Hello everyone,
I'm using MoveIt! with ROS Kinetic on Ubuntu 16.04 Xenial to plan the motion of a climbing legged robot. I want to try using sample-based planners to produce a feasible complete joint path/trajectory. To do that I need to create a custom state sampler, because in my application I can't rely on rejection-sampling to find suitable new states.
I have read the [MoveIt! custom constraint sampler tutorial page](http://docs.ros.org/kinetic/api/moveit_tutorials/html/doc/custom_constraint_samplers/custom_constraint_samplers_tutorial.html), but I guess it is still incomplete, because I can't find the
> ROBOT_moveit_plugins/ROBOT_moveit_constraint_sampler_plugin
template that they talk about.
I have also taken a look at the [MoveIt! planning with approximated constraint manifolds](http://docs.ros.org/kinetic/api/moveit_tutorials/html/doc/planning_with_approximated_constraint_manifolds/planning_with_approximated_constraint_manifolds_tutorial.html) but, from what I understand, that is not what I want.
I wish to program in C++. I am not unfamiliar with the OMPL library (so [such a thing](https://ompl.kavrakilab.org/samplers.html) doesn't scare me), but I'm very far from being an expert. My guess is that I need to repeat the procedure provided in the last link, and somehow make it so MoveIt! knows to use this sampler instead of the default one.
So my question is: how can I implement a custom sampler to use in the MoveIt! framework? I say "custom" in the sense that my sampler will use a specific method to produce valid new states without having to rely extensively on rejection-sampling.
Any input is appreciated.
Best,
Maxens
↧