Working in a real robot project

External Requests
1

Hi developers,

I bough one month ago a sainsmart 6 DOF robot. I wanted to try to integrate it with a raspberry and OS to try to control it. I made the design of the pieces and then I start to implement it on Moveit. At first it was easy but the robot have a paralell joint that make a close kinematic chain in the robot. That made that I couldn’t continue working on it. I search information about it and I know more or less what are my future plans about this project. Let’s talk about them:

1-I want to implement my own inverse kinematic for the robot with the actual representation. Try to know what is the relation between the angles of the joints to forget the paralell joint for the moment. ¿How can I do this? I didn’t found nothing. I think that is something that there isn’t too much information about it, and i wanna know about it, I think this is really useful.

2- Then when I do the first step, I want to close the loop in gazebo representation. I think that this is posible in SDF.

3-Then I will continue with the implementation of the planning, but without the first step I thik that this is imposible now.

I will thank you a lot if you can give me bibliography or some material that could help me in the progress of my project. I haver search a lot about mimic joints, about kinematic plugins, about how to deal with this type of robots but with any results. I don’t know how to continue.

3

Hi,

So y take on this is:

  1. First try to simplify the system in a way that you don’t have a closed loop in a gazebo simulation. Then , you will be able to use the Moveit InversKinematics algorithms without issues.

  2. For use in closed loops, you need to create it in SDF, not in URDF of XACROS. In that case you will then have to creat your own Inverse kinematics system. Hve a look at this paper

ope this helps and we would be delighted to see a ROSject about this tpic. It will surely be very useful froa lot of other ROS users.

4

Obviusly I have simplified the robot to an open chain one. It’s in my plans to share ir with the ros comunity when have finished it. There isn’t too much documentation about this in the ROS community and I’m finding lot of problems in finding help. I tried to implementmy own kinematic solver to the robot by using some nodes that calculate and publish in joint_states topic. I don’t know if it’s the right topic where to publish, but I found that when publishing there, it’s always changing between the start pose and the desired pose that I have implemented with my node. Can you help me in that?

I read the paper that you sent me but I’m not sure that this could help me in my project, Can you explain that to me?

Thanks for all and waiting for your response