference is celebrating its anniversary in Rotterdam in September 2024, with as highlight the Happy Birthday ICRA Party at the iconic Holland America Line Cruise Terminal. One month later the IROS conference will take place, which will include the Earth Rover Challenge. In this challenge open-world autonomous navigation models are studied truly open-world settings. As part of the Earth Rover Challenge several real-world navigation sets in several cities world-wide, like Auckland, Australia and Wuhan, China. The only dataset recorded in the Netherlands is the small village Oudewater. The proposal is to record a dataset with the robot used in the Earth Rover Challenge in Rotterdam, in front of the Holland America Line Cruise Terminal, before the festivities of the Happy Birthday ICRA Party start. See: https://github.com/SlamMate/vSLAM-on-FrodoBots-2K
Paper Structure (4 sections, 4 figures)

This paper contains 4 sections, 4 figures.

Table of Contents

  1. Introduction
  2. Dataset
  3. Situation Awareness
  4. Conclusion

Figures (4)

  • Figure 1: The FrodoBot Earth Rover Zero.
  • Figure 2: Holland America Line Cruise Terminal in Rotterdam.
  • Figure 3: The first row shows the target boxes obtained by YOLOX from the frames. The second row of images shows the dense depth maps predicted by LiteMono from a single RGB image.
  • Figure 4: A part of the trajectory plot of ORBSLAM3 during operation on the output_rides_21 sequence (see Fig. \ref{['fig:yolox']}). Squares represent individual frames, with red indicating tracking failure frames and blue indicating tracking success frames.