bipedal-locomotion-framework

The bipedal-locomotion-framework project is a suite of libraries for achieving bipedal locomotion on humanoid robots.

📚 Interesting Math Notes

What follows is a list of interesting math notes that can be usefull to fully understand the usage of the library

🧮 SO3 Minimum jerk trajectory

📄 Mandatory dependencies

The bipedal-locomotion-framework project is versatile and can be used to compile only some components.

The minimum required dependencies are Eigen3, iDynTree and spdlog. If you want to build the tests please remember to install Catch2. If you are interested in the python bindings generation please install python3 and pybind11 in your system.

📙 Exported components

The bipedal-locomotion-framework project consists of several components. The components are stored in the src folder and their compilation depends on the installed dependencies.

Component	Description	Additional Dependencies
`Framework`	Interface library that gathers all the exported components, includable with the file `BipedalLocomotion/Framework.h`	-
`AutoDiff`	Bridge between CppAD and Eigen	`CppAD`
`ContactModels`	Models to describe the contact between robot and enviroment	-
`Contacts`	Syntactic description of a contact	`manif` `nlohmann json`
`CommonConversions`	Common conversion utilities used in the framework	-
`ManifConversions`	Library related conversion utilities used in the framework	`manif`
`matioCppConversions`	Library related conversion utilities used in the framework	`matio-cpp`
`CasadiConversions`	Library related conversion utilities used in the framework	`CasADi`
`Estimators`	Library containing observers	-
`FloatingBaseEstimator`	Library containing floating base estimators	`manif`
`RobotDynamicsEstimator`	Library containing floating base estimators	`manif`
`GenericContainer`	Data structure similar to `span` but resizable.	-
`IK`	Inverse kinematics	`manif` `osqp-eigen`
`Math`	Library containing mathematical algorithms	`manif`
`ML`	Library containing deep learning algorithms	`onnxruntime`
`ParametersHandler`	Library for retrieving parameters from configuration files	`YARP` (only if you want the `YARP` implementation) `tomlplusplus` (only if you want the `toml` implementation)
`PerceptionInterface`	Generic interface classes to adapt to perception data formats like images and point clouds	`OpenCV` `PCL`
`PerceptionCapture`	Library containing driver classes for perception devices	`OpenCV` `PCL` `realsense2`
`PerceptionFeatures`	Library containing perception algorithms useful for locomotion	`OpenCV`
`Planners`	Library containing planner useful for locomotion	`manif` `CasADi` `qhull`
`ReducedModelControllers`	Controller based on reduced models, e.g., `CentroidalMPC`	`CasADi`
`RobotInterface`	Library for communicating with the robot	`YARP` (only if you want the `YARP` implementation)
`SimplifiedModelControllers`	Controllers based on simplified models, i.e, LIPM.	-
`System`	Description of discrete and continuous dynamical systems	-
`TSID`	Task space inverse dynamics	`manif` `lie-group-controllers`
`TextLogging`	Helper library to display messages into the console	`YARP` (only if you want the `YARP` implementation) `ros-humble` (if you want the `ros2` implementation)
`YarpUtilities`	Utilities library for retrieving data and from YARP structures	`YARP`

📦 Install with conda (recommended)

You can easily the library with conda using the following command

conda install -c robotology bipedal-locomotion-framework

conda will automatically install all the required dipendencies.

🛠️ Build the suite

The bipedal-locomotion-framework can be built on Windows, macOS, and Linux. The easiest way to compile the library is to use the robotology-superbuild. If you enable the profiles ROBOTOLOGY_ENABLE_DYNAMICS and ROBOTOLOGY_ENABLE_DYNAMICS_FULL_DEPS in the robotology-superbuild you will automatically clone and build bipedal-locomotion-framework and all the dependencies.

If you do not want to use the robotology-superbuild you can manually compile the code in the repository running the following command in the terminal

git clone https://github.com/ami-iit/bipedal-locomotion-framework.git
cd bipedal-locomotion-framework
mkdir build && cd build
cmake -DCMAKE_INSTALL_PREFIX=<path/where/you/want/to/install> \
      -DCMAKE_BUILD_TYPE=Release \
cmake --build . --config Release --target install

🖥️ Some utilities

The bipedal-locomotion-framework ships also some utilities that can help you in the everyday tests on a real robot. You can find them in the utilities folder. Each utility contains a well-documented README where you can find further details.

🐍 Python

bipedal-locomotion-framework provides also python bindings. Only a small set of the components implemented in the library have the corresponding python bindings.

If you want to compile the bindings please install python3 and pybind11 in your system then you can run the following CMake command in your build folder. You also need to install manifpy.

cmake -DCMAKE_INSTALL_PREFIX=<path/where/you/want/to/install> \
      -DCMAKE_BUILD_TYPE=Release \
      -DFRAMEWORK_COMPILE_PYTHON_BINDINGS:BOOL=ON \
      -DPython3_ROOT_DIR=$(python3 -c "import sys; print(sys.prefix)") \
      -DFRAMEWORK_USE_Python3:BOOL=ON \
      -DFRAMEWORK_USE_pybind11:BOOL=ON .
cmake --build . --config Release --target install

🏃 How to use the libraries

bipedal-locomotion-framework provides native CMake support which allows the library to be easily used in CMake projects.

bipedal-locomotion-framework exports the CMake targets presented in Exported components section. The targets can be imported using the find_package command and used by calling target_link_libraries.

For instance, Math component can be used as follows:

cmake_minimum_required(VERSION 3.0)
project(myproject)
find_package(BipedalLocomotionFramework COMPONENTS Math REQUIRED)
add_executable(example example.cpp)
target_link_libraries(example PRIVATE BipedalLocomotion::Math)

The list of the targets can be found in the table 📙 Exported components

⚙️ Contributing

The bipedal-locomotion-framework is an open-source project, and is thus built with your contributions. We strongly encourage you to open an issue with your feature request. Once the issue has been opened, you can also proceed with a pull-request 🚀

🧑‍💻 Maintainers

Giulio Romualdi (@GiulioRomualdi)
Stefano Dafarra (@S-Dafarra)