chhinze/panda_tutorial

Tutorial to get familiar with the Franka Emika Panda and `libfranka`.

---

title: 'Guide: Get the Franka Emika Panda running with C++' author: Christoph Hinze date: \today institute: IRTG Soft Tissue Robotics (GRK2198/1) ...

Preliminaries:

1. The tutorial will run on Linux, explanations below are for Ubuntu. If you haven't already installed Ubuntu on your Computer, or enabled the Windows Sybsystem on Linux (WSL), you should install it, following the tutorial on https://docs.microsoft.com/en-us/windows/wsl/install-win10. As distribution install e.g. Ubuntu 18.04. This will enable you to run a virtual Ubuntu terminal on Windows 10.

2. Install Git on Windows (https://git-scm.com/downloads), or directly in the WSL (sudo apt install git)

3. Install requirements for building the C++ project that we will create throughout this tutorial. We need libfranka (unfortunately not directly available through sources) and libeigen3-dev

   mkdir -p ~/dev && cd ~/dev
   libfranka_version="0.6.0"
   
   # install Eigen3 and libfranka 
   sudo apt update && sudo apt install --yes build-essential cmake git libpoco-dev libeigen3-dev
   
   # from here: build libfranka dependency from source code and install it:
   git clone --recursive https://github.com/frankaemika/libfranka.git
   cd libfranka
   
   git checkout ${libfranka_version}
   git submodule update
   
   mkdir build && cd build
   cmake -DCMAKE_BUILD_TYPE=Release -DBUILD_TESTS=OFF -DBUILD_EXAMPLES=OFF ..
   make -j4
   sudo make install

4. A manual on the Franka Emika Panda and libfranka is at https://frankaemika.github.io/docs/getting_started.html

5. A documentation for libfranka can be found at https://frankaemika.github.io/libfranka/

6. We will use Eigen3 as math library, where an introduction may be found at http://eigen.tuxfamily.org/dox/group__QuickRefPage.html

Theory

• How the path planning is done

Exercises:

The exercises without solution are directly available on the master branch. The solution is under branch solution.

Tutorial Project

Get the tutorial from Github with

# Drive C: on Windows (with WSL) is available with /mnt/c/..., so you might want to `cd` to this directory
git clone https://github.com/chhinze/panda_tutorial

Ex. 1: Build the demo project and its documentation

Install the requirements for building the documentation

# --no-install-recommends needed to prevent installing doxygen-latex
sudo apt install --no-install-recommends doxygen graphviz

The project can be built with cmake with GNU make. I.e.

cd ~/dev/panda_tutorial
mkdir build 
cd build

# build the makefiles
cmake ..

# build the executables from C++ code
make -j4

# build the documentation with doxygen (needs doxygen installed)
make doc

The binaries are generated within the ./build/.... You can find the documentation at ./build/documentation/html/index.html

Ex. 2: Math operations with Eigen3

1. Create two Eigen::Vector3d objects
   • One with all entries 10 and
   • one with [0.1, 0.5, 0.7];
2. Create two Eigen::Matrix3d objects:
   • An Identity matrix
   • A Matrix with

     | 1, 2, 3 |
     | 4, 5, 6 |
     | 7, 8, 9 |
     

3. Try, how to do vector-vector matrix-vector and matrix-matrix multiplications
4. Find out, how to perform element-wise operations. (E.g. find the row-wise maximum of a matrix, multiply element-wise).
5. Access the first two elements of a vector. Write new values to it.

Ex. 3: Drive the robot on axes level

1. Have a look at the MotionGenerator, which is originally provided by Franka Emika for their code samples. It is included as "examples_common.h"
2. Define a goal position as std::array<double,7> for the axes. The joint positions should be [0, -pi/4, 0, -3/4*pi, 0, pi/2, pi/4] (0, -M_PI_4, 0, -3 * M_PI_4, 0, M_PI_2, M_PI_4). Create a MotionGenerator object with this goal position.
3. Set the robot controller to axis position control by just passing the MotionGenerator object. Refer to the libfranka documentation for different control modes.
4. Build and run the application (make -j4 axes_motion). Be careful, the robot will be moved. Be prepared to push the user stop button.

Ex. 4: Drive the robot in a linear trajectory

1. Refer to the documentation to find out how to read a franka::RobotState. Read it once to get the sart position.
2. O_T_EE_c contains the homogeneous transformation matrix as array with 16 values. Convert it to a 6d vector (3 translations, 3 RPY rotations) by using the function homogeneousTfArray2PoseVec(std::array<doublem 16>) -> Eigen::Vector6d
3. Calculate a 6d goal pose by adding 0.1 m to each of the translational coordinates.
4. Create a LinearTrajectory between start an end pose. Use v_max = 0.05, a_max = 0.5 and j_max = 1e-3.
5. Create a TrajectoryIteratorCartesianVelocity object. It overloads the function call operator, such that it can directly be used in franka::Robot.control(...).
6. Specifiy the robot controller (panda.control(...);)
7. Build and run the application (make -j4 cartesian_trajectory). Be careful, the robot will be moved. Be prepared to push the user stop button.