ROS Resources: Documentation | Support | Discussion Forum | Index | Service Status | ros @ Robotics Stack Exchange
Ask Your Question

Planning multiple sequential motion in MoveIt RViz

asked 2021-09-03 05:26:59 -0500

Rahutam gravatar image

I am using universal robot 10 (Ubuntu 18 - Melodic) in MoveIt RViz, and would like to test different planners with motion planning. I can however only execute 1 move with command roslaunch moveit_ros_benchmarks demo_ur10.launch at once (it will generate a different file for each move). Is there a way I can plan a motion of multiple sequential points and then moving the robot arm from A to B to C to D etc. and generate a file for this entire motion?

There is an answer on a question ( "you can use the plan method to get the path that will be executed, without actually executing it yet. You can then plan the next path by setting the start to be the goal state of the previous motion with setStartState." Is there a way you can do this is RViz only or is there an external code you need to implement in order for this to work?

Furthermore, is there an option to randomly generate 3 out of X predefined points and then move through there points? So if I have let's say "1000 runs" defined in my .yaml file, the robot would generate 3 different points each run and then generate a file I could upload to Planner Arena to benchmark the planners?

Thank you for any answers.

edit retag flag offensive close merge delete

1 Answer

Sort by ยป oldest newest most voted

answered 2021-09-03 09:10:15 -0500

fvd gravatar image

You can still do that in Rviz, you just lose the previous plan when you calculate the next one. Only one trajectory is stored in the Rviz MotionPlanning plugin. If you want to store a sequence of them, you will need to write your own code.

I don't know of a ready-made test of the kind you describe, but you could write your own test case and try to use moveit_benchmarks to benchmark it locally. But it sounds unnecessary, since the three points would just be three independent planning requests, and you can get plenty of performance data on planners for single planning requests.

edit flag offensive delete link more


Thanks! By "single planning requests" you mean moving the robot from start state to a single point and that's enough to give good and representative information about a specific planner? Wouldn't benchmarking a sequence of points for the robot to move to and fro be more representative for efficiency and performance of each planner?

Rahutam gravatar image Rahutam  ( 2021-09-06 02:49:12 -0500 )edit

Question Tools



Asked: 2021-09-03 05:26:59 -0500

Seen: 386 times

Last updated: Sep 03 '21