From: Mark M. <mm...@ri...> - 2021-08-05 18:30:55
|
Hi Cemal, On Aug 4, 2021, at 2:57 PM, E. Cemal Gonen <emr...@sa...> wrote: > While we define our problem in constrained motion planning, we defining the geometric constraints as mathematical equations, and the sampler creates samples based on the constrained methodology we chose. Is there a way to define some "soft constraints" meaning constraints that are preferred? For example, consider a 2 dof manipulator on the plane, and how can we define our preference on elbow configuration of the manipulator. Although it can be set by validity, this eliminates the other elbow configuration. I wonder if there is a proper way to do this. For soft constraints you can use the cost infrastructure in OMPL to define the cost for states/motions. The cost is defined using a OptimizationObjective-derived class. There are lots of examples predefined. See this page <https://ompl.kavrakilab.org/optimalPlanning.html> for details. Best, Mark |