This is an OpenRAVE wrapper to an algorithm for converting a linear joint space path into a timed trajectory with bounded acceleration and velocity. It does this by adding circular blends between consecutive linear path segments, and computing the time-optimal motion along the linear-circular-linear segments.
The resulting trajectory is then reinterpolated at a fixed discretization to return a quadratic polynomial spline approximation of the solution.
It is based on code provided by Tobias Kunz: https://github.com/tobiaskunz/trajectories
which implements the technique described in the paper:
- Tobias Kunz and Mike Stilman, Time-Optimal Trajectory Generation for Path Following with Bounded Acceleration and Velocity, RSS 2008
This algorithm takes the following parameters (with the specified default values):
<check_collision>False</check_collision>: If the blended trajectory should be re-checked for collisions<integration_step>0.001</integration_step>: The integration step size.<interpolation_step>0.01</interpolation_step>: The interpolation step size.<max_deviation>0.1</max_deviation>: The maximum allowed deviation from the original trajectory