DiTer (Diverse Terrain and Multi-Modal) Datasets
Our project page is renewaled. Please visit the new DiTer site.
HILL01 HILL02 FOREST LAWN
DiTer(Diverse Terrain and Multi-Modal) Datasets
Construction robots require autonomy in diverse environments to navigate and map their surroundings efficiently.
However, the lack of diverse and comprehensive datasets hinders the evaluation and development of autonomous construction robots.
We hope that our dataset support research that is designed to traverse terrain where the wheeled platform is difficult to navigate.
Navigation Sequences
All sequences are obtained within the outdoor sites on the campus.
Since each sequence takes two laps around the same path, we also hope to contribute to the study of place recognition or SLAM.
HILL01-A,B
HILL01 is mainly composed of a gravel road.
HILL02-A,B
HILL02 focuses on poor environments including under-canopy. In this sequence, difficult situations such as the robot stepping on a high grass or hitting it directly are included.
FOREST-A,B
FOREST is a terrain where trees are sporadically planted. People are continuously appearing in this sequence.
LAWN-A,B
LAWN is obtained in a rectangular park with hills scattered. In this sequence, the robot travels around with the shape of an infinite symbol (∞) for two laps using the stones in the center as a base point.
News!
[24.01.09] Our paper is accepted to IEEE Sensors Letters!
[23.11.20] We submit the extended version of the dataset to IEEE Sensor Letters.
[23.08.03] We submit the extended version of the dataset to IEEE journal.
[23.08.03] Our short paper is the under-review state in the IEEE Sensors Letters.
[23.08.03] We plan to add each sequence in the night, respectively.