Doubt on DH parameters

[moreuser24]

I was trying to build from the ground up the Fanuc model. I started from an empty gameobject, added the Robot6RLobsterWirst script and used the context menu option "Create Default Hierarchy" to create the joints hierarchy.

My problem is the modified DH parameters I input in each joint's Frame script public fields give me a wrong kinematic chain. I looked into a few papers for the DH parameters but each one has different values for the alpha and theta angles. Moreover, it seems the classical DH convention is used in the majority of the papers, not the modified one which this package is based upon, as stated in the README.

Did you worked out the DH parameters by hand or followed a specific paper? Did you tweak something to get the correct kinematic chain? Can you confirm the Modified Denavit-Hartenberg convention for assigning the frames? (i-th z axis along i-th joint axis)

I'd like to understand it because I need to model an ABB GoFa CRB15000 and use the forward kinematics of your package but I'm facing the same issue in setting up the kinematic chain as this robot too has neither a spherical wrist nor a structure you already defined. I was only able to get the correct chain with a standard ABB spherical wrist robot.

Also, if I don't need the inverse kinematics solution, can I use any of the solver script, just to setup the kinematic chain and then use the MechanicalGroup.SetJoints from the Controller script?

Thank you in advance

[Preliy]

Yes, you're right — the modified DH convention can be confusing, especially since most papers and examples use the classical DH method instead. I followed this guide when figuring it out: https://linuxcnc.org/docs/html/motion/dh-parameters.html

Here's how I usually set up a robot:

1. I start with an empty GameObject and use Create Default Hierarchy to build the joint structure.
2. If modified DH parameters are available, I use them directly.
3. If not, I set A and D first, then adjust alpha (α) and theta (θ) until the skeleton gizmos align with the robot’s geometry. That’s usually the easiest way to get it right.

The goal is to make sure the skeleton gizmos are properly aligned with the actual geometry. You might need to tweak the values a bit to get it right.

Yes, the package uses the Modified DH convention, but with some caveats. The rotation axis is along Z (which maps to Unity’s Y axis) from the previous frame. This can be confusing, especially because of the transformation between left-handed and right-handed coordinate systems.

To be honest, I’m also questioning if this is the best approach — it definitely needs a deeper investigation to be sure everything is consistent.

By the way, I’ve created a prefab for the ABB GoFa robot, and it works fine.
And yes — you can definitely use the package for forward kinematics only. Just skip the IK solver and use MechanicalGroup.SetJoints() from the Controller script to set joint values directly.

2025-07-03.21-49-57.mp4

ABB CRB15000_10kg.zip

[Preliy]

DH Parameters

	Theta	D	Alpha	A
1	0	0.4	-1.570796	0.15
2	-1.570796	0	0	0.707
3	0	0	-1.570796	0.11
4	0	0.6375	1.570796	0
5	0	0	-1.570796	0.08
6	3.141593	0.101	0	0

[moreuser24]

Hello, thank you for the clarification. I will try to follow your suggestions should I have the need to model another robot.

However, I also tried to attach the frames according to another DH convention and surprisingly I got the correct kinematic chain! Is it true that there is confusion about the multiple DH conventions for affixing frames and when I read Modified I thought of that as described in Craig textbook, with the i-th z axis along i-th joint axis. Instead, I then applied the Standard one, as described in Siciliano textbook, with i-th z axis along i+1 joint axis, which worked better for the ABB GoFa with 5 Kg payload. The parameters as very similar to the ones in the table above