Home

A Ravelin

[Installation instructions](Installation instructions)

What is Ravelin?

Ravelin is a C++ library for rigid body and multi-rigid body kinematics and dynamics.

Reference points for rigid bodies need not be collocated with centers-of-mass
Composite Rigid Body Algorithm and Featherstone Articulated Body Algorithm implementations
Rigid body and multi-rigid body kinematics and dynamics computation, including poses, velocities, accelerations, inertias, momenta, forces, and Jacobians
Both explicit and implicit joint constraints supported: dynamics computable in minimal coordinates and absolute coordinates
Floating bases, tree-like kinematic topologies, and mechanisms with kinematic loops supported
Frame checking: operations on kinematic and dynamic quantities defined in different frames throw exceptions
Supports reading in multi-rigid body kinematics and dynamics from URDF
Results from all dynamics algorithms have been verified against Featherstone's MATLAB-based code

Ravelin wraps BLAS and LAPACK in place of Eigen.

Arithmetic with zero-dimensional matrices/vectors supported
Matrix/vector arithmetic and linear algebra through very reliable BLAS and LAPACK libraries
Efficient block operations on matrices and vectors (similar to Eigen's segment() and block() functions)
Minimum templated C++ code for fast compilation
Iterators, which provides compatibility with STL and Boost algorithms
Straightforward semantics on memory allocation toward programming for real-time systems

You should use Ravelin if one or more of the following applies to you:

You are computing robot kinematics or dynamics
You want to reduce programming errors by catching frame (pose) mismatches
You want to use the most robust matrix/vector arithmetic and linear algebraic implementation available
You are programming for real-time systems (like control loops) and need to ensure that memory is not allocated during your real-time code

Ravelin supports floating bases and kinematic loops (KDL does not)
Ravelin's dynamics computations are complete (KDL's are not)
KDL contains methods for trajectory formation (Ravelin does not)
KDL contains various methods for inverse kinematics using resolved motion rate control (Ravelin does not implement IK, though it provides the Jacobians necessary to do so)
KDL uses Eigen (Ravelin wraps BLAS/LAPACK)