Why is DH convention still used in Robotics?

Question

I do not know if this is a correct question for the main portal, hence asking here.

I am aware that DH convention is the first and the important thing that is taught in robotics. But it has drawbacks. For example, defining the axes properly.

There is an another intuitive and easy(at least in my opinion) method that uses twist vectors to define the forward, inverse kinematics , and also forward and inverse dynamics (Recursive Newton-Euler method).

It seems like a lot of industries still uses DH convention. So my question is why isn't the twist vector method adopted as much as DH convention?

Answer 1

This question has been tackled on our site a few times:

• Why do we generally prefer DH parameters over other kinematic representations of robot arms?
• What are the advantages of using the Denavit-Hartenberg representation?

Without rephrasing everything in the previous questions, I have a few thoughts:

DH convention is the first and the important thing that is taught in robotics.

This is probably because of it's relative simplicity. Allowing students to quickly go from arm dimensions to forward kinematics. It may also simply be due to its historic nature.

There is an another intuitive and easy method...

Yes, in fact there are probably several other methods. Each with their own pros and cons.

It seems like a lot of industries still uses DH convention.

I'm not so sure about this actually. I don't know what representation industrial arms use. It is quite possible they use something else internally, even if they publish the DH parameters externally. (I especially believe this since my efforts in kinematic calibration). I'd be very interested if someone from ABB, Kuka, UR, etc. could chime in. But I doubt they are too willing to give away the secret sauce.