3D Soft Trunk repository

SoPrA fabrication guide

check the Wiki of this repo for details of fabrication process, links to STL files, etc.

Installing this repository

Set up SSH key authentication for GitHub, then in the desired directory, run...

git clone --recursive [email protected]:srl-ethz/3d-soft-trunk.git

(--recursive option will automatically clone the submodules as well)

Install necessary packages for mobilerack-interface

Refer to README of mobilerack-interface

Also install these packages:

sudo apt install python3-pip
pip3 install xacro # used to convert robot model files from XACRO to URDF

Install Drake

Refer to Drake documentation- binary installation and Drake sample CMake project.

For Ubuntu 20.04, basic steps are:

curl -O https://drake-packages.csail.mit.edu/drake/nightly/drake-latest-focal.tar.gz
## decompress and place drake files into /opt/drake
sudo tar -xvzf drake-latest-focal.tar.gz -C /opt
## install prerequisites
sudo /opt/drake/share/drake/setup/install_prereqs

For macOS, .... TODO

Compile

cd /path/to/3d-soft-trunk
cmake -DCMAKE_PREFIX_PATH=/opt/drake .
make

Executables are output to bin, libraries are output to lib/. For visualization of the model, use the Meshcat Visualizer (Drake visualizer has been deprecated). Launch /opt/drake/bin/meldis (which relays LCM connections from Drake to the Meshcat visualization), and go to the URL shown (probably http://localhost:7000) in your browser.

Python interface

In its current implementation, you must set the $PYTHONPATH environment variable to point to the directory containing the library binaries in order to run. (probably 3d-soft-trunk/lib)

## run this everytime you open a new terminal to run a python script using this library
export PYTHONPATH=$PYTHONPATH:/path/to/lib
## Alternatively, append the line to ~/.bashrc if you don't want to run it every time.
python3
>> import mobilerack_pybind_module
>> vc = mobilerack_pybind_module.ValveController("192.168.0.100", [0, 1], 200)
>> vc.setSinglePressure(0, 100)

>> import softtrunk_pybind_module
>> st_params = softtrunk_pybind_module.SoftTrunkParameters()
>> st_params.finalize()
>> stm = softtrunk_pybind_module.SoftTrunkModel(st_params)
>> stm.getModel()

see more examples in examples_python/ and mobilerack-interface/examples_python.

Generating Documentation

Uses Doxygen to generate documentation from inline comments in code. Install Doxygen, and run doxygen Doxyfile in this directory to generate HTML (can be seen with browser at html/index.html) & LATEX output.

Example code are in apps/example_*.cpp and examples_python/example_*.py.

I want to add my own sensor/controller/model

Refer to the README files in src/.

I am not a member of SRL

You will likely be using different sensors and actuators. Refer to the wiki for how to add new sensors to the StateEstimator. To add your own actuation method, you will need to add a submodule class which allows you to set inputs for the actuators, and replace the ValveController object in the ControllerPCC class. Then you simply need to adjust the ControllerPCC::actuate(VectorXd p) to use your new actuator instead, and the framework should work.