Debridement

This contains source code for autonomated surgical debridement for our ICRA 2018 paper.

• Supplemental website with videos: https://sites.google.com/view/calib-icra/
• arXiv: https://arxiv.org/abs/1709.06668

General Structure (davinci-skeleton)

(By Sanjay Krishnan)

The base directory structure for da Vinci utilities. This repository defines the basic structure and utilities for the Da Vinci surgical robot in the AUTOLAB at UC Berkeley.

Starting up the robot

Move the robot arms so they have at least a 1cm box of clearance around the tool tip, make sure that the tool tip is out of the cannula. Open a terminal and run

roscore

Create a new tab and run

./teleop

A teleop interface should load. In the telop interface, click the radio button for Home. You will hear a nasty fan sound from the PSM1 controller, don't worry about this (Sanjay is being negligent right now). Wait until the messages stop and then run Teleop. The robot is now engaged. Do not move the arms unless the clutch is engaged.

Setting up your development environment

If this your first time on davinci0, create a directory for yourself in the home directory. Open a terminal and type

mkdir awesome_autolab_grad
cd awesome_autolab_grad

Then, create a virtual environment inside this directory

virtualenv -p /usr/bin/python --system-site-packages my-new-project

This creates a new sub directory that will contain your project. Then, run following command:

cd my-new-project && source bin/activate

You know you are successful when your terminal prompt changes to something like (my-new-project)davinci0@davinci0. NEVER install anything outside your virtual environment. Inside the project directory clone the bare skeleton repository:

git clone --bare https://github.com/BerkeleyAutomation/davinci-skeleton.git

Go to your own github and create a new repository, don't initialize it with anything. For example https://github.com/sjyk/my-new-project.git. Then, run the following commands:

cd davinci-skeleton.git
git push --mirror https://github.com/sjyk/my-new-project.git

Then, delete the skeleton repository:

cd ..
rm -rf davinci-skeleton.git/

You can now clone your own repository:

git clone https://github.com/sjyk/my-new-project.git
cd my-new-project

Commit and push whatever you want to this personal repository.

Robot API

There are basically three modules that are important (for now) dvrk.robot, config.constants, and autolab.data_collector:

from config.constants import *
from dvrk.robot import *
from autolab.data_collector import *

First, create a robot object:

psm1 = robot("PSM1")

To move the robot to a home position:

psm1.home()

To move the robot's tooltip to a new position:

import tfx

#A new position
post,rott = ((0.05, 0.02, -0.15), (0.0, 0.0,-160.0))

#creating the proper data structures
pos = [post[0], post[1], post[2]]
rot = tfx.tb_angles(rott[0], rott[1], rott[2])

#execute move with a SLERP motion planner
psm1.move_cartesian_frame_linear_interpolation(tfx.pose(pos, rot), 0.03)

Talk to Sanjay before using more advanced features of the API (all of which are in dvrk.robot).

Doing Science

Unique experimental identifiers are useful, and every time you create a DataCollector class (which arbitrates the sensor readings) it creates a new one for you.

d = DataCollector()
print(d.identifier)
>>> 80PTSJF68HLISWKQY1ZL

This can be used to make a directory in the results folder:

import os
try:
    os.stat('results/'+d.identifier)
except:
    os.mkdir('results/'+d.identifier)

This folder is a good place to put all of your experimental data.

If you run the program scripts/runExp.sh it will execute the main program and commit the config, parameters, and results to github:

(my-new-project)davinci0@davinci0:~/davinci-skeleton$ bash scripts/runExp.sh 
PID of running experiment  9666
Thu Jul 6 12:04:24 PDT 2017 Still working...log in: 0 /tmp/exp.Y8cir8gwG
Thu Jul 6 12:04:25 PDT 2017 Still working...log in: 0 /tmp/exp.Y8cir8gwG

It logs all standard error and out to a temporary directory and prints out the PID (really useful if something goes wrong and you want to kill). The cool part is that even if you kill the python program the bash script will commit whatever intermediate results to github so you won't lose them!

Citation

If you find the code or the paper useful, please consider citing:

@inproceedings{seita_icra_2018,
    author = {Daniel Seita and Sanjay Krishnan and Roy Fox and Stephen McKinley and John Canny and Kenneth Goldberg},
    title = {{Fast and Reliable Autonomous Surgical Debridement with Cable-Driven Robots Using a Two-Phase Calibration Procedure}},
    Booktitle = {IEEE International Conference on Robotics and Automation (ICRA)},
    Year = {2018}
}