Between which joints does a given DH table row represent?

Question

Say for example

                          __
         _____           |  |          ____
--------|_____|----------|  |---------|____
                         |__|

That ASCII art is supposed to represent a "rolling" revolute joint followed by a "Yawing" revolute joint, followed by the tool frame.

Would the rows in the DH table represent

BASE-J1
J1-J2

Or would they represent

J1-J2
J2-Tool Frame

score 1 · Accepted Answer · answered Nov 06 '17 at 02:45

tl;dr The rows (I guess you refer a row to a set of 4 DH-parameters) represent, roughly speaking, every link. So it's BASE-J1, J1-J2, and J2-Tool Frame.

When deriving DH parameters, the base frame usually situates at the robot base, where it is mounted on to floor. This DH convention is a convention that tells you how you put reference frames on a robot. How many frames you need (apart from the base frame) depends on how many joints there are. This corresponds to how many links you have.

So for a 2 degrees-of-freedom (DOF) robot as shown, you would have 3 sets of DH parameters. Each set, consisting of link length, link twist, link offset, and joint angle, describes each link.

I asked my professor and he said something similar, albeit he said don't give the tool frame if not asked for it. — DeepDeadpool, Nov 06 '17 at 16:32

score 0 · Answer 2 · answered Nov 06 '17 at 08:47

There are a few DH convention out there, with main difference the way to chose the first and last frame to consider. Usually the emphasis is put on minimal representation, i.e., minimal number of frame. In my opinion this often leads to parametrization which are a bit more obscure to the user.

The most common DH parametrization would recommend to consider J1-J2 and J2-tool. Because in the end J1 is fixed in a frame, whether in the global frame if the arm is rigidly attached to the environment or a mobile frame if the arm is mounted on a mobile robot. In both cases the base frame of the arm is often irrelevant. Personally I don't agree that much with this approach and would consider an extra frame at the base o the robot, because by experience it's is more intuitive for debug and inverse kinematic. In particular the inverse kinematic is then conducted considering the base frame and not the first joint frame, which I think improves the intuitiveness, e.g. the arm is fixed on a table and should grasp an object on the table you can check easily that the base frame end the end-effector have the same z coordinate, if you consider the first frame in J1 you need to do an additional translation and this get more 'fun' when you attach your arm to the environment with funny configuration like this

Between which joints does a given DH table row represent?

2 Answers2