Skip to content

This library implements a model predictive control algorithm to generate walk trajectories with automatic foot step placement.

Notifications You must be signed in to change notification settings

aldebaran/mpc-walkgen

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

mpc-walkgen

This software provides...

Setup

Dependencies

The package mpc-walkgen depends on several packages which have to be available on your machine.

  • System tools:

    • CMake (>=2.8)
    • pkg-config
    • usual compilation tools (GCC/G++, make, etc.)
  • Libraries

    • boost (>=1.40)
    • eigen (>=3.0.0)
    • qpoases (version 2.0, compiled twice for both float and double, with special care taken to avoid symbol collisions)

Compile

The project is installed and built thanks to qibuild (https://github.com/aldebaran/qibuild).

Once you have qibuild installed, simply run

qibuild configure
qibuild make

Design choices

API

At some stage, the lib might be split in two in order to separate the humanoid and zebulon parts (the public and currently secret parts). The common pieces used by these two might go in a third "common" library.

In this spirit it was decided that most of the header could be kept public.

Template instantiations

Just like Eigen, mpc-walkgen is templated by a scalar type. However unlike Eigen, mpc-walkgen is not a head-only library. Its algorithms are not implemented in headers the user includes, but in a shared library the user links with.

This means:

  • the algorithms are not implemented in (public) headers but in (private) cpp files.

  • the templates are not instantiated in the user code but in the mpc-walkgen library (there are explicit instantiation for float and double in each cpp files)

  • the user code will be compiled faster (no need to instantiate mpc-walkgen templates in it)

  • the user code can be compile in debug while using mpc-walkgen compiled in release

  • the user cannot extend mpc-walkgen to support scalar types besides float and double. (Note that we currently have no QP solver for other scalar types anyway.)

About

This library implements a model predictive control algorithm to generate walk trajectories with automatic foot step placement.

Resources

Stars

Watchers

Forks

Packages

No packages published