Path and motion planning
INFO: The old path/motion planning functionality is still functional for backward compatibility and available if you click here, but it is recommended not to use it anymore.
V-REP offers path/motion planning functionality via a plugin wrapping the OMPL library. The plugin is courtesy of Federico Ferri.
Following points should be considered when preparing a path/motion planning task:
Above procedure is the regular approach, which sometimes lacks flexibility. Additionally, following callback functions can be set-up:
The path provided by simOMPL.compute is usually just one path among an infinite number of other possible paths, and there is no guarantee that the returned path is the optimal solution. For that reason it is common to compute several different paths, then to select the better one (e.g. the shorter one).
In a similar way, if the goal state had to be computed from a goal pose, then several goal states are usually tested, since not all goal states might be reachable or close enough (in terms of state space distance). A common practice is to first find several goal states, then order them according their state space distance to the start state. Then perform path planning calculations to the closest goal state, then to the next closest goal state, etc., until a satisfactory path was found.
Make sure to refer to following demo scenes for additional details: