Skip to content

ctu-vras/adaptive-skin

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

2 Commits
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Adaptive Electronic Skin Sensitivity for Safe Human-Robot Interaction

This repository contains code for the article:

Rustler, L.; Misar, M. & Hoffmann, M. (2024), Adaptive Electronic Skin Sensitivity for Safe Human-Robot Interaction, in 'IEEE-RAS International Conference on Humanoid Robots (Humanoids)', IEEE.

Preprint available at arxiv.

Video is available at https://youtu.be/kWt2J1Qgzpk.

Maintainer: Lukas Rustler, lukas.rustler@fel.cvut.cz

Contents

Licensing

The code created in this work is under GNU-3 licence.
For licensing information about other used repositories see: LICENSE, and the original repositories:

Installation

  • clone this repository

    cd SOME_PATH
    git clone https://github.com/ctu-vras/adaptive-skin.git adaptive_skin_ws
    
  • use Docker (see Docker Installation or install ROS noetic (+ all necessary libraries; see Dockerfile) and python3.8 (with pybullet, open3d and their dependencies)

  • if using Docker

    cd adaptive_skin_ws/Docker
    `./deploy.py -c adaptive_skin -p SOME_PATH/adaptive_skin_ws -b`
    
    • the above will build the Docker image, rename the container to 'adaptive_skin' and run it. You can also:
      • ./deploy.py -c adaptive_skin -e to run already built container
      • ./deploy.py -c adaptive_skin -t to open new terminal in the container
      • see easy-docker for more deploy.py options
  • build the workspace

    cd adaptive_skin_ws
    catkin config --extend /opt/ros/noetic --init
    catkin build 
    
  • source the workspace

    source devel/setup.bash
    

Docker Installation

  • Works only on GNU/Linux systems!
  • install docker-engine
  • DO NOT INSTALL DOCKER-DESKTOP!
    • docker-desktop is not the same as docker-engine, and it does not work the same way
  • do not forget about post-installation steps
  • (optional) install nvidia-docker for GPU support

For more information, more commands and FAQ see easy-docker

Code

This section will talk mainly about the experiment code. If you want to learn more specifically only about the simulation, please look into README in ur10e_simulator.

The main package with experiment code is airskin_pain with the following files:

  • main.py - main file that contains the task
  • tester.py - script that reads config files and activates the skin (simulation only)
  • utils.py - file with utility files (dynamics computation, airskin processing, etc.)
  • do_exps.py - script that runs all necessary script to run the experiment
  • airskin_node.py - ROS node that reads the skin data and publishes touch events
  • eff_mass.py - script that computes effective mass of the robot and publishes thresholds

Run

  1. The easiest way how to run your own experiments is to use do_exps.py - run rosrun airskin_pain do_exps.py -s SETUP -c CONFIGS -e EXP_NAME -i ITERATIONS, where

    • SETUP can "real" or "sim"
      • default: "sim"
    • CONFIGS is the path to the config file/directory relative to configs directory
    • EXP_NAME is the name of the experiment. This will be used for folder name in exps directory
      • you can omit this argument and the script will name the folder with current timestamp
    • ITERATIONS is the number of iterations to run the experiment for each config
      • default: 10 - you can further specify in the command:
    • --save_bag to also save .bag file with all the data
      • default: not saving
    • --offscreen to run the experiments offscreen
      • default: RVIZ is shown with visualization
  2. you can run everything by hand. Use the following commands and check parameters of each script in the corresponding file - run roslaunch airskin_pain main.launch to run ROS + start drivers - rosrun airskin_pain main.py to run the movement - rosrun airskin_pain airskin_node.py to run airskin events - rosrun airskin_pain tester.py to run the testing script - rosrun airskin_pain eff_mass.py to run the effective mass computation, if used

Real setup

The code is working the same for both simulation and real setup. However, to use it in real world you need the HW and also drivers for the HW. We can not provide the drivers here.

Drivers for robot are available at Universal_Robots_ROS_Driver

Drivers for AIRSKIN are not publicly available (and you need the skin + special access to the HW).

Data

All data from the experiments are stored at OSF.io. If you want to run the same experiments as us, please download the directory 'exps' and place it in the data directory.

About

No description, website, or topics provided.

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published