Great project

[SabineChrista]

Good Morning!

I love your project! I really appreciate the well-thought-out design and how you've built everything in small, modular components. That's such a clean and scalable approach.

It's also incredibly helpful (and rare!) to find analytical solutions for the inverse kinematics of different robot designs and your package includes several! I'm currently working with quite a few of your solutions and have even implemented some additional robots myself.

Right now, I'm trying (and still struggling a bit) to find an analytical solution for the Fanuc CRX10iA/L cobot, so I can implement it in the same style — but no luck so far.

[Preliy]

Thank you so much @SabineChrista for the kind words — really glad to hear you're finding the project helpful! 🙌

The Fanuc CRX10iA/L is definitely a tricky one with its unique geometry. I’ve also started looking into its kinematic chain to see how we might build a simple analytical solver, but it's not straightforward. If you're open to it, feel free to share your current approach or thoughts — maybe we can figure it out together or crowdsource ideas here.

[moreuser24]

Hello everyone,
what a lucky coincidence!

After having succesfully tested this great package on a standard 6-axis robot with spherical wrist, I was also trying to use it with a Fanuc CRX-10iA. This time, though, I can't even seem to be able to make the kinematic diagram. 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.

Anyway, I'd also like to contribute!

[moreuser24]

This is the list of papers on the topic I found till now:

1. Inverse Kinematics of a Class of 6R Collaborative Robots with Non-Spherical Wrist
2. Geometric Approach for Inverse Kinematics of the FANUC CRX Collaborative Robot
3. The inverse kinematics of lobster arms
4. A review of cuspidal serial and parallel manipulators
5. Path Planning and Optimization for Cuspidal 6R Manipulators
6. Cuspidal robots: theoretical study, classification and application to commercial robots, PhD thesis

[SabineChrista]

Hello!
It's been a while since I looked into this topic, but I believe I worked with the first paper you mentioned. They do present an analytical solution, but when I looked more closely, I noticed that Equation 1 still depends on Equation 2, and so on. So although it's not a numerical loop, the solution still feels somewhat sequential or dependent, rather than being fully decoupled or straightforward to generalize.

[moreuser24]

As a noob, if such an analytical solution is hard to compute, I wonder how do Fanuc engineers solve the IK problem then

[Preliy]

#39 Issue created

[moreuser24]

How can we of any help to you?

[Preliy]

A new branch https://github.com/Preliy/Flange/tree/39-fanuc_lobster_kinematic has been created for this issue. I've added the Fanuc robot geometry and set up the kinematic chain. The new Solver script is already attached to the robot, but the calculation section is currently commented out.

Feel free to implement the calculation logic in your own way. Once you're done, you can create a pull request so we can review your code together.

[Preliy]

@moreuser24 @SabineChrista Interesting problem. I found a Python solution, but need to adapt it to Unity’s use case. The approach using a candidate point and the dot product is definitely clear and makes sense.

[moreuser24]

I will also try to understand this python solution, but I'm sure you'll be faster ;)
There is also a guy on this forum who claims to have successfully implemented the solution proposed in the paper The inverse kinematics of lobster arms

[Preliy]

Its based on Geometric Approach for Inverse Kinematics of the FANUC CRX Collaborative Robot

[Preliy]

Hello, I've made some progress. I followed the exact same approach as described in the paper, and it seems to be working well. Unfortunately, the calculation is iterating through 3,600 candidate points, which impacts performance. However, I assume this will be properly handled in a separate asynchronous thread.

2025-07-02.00-02-03.mp4

[moreuser24]

Great job! Have you already committed the new code?

[Preliy]

I am working on the solution selection

[SabineChrista]

Looks great! Why are there so many iteration steps? Is it still a analytical solution?

[Preliy]

Looks great! Why are there so many iteration steps? Is it still a analytical solution?

The IK solver is still analytical, but not in closed form. A 1D search algorithm is required to find a valid combination of frames 3 and 4.

[Preliy]