Amp provides a user-friendly API for loading robot models and generating robot joint and Tool Center Point (TCP) trajectories for point-to-point motion planning problems. Amp provides standalone trajectory parameterization and global search algorithms. Robot kinematic and dynamic algorithms are not a part of the library, but can be loaded at run-time allowing the user flexibility to suit their needs. The library includes general modules for:

• Robot trajectory and state representation
  • Using Eigen matrix objects http://eigen.tuxfamily.org
• Trajectory parameterization with general BSplines
  • Supports linear fitting
  • Supports quadratic minimization,
• Interface to various robot kinematics and dynamics libraries:
  • RBDL https://bitbucket.org/rbdl/rbdl
  • Orocos KDL http://www.orocos.org/kdl
• Forward and Inverse kinematics trajectory generation
  • Where IK solutions are computed numerically
• Sampling-based motion planning (RRT, RRT*)
  • Using a simple list-based tree structure and kd-tree nearest-neighbor search: https://github.com/jtsiomb/kdtree
  • In any Tool Center Point (TCP) space
  • In the robot's joint space
  • With bound constraints on states and inputs,
  • and collision constraints using the Bullet Collision Library http://bulletphysics.org

The code is developed by Samantha Stoneman [email protected] and Roberto Lampariello [email protected] at the Institue of Robotics and Mechatronics http://www.dlr.de/rmc/rm/en at the German Aerospace Center.

Amp requires at least C++11 and is dependent on several other open-source libraries. The required dependencies are:

• Eigen http://eigen.tuxfamily.org (Core, Linear Algebra and Geometry modules)
• Boost http://www.boost.org/ (odeint library)
• kdtree https://github.com/jtsiomb/kdtree
• EigenQuadProg

And the optional dependencies:

• RBDL https://bitbucket.org/rbdl/rbdl (rbdl, urdf/luamodel libraries)
• Orocos KDL http://www.orocos.org/kdl
• Bullet Collision Library http://bulletphysics.org (Collision library)

Amp will try to find the depencies using the native CMake find_package() tool as well as with the PkgConfig module. Valid paths to the required dependencies and the desired optional ones should be set in the CMAKE_PREFIX_PATH. Static versions of the kdtree and EigenQuadProg (header only) are also provided in the contrib dir.

Amp is built using CMake. From the package root directory, the simplest way to build amp using the default build options.

mkdir build
cd build && cmake ../ 
make 
make install

This will install the amp shared objects and binaries in lib and bin directories directly in the package root directory.

Amp comes with several build options which affect what is built and how. The standard cmake variable CMAKE_INSTALL_PREFIX allows changing the location of the install dirs lib, bin, and incude. Amp can be built in Debug, RelWithDebInfo or full Release mode with CMAKE_BUILD_TYPE.

All of the unique amp build options are detailed in the root CMakeLists.txt. The most important of which are the options which turn on and off compilation of code sections:

• DEBUG - allows computations which are not strictly necessary e.g. matrix determinants
• VERBOSE - streams verbose output to std::cerr during runtime
• LOG_ALL_DATA - writes intermediate data to files during runtime

The default is always OFF, setting these options to ON will greatly affect computation time, but can be helpful for debugging purposes.

The inverse kinematics -based RRT algorithm was tested on a difficult point-to-point TCP motion which could not be solved in one trajectory. The following table shows the performance results for several parameter settings. This benchmark problem is described in more detail in the Examples section.

A sequence of examples are provided, increasing in complexity to demonstrate the usage of the library.

• A simple example for loading a robot model and environment from a .urdf file(s) and accessing model data members using the C++ API.
• An example demonstrating the inverse kinematics- based trajectory generation.
• An example demonstrating the use of the rrt planner module in a robot TCP space.

The library is published under the GNU Lesser General Public License (LGPL). Here is the full license text:

Amp - Articulated-robot Motion Planner library. Copyright 2017 Samantha Stoneman Research Associate, German Aerospace Center (DLR)

Amp is free software: you can redistribute it and/or modify it under the terms of the GNU Lesser General Public License as published by the Free Software Foundation, either version 3 of the License, or (at your option) any later version.

Amp is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU Lesser General Public License for more details.

You should have received a copy of the GNU Lesser General Public License along with Amp. If not, see http://www.gnu.org/licenses/.