Goal tolerance and trajectory repeteability with MoveIt on real robots
While planning pose target goals and testing the repeteability of the trajectory (Planning and executing trajectories with the same pose goal) its very common to get different plans and trajectories. As far as I've tested for the same goal usually gets up to 3-5 different trajectories. So my first question is:
¿Is it possible to get always the same behaviour when planning pose goals (Without saving the trajectories, planning them at runtime)?
The next question has also some relation with the first one. When executing plans obtained from pose goals in simulation everything works as expected. On the other hand, when executing them in real robots the goal tolerance its much higher. Experimenting with the real robots I usually get up to 5 different trajectories when planning and executing poses, as I described earlier. Some of the trajectories get fine to the goal itself with unappreciable errors, but for some reason other trajectories diverge up to 2mm from the goal and always behave this way (For instance, each trajectory always get to the same goal, some of them have no error but others always differs the same amount from the goal). I've tried to tighen MoveIt tolerances, wich seems to have no effect at all. After some research I've found some older posts https://groups.google.com/g/moveit-us... wich seem to have the same problem and relate it to the robots drivers. I've also tried to lower the tolerances in the driver unsuccesfully.
¿Has someone manage to solve this situation (I dont know if its a specific UR problem or affects all robots)?
I'm working with a UR3e with the ur_robot_driver
Planning the trajectories with RTTConnect
ROS noetic on Ubuntu 20.04
Quick comment: please try to avoid including more than a single question in a post. ROS Answers is not a forum, it's a Q&A site, similar to Stack Overflow.
Those work best with a 1-to-1 ratio of questions-to-answers.
Your "related question" has almost 0 visibility, and it's difficult to answer them both properly in a single answer.
Thanks for your comment I'll take it into account for the next time, in this particular case as both questions have a close relation I thought that it could work. Anyways, the main question is the second one, about the tolerance matter. In the link shared I saw that you participated back then, ¿Did you manage to solve the situation in the end?