Inverse Kinematic changes between CoppeliaSim 4.4 and 4.6 in Hand Guidance model

[mnini]

Hi everyone,

I am currently working on a Hand Guidance model in CoppeliaSim 4.4, that uses a robot with an handle on its EE, and a maneqquin that represents the operator. What I want to do is to simulate how the maneqquin responds to the robot movement.

Basically I control the robot joint positions, and the maneqquin hand follows a target positioned on the EE handle.
I want the robot to follow a motion profile, so i use the sim.setJointTargetPosition function.

Everything was working fine, then for other reasons I had to upgrade my virtual machine and I installed the new version of CoppeliaSim 4.6, and now the scene is broken. The maneqquin movement are very slow and they seem to block the movement of the robot (but they shouldn't, considering that the robot is following a target positions, and they are not 'connected' to the maneqquin hand).

I would like to understand why the behaviour changes in the handling of the Inverse Kinematic for the maneqquin, considering that everything works fine if I use sim.setJointPosition, instead of sim.setJointTargetPosition. But I need the latter.

To better understand my error I will leave the CoppeliaSim 4.4 Scene: https://drive.google.com/file/d/1Q5J27_ ... sp=sharing

Thank you in advance!

[coppelia]

Hello,

you are right, there is a bug. Will be fixed in V4.7 rev2, out today or tomorrow.

Cheers

[mnini]

Hi,

Thank you very much. I just tried the new version (4.7 rev2) and now it moves. But there is something still bugging me.

If you notice in the scene the maneqquin is positioned on top of a moving base, so it can adapt better to the movement of the robot. In coppelia 4.4 the IK of the maneqquin is computed starting from the base, but in the new version the base is not moving, and the maneqquin stand still with only its arms moving.

I did some testing, and it seems that the prismatic joint is not considered in the IK chain, conversely if I substitute it with a revolut joint it works. I think there is still a bug that affects only the prismatic joints!

Thanks again in advance!
Cheers

-Matteo Nini

[coppelia]

The IK routines have greatly improved since V4.4, and so, it behaves (normally) in a more logical and appropriate manner. Regarding the mobile base, it won't move to the left because its range is limited in that direction (its range is -0.5 to 0.0). If you extend the range of that prismatic joint to -0.5 to 0.5 (i.e. you'll have to enter 1.0 for range), then the base joint will also be involved in IK.

Cheers