What is the standard way of getting the signed distance and Jacobian matrix (Trajopt)?
I am not sure if it is the right place to ask for help.
I am trying to implement discrete time collision term for Trajopt planner, which I forked from @Omid's work.
I am currently using distanceRobot
and distanceSelf
of CollisionEnvFCL
to get the signed distance which is necessary for Trajopt. However, I am not sure if it works properly, since I sometimes get huge collision distance for the the panda robot (larger than 6 meters). Also, robot_state.getJacobian(...)
sometimes returns nan
values, but dont know the reason for that.
Is this the standard way of getting the signed distance and Jacobian matrix?
Related distance calculation part:
https://github.com/adam-vonderviszt/moveit/blob/b18b8c66963907aa6d03cbee4450fc3d6e740162/moveit_planners/trajopt/trajopt/src/collision_terms.cpp#L110-L122
Related Jacobian calculation part:
https://github.com/adam-vonderviszt/moveit/blob/b18b8c66963907aa6d03cbee4450fc3d6e740162/moveit_planners/trajopt/trajopt/src/collision_terms.cpp#L48-L50
Asked by blabla on 2019-12-18 21:42:51 UTC
Comments
hi , did you able to solve this? if yes , can you explain please.
Asked by prajwal98 on 2022-05-10 13:24:41 UTC