F4bich's profile - activity

Hi,

I'm using a simple 4 DoF arm and want to use MoveIt! to calculate and execute trajectories for it. I followed the setup tutorial and the Move Group Python Interface Tutorial which allow me to give the pr2 a pose goal to which he then plans a trajectory.

However, when I'm using this approach on my own model it doesn't work. I tried various Cartesian coordinates as pose goal, some that he is definitively able to reach, but I always receive:

Fail: ABORTED: No motion plan found. No execution attempted.

and

RRTConnect: Unable to sample any valid states for goal tree

I even (unsuccessfully) tried to move to the current position using:

pose_target = group.get_current_pose()
group.set_pose_target(pose_target)

The second application from the tutorial of planning to a joint-space goal is working correctly.

The planning group I use contains all 4 joints of the arm nothing else (I also tried a group containing all joints and links). I also added an end-effector in the setup, but it didn't help.

This problem is annoying me for some time already, every hint or solution is welcomed :)

Hi,

I'm trying to implement a robot model into MoveIt!. URDF-file and meshes have been build by my colleagues.

The robot can be viewed using RViz. However, when I'm loading the URDF into the MoveIt Setup Assistent such an error is given:

[ INFO] [1472814379.176343132]: Loaded complete robot model.
[ INFO] [1472814379.176441870]: Setting Param Server with Robot Description
[ INFO] [1472814379.184765149]: Robot semantic model successfully loaded.
[ INFO] [1472814379.184804317]: Setting Param Server with Robot Semantic Description
[ INFO] [1472814379.200211356]: Loading robot model 'complete'...
[ INFO] [1472814379.200256287]: No root joint specified. Assuming fixed joint
================================================================================REQUIRED process [moveit_setup_assistant-2] has died!
process has died [pid 24116, exit code -11, cmd /opt/ros/indigo/lib/moveit_setup_assistant/moveit_setup_assistant __name:=moveit_setup_assistant __log:= 
[...]

The problem seems to be caused by certain .stl meshes that are used. These meshes can be used as collision link elements, but not as visual link elements. If these specific meshes are not loaded the error doesn't occur.

Has someone experienced something similar? What might cause this problem that only MoveIt! seems to have with these .stl files.

Hi there!

I implemented a small program to test for a sequence of positions if a solution for inverse kinematics exists. I used some of the code from the tutorial. Now, I would like to set the GoalOrientationTolerance and GoalPositionTolerance as it can be done in the move_group API, however it seems like I only can set the number of attempts and the timeout.

Is there a way to initiate inverse kinematics computation with different tolerances? And what is the default tolerance?

My thanks, F4bich

Do you know if it is possible to use this through the MoveIt! API?

[MoveIt!] Goal tolerance for inverse kinematics Hi there! I implemented a small program to test for a sequence of posit

Currently I can't run sudo apt-get install ros-indigo-moveit nor sudo apt-get install ros-indigo-moveit-full. I tested it on multiple mashines. Is this still a problem with the official public repo or is it something on my side?

Is it possible to do this with the python api aswell? I don't know how to do the step

moveit_msgs::Constraints c = kinematic_constraints::constructGoalConstraints(
    tool_link_name, stamped,pos_constraint,orient_constraint);

with python

As far as I understand these are constraints one can limit the possible paths with. Therefore, an OrientationConstraint would only solve the problem, when the scenario is something like moving in a straight line, otherwise the constraint has a high chance of preventing us to find any path.

Hi,

is there a way to plan a path to a goal which is defined with a position (x, y, z) and a value for pitch and roll, but leave yaw open for variation? For my use case the yaw angle is not relevant, so I don't want to restrict possible paths by defining a value for it.

As far as I understand not giving a value for the axis will not work, because the value is initialized to zero.

Another approach would be to set the goal_orientation_tolerance. However, this is quite unsatisfying as this sets the tolerance for the whole pose and not only one specific axis.

Using the robot_state doesn't work for me. I used these functions:

state = robot.get_current_state()
group.set_start_state(state)

If the state is set like that, and then a path is planned somewhere else, the path will start in the state it was before setting it to something else.

Hi guys,

is it possible to set the pose (e.g. described as a list of joint angles) of the robot to certain values without planning a path, using the python move_group interface? Possibly there doesn't exist a path from my current pose A to the desired start pose B, but I want to find a path from B to C.

I think the 'set_start_state' function is supposed to provide this; however, I don't know how the start point is supposed to be specified (as msg).

Yes, no problem

Not yet, I used the 4DoF robot only for testing purposes and now moved to a more complex robot.

As gvdhoorn suggested, I reduced the complexity of the meshes (using Blender) , which was the solution to the problem.

As gvdhoorn pointed out correctly, the IK solver had problems with the < 6 DoF robot. Using trac_ik didn't help, however, adding more DoF did.

Hello there,

I wanted to know if there is any easy way of saving a plan object to disk or write it as a string.

E.g. you calculated a motion plan for your robot and the next time you load the simulation the same plan should be used for execution.

Couldn't find anything about that.