agx_dual_arm

Dual arm control using Algoryx simulator AGX

Control the dual arm in AGX simulation

Dependencies

The main dependencies are the following ones:

• AGX simulator: Working AGX urdf-application environnement is required
• AGX custom project scene for IIWA: Custom AGX scene (https://github.com/epfl-lasa/agx_scene)
• dual_iiwa_toolkit: Optional, if the build of the passive controller is needed: The python binding of dual_iiwa_toolkit (https://github.com/epfl-lasa/dual_iiwa_toolkit)

File hierarchy

The file system is divided in several subfolders:

• builded_libs: contains the cpp library and the python binding of the passive controller
• config: contains .yaml and _ LearnedModel_ used by the controller
• urdf: contains .urdf of robot used by the controller
• script:
  • python_agx_dual_arm: Contains the control loop

Setup environment

Tested with python 3.8.10. It is recommanded to create a venv

1. cd agx_dual_arm && pip install -r requirements_agx.txt
2. Optional: for the keyboard control: sudo pip3 install keyboard
3. Add the cpp lib and python bindings to you path

echo 'export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/path_to_iam_dual_arm_control/dual_arm_control/python_binding/build/' >> ~/.bashrc
echo 'export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/path_to_iam_dual_arm_control/agx_dual_arm/builded_libs' >> ~/.bashrc 

. ~/.bashrc

sudo ldconfig

If the libraries of i_am_project or iiwa_toolkit need to be build:

cd python_binding && mkdir build && cd build && cmake .. && make -j

Run simulation

1. Launch AGX simulator with the custom project scene for the double iiwa

sudo python3 ../run-in-docker.py python3 click_application.py --model models/Projects/agx_scene/Scenes/DualIiwaClickSceneTable.yml:IiwaTorqueClick --timeStep 0.001 --agxOnly --rcs --portRange 5656 5658  --disableClickSync

2. Run the controller:

 python script/python_agx_dual_arm.py

With the keyboard commands (sudo rights required)

 sudo venv/bin/python3.8 script/python_agx_dual_arm.py