In this repository, different representations are used to generate new smaller datasets based on the CRL-Dataset-CTCR-Pose. Code and derived datasets will be provided.
We use a dataset captured from a three-tube concentric tube continuum robot for use in learning-based kinematics research. The dataset consists of 100 000 joint configurations and the corresponding four 6 dof sensors in SE(3) measured with an electromagnetic tracking system. The dataset has been collected in eight sequences. Each sequence encompasses 12 500 dataset points. A dataset point consists all 6 dof of Cartesian space SE(3) described by the singularity free quaternion/vector-pairs for each sensor pose, all 6 dof of joints space Q, and the difference in joint space configuration to the previous configuration.
CRL-Dataset-CTCR-Pose is a lightweight dataset and has a size of 56.5 MB. It is provided as a CVS file. A dataset point consists:
- six absolute joint values
- six relative joint values
- pose of the base
- pose of the proximal sensor attached to the outermost tube
- pose of the sensor attached to the middle tube
- pose of the distal sensor attached to the most inner tube
The annotation of the dataset is provided in table below.
For more details on the dataset, our paper provide implementation details on the data acquisition and a brief overview of the used testbed including the robotic prototype. Furthermore, insights on learning the kinematics of this type of robot and a discussion on open challenges are provided. If you want to cite our CRL-Dataset-CTCR-Pose, you can use our IROS 2022 paper (Preprint):
@inproceedings{GrassmannBurgner-Kahrs_et_al_RSS_WS_2022,
title = {A Dataset and Benchmark for Learning the Kinematics of Concentric Tube Continuum Robots},
author = {Grassmann, Reinhard M. and Chen, Ryan Zeyuan and Liang, Nan and Burgner-Kahrs, Jessica},
url = {https://openreview.net/pdf?id=DW9uz_GZ0og},
year = {2022},
booktitle = {Robotics: Science and Systems -- Workshop on Learning from Diverse, Offline Data (L-DOD)},
}Concentric tube continuum robots (CTCR) are a class of continuum robots introduced in 2006. A CTCR consists of multiple nested tubes that are concentric, pre-curved, and super-elastic. To generate a motion by changing the centerline of the nested tubes, each tube can be rotated and translated as shown in this video. The kinematics of a CTCR is characterized by the highly non-linear behavior of the elastic interaction between the tubes.
CTCR are proposed and introduced simulatneously by
- Webster et al. Toward Active Cannulas: Miniature Snake-Like Surgical Robots
- Sears and Dupont A Steerable Needle Technology Using Curved Concentric Tubes
More details on CTCR, ckeck the following review paper
- Gilbert et al. Concentric Tube Robots: The State of the Art and Future Directions
- Mahoney et al. A review of concentric tube robots: modeling, control, design, planning, and sensing
- Mitros et al. From Theoretical Work to Clinical Translation: Progress in Concentric Tube Robots
An attempt to define a CTCR and some details on retrospective on the development of CTCR prototypes can be found in our RSS workshop paper
- Grassmann et al. CTCR Prototype Development: An Obstacle in the Research Community?
Take a look at the following paper using learning-based approaches. They are presented in chronological order.
- Bergeles et al. (HSMR 2015) Concentric Tube Robot Kinematics using Neural Networks
- Fagogenis et al. (IROS 2016) Adaptive Nonparametric Kinematic Modeling of Concentric Tube Robots
- Grassmann et al. (IROS 2018) Learning the Forward and Inverse Kinematics of a 6-DOF Concentric Tube Continuum Robot in SE(3)
- Grassmann and Burgner-Kahrs (RSS 2019) On the Merits of Joint Space and Orientation Representations in Learning the Forward Kinematics in SE(3)
- Kunzt et al. (TMRB 2020) Learning the Complete Shape of Concentric Tube Robots
- Iyengar et al. (IJCARS 2020) Investigating Exploration for Deep Reinforcement Learning of Concentric Tube Robot Control
- Donat et al. (TMRB) Estimating Tip Contact Forces for Concentric Tube Continuum Robots Based on Backbone Deflection
- Liang et al. (ICRA 2021) Learning-based Inverse Kinematics from Shape as Input for Concentric Tube Continuum Robots
- Iyengar and Stoyanov (ICRA 2021) Deep Reinforcement Learning for Concentric Tube Robot Control with a Goal-Based Curriculum
- Grassmann et al. (IROS 2022) A Dataset and Benchmark for Learning the Kinematics of Concentric Tube Continuum Robots (Preprint)
Note that the list is not exhaustive and will change as new publications appear. If you want to add your publication, shoot me a message.