jchen396 commited on
Commit
5a059c0
·
verified ·
1 Parent(s): 914b65d

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +86 -3
README.md CHANGED
@@ -1,3 +1,86 @@
1
- ---
2
- license: cc-by-4.0
3
- ---
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ license: cc-by-4.0
3
+ language:
4
+ - en
5
+ tags:
6
+ - medical
7
+ ---
8
+ # SutureBot: A Precision Framework and Benchmark for Autonomous End-to-End Suturing
9
+
10
+ This dataset is part of the **SutureBot** project, a benchmark for developing and evaluating autonomous surgical robotic policies for end-to-end suturing. It contains high-resolution, multi-camera video and robot kinematics from real robot demonstrations on tissue phantoms, supporting research in imitation learning, vision-language-action modeling, and surgical autonomy.
11
+
12
+ ## Dataset Summary
13
+
14
+ The dataset includes:
15
+
16
+ - **Stereo endoscope images** (`left_img_dir`, `right_img_dir`)
17
+ - **Wrist camera views** (`endo_psm1`, `endo_psm2`)
18
+ - **Low-level kinematics** (end-effector pose, joint angles, jaw state, etc.) in `ee_csv.csv`
19
+ - **Multiple tissues** and **multiple task types** (e.g., needle pick-up, suturing)
20
+ - All data is organized by:
21
+ tissue_<id>/<task_name>/<episode_timestamp>/
22
+
23
+
24
+ Each episode is a complete trajectory from a real robotic execution using a da Vinci Research Kit (dVRK) platform.
25
+
26
+ ---
27
+
28
+ ## Usage
29
+
30
+ This dataset is designed for training and evaluating autonomous policies that take as input multi-view RGB images (optionally language-conditioned), and output continuous control actions.
31
+
32
+ It supports a wide range of applications including:
33
+ - Behavioral cloning and offline RL
34
+ - Multi-modal policy learning (images + kinematics + language)
35
+ - Visual servoing and tool tracking
36
+ - Surgical robot benchmarking
37
+
38
+ ---
39
+
40
+ ## File Structure
41
+
42
+ Each episode contains:
43
+ - `left_img_dir/`: Left endoscope camera images
44
+ - `right_img_dir/`: Right endoscope camera images
45
+ - `endo_psm1/`: Right wrist (PSM1) camera images
46
+ - `endo_psm2/`: Left wrist (PSM2) camera images
47
+ - `ee_csv.csv`: Kinematic data including joint states, poses, and jaw angles
48
+
49
+ ---
50
+
51
+ ## Format
52
+
53
+ - Image format: `.jpg`, RGB
54
+ - Kinematics: CSV with columns such as:
55
+ - `timestamp`
56
+ - `psm1_pose.position.{x,y,z}`
57
+ - `psm1_pose.orientation.{x,y,z,w}`
58
+ - `psm1_js[0-5]`, `psm2_js[0-5]`, etc.
59
+ - `ecm_pose.*`, `suj_pose.*`, etc.
60
+
61
+ A complete list of columns is available in the `dataset.croissant.json` schema.
62
+
63
+ ---
64
+
65
+ ## Citation
66
+
67
+ If you use this dataset, please cite:
68
+
69
+ ```bibtex
70
+ @misc{suturebot2025,
71
+ title = {SutureBot: A Precision Framework and Benchmark for Autonomous End-to-End Suturing},
72
+ author = {Juo-Tung Chen and others},
73
+ year = {2025},
74
+ note = {Under review at NeurIPS 2025 Datasets and Benchmarks Track},
75
+ howpublished = {\url{https://huggingface.co/datasets/YOUR_ORG/suturebot}}
76
+ }
77
+ ```
78
+ ## License
79
+ This dataset is licensed under CC-BY-4.0. Please credit the authors when using this dataset.
80
+
81
+ ## Contact
82
+ For questions or collaboration inquiries, please contact:
83
+
84
+ Juo-Tung Chen
85
+ Johns Hopkins University
86
+ 📧 jchen396@jh.edu