# Posts by Hey Hey

• ## Problem with solution (equation) for wrist singularity

In my textbook about industrial robots it says that if the angle q5 in this robot is equal to 0, there are infinite solutions for the angles q4 and q6. I understand this part, but I have a problem with the equation below.

As a solution for this the book says f(q4,q6) = c1*(q4A - q4)^2 + c2*(q6A-q6)^2 = MIN

q4A and q6A are the joint angles calculated from the previous backward transformation when driving a robot path.

By the above equation, a solution for q4 and q6 is taken which is closest to the previous solution.

Between two IK solutions (intermediate points) on a line that the TCP must travel, Q4 must rotate from -pi/2 to -3/2*pi, and Q6 must rotate from pi/2 to -pi/2. Both joints must therefore cover a rotation of -pi.

So the formula would be: c1*(-pi/2 - (-3/2*pi))^2 + c2*(pi/2-(-pi/2))^2

Since I do not know c1 and c2 and the book does not tell me how to determine them, I cannot calculate a solution. How do I calculate the angles for Q4 and Q6?

• ## How to deal with a joint suddenly having to make a large change in angle during linear interpolation?

HawkME Oh ok, thanks for your reply. I now know that the problem is a singularity occurs.

In my textbook it says that if the angle q5 in this robot is equal to 0, there are infinite solutions for the angles q4 and q6.

As a solution for this the book says

f(q4,q6) = c1*(q4A - q4)^2 + c2*(q6A-q6)^2 = MIN

q4A and q6A are the joint angles calculated from the previous backward transformation when driving a robot path. By the above equation, a solution for q4 and q6 is taken which is closest to the previous solution.

I dont understand this equation.

• ## How to deal with a joint suddenly having to make a large change in angle during linear interpolation?

After having set up the inverse kinematics for each intermediate point on a line that the tcp is supposed to travel, I encountered a problem that stops the tcp when it is supposed to travel along the line. While all other joints do not have to make big changes in the joint angles between the intermediate points, a joint has to rotate from -1 rad to 1 rad between two intermediate points, which is a very big rotation compared to the other joints. I understand that all joints have to make a different amount of angle change between two intermediate points, but the joint that has to do the large angle change cannot rotate as fast to compensate for this.

Should the other joints wait until the rotation of the joint is finished and then continue with the interpolation? This seems to be the only reasonable solution for me. However, this would mean a short interruption of the movement of the tcp on the line.

Therefore, I am confused by tasks that require the tcp to move at a constant speed on a line

• ## What parts do I need to buy in order to build a small industrial-like robot?

I want to build and program the robot as far as possible in the same way as it is done in industry. The robot should have six servo motors for a free choice of the orientation of the effector. Which parts should I buy for this? Should I buy some small servo motors and 3D-print the arm parts and program the controller with an AVR chip? If so, which servo motors should I buy, how should I wire the servo motors, etc? And I don't want to do it with Arduino at all, because I want to learn how something like this is done in the industry.

• ## Is the Cyclic Coordinate Descent algorithm suitable for industrial robots?

Also, which algorithms are suitable for industrial robots?