for Robot Artificial Inteligence

5. Inverse Kinematics

|

Inverse Kinematics

1. General Derivation(起源)

  • Not possible for serial Kinematics
  • Must be derived(获得) with geometrical intuition
  • Not always possible to find in closed-form
  • Numerical solutions sometimes required

2. Non-Linear Porblem

  • Example: TCP moves linearly along axis z
  • Joints movements are highly non-linear

2.1 Non-unique solution

  • Same TCP pose can be reached with different joints configurations
    • to solve this, we give singularities
  • infinite joints solution exist for one single TCP pose to reduce
  • Dependent linear value

3. Inverse Kinematics Steps

  1. Decouple wrist from arm
    • find out that the position of the wrist point only depends on the first three joint axes of the robot
    • moving the fourth, fifth, sixth joint will not modify the position of the wrist point
    • so we can split into two part (wrist, arm) it called Decoupling

3. Base Frame and Tool

4. Adding a tool

5. Couple

  • output of inverse kinematics is position of real joints

  • They must be adjusted to calculate correct motor on target positions

Comments