Runtime Detection of Executional Errors in Robot-Assisted Surgery

This is the code of Runtime Detection of Executional Errors in Robot-Assisted Surgery
Presented at 2022 International Conference on Robotics and Automation (ICRA 2022)
The repo includes the models (LSTM, CNN, Siamese-LSTM and Siamses-CNN) and the experimental setups(GSTS,GST*,G*TS,G*T*).

A video describing this work is available here

Install

conda install --file requirements.txt

Dataset

The error labels for the Suring/Needle Passing task from the JIGSAWS dataset can be found here. We preprocessed the data with downsampling and normalization. The preprocessed data can be found here and here.

How to run

We have 4 main scripts. The 'type' variable can be changed to 'double' or 'single' for performance evaluation on the Siamese network or the LSTM,CNN.

• GSTS.py : training with gesture specific task specific setting
• GST*.py : training with gesture specific task non-specific setting
• G*TS.py : training with gesture non-specific task specific setting
• G*T* : training with gesture non-specific task non-specific setting

The util.py contains utility functions including data loading and parameter tuning.

Contact

Please let us know if you have any questions. You can send an email to Zongyu Li ([email protected])

Citation

Bibtex

@article{li2022runtime, title={Runtime Detection of Executional Errors in Robot-Assisted Surgery}, author={Zongyu Li and Kay Hutchinson and Homa Alemzadeh}, year={2022}, booktitle={2022 IEEE International Conference on Robotics and Automation (ICRA)}}