Introduction
In all the control schemes before, the desired trajectory is given in joint space (, , ). But you often want the end effector to follow a straight line or another path in Cartesian coordinates. Control schemes that form the errors in Cartesian space are called Cartesian based control schemes. They extend Computed Torque Control.
Approach 1: trajectory conversion
Change the Cartesian path into a joint path, then use a joint space controller:
| Symbol | Meaning |
|---|---|
| desired position and orientation of the end effector in Cartesian space | |
| vector of desired joint positions | |
| inverse kinematics | |
| Jacobian matrix of the manipulator |

Approach 2: Cartesian error
Measure the error directly in Cartesian space, with forward kinematics :

Force and torque relationship
The work done must be the same in any set of coordinates. Let be a small Cartesian displacement of the end effector and a small joint displacement:
| Symbol | Meaning |
|---|---|
| vector of Cartesian forces and moments at the end effector | |
| vector of joint torques |
Since :

Cartesian computed torque control
Start from the joint space dynamics, , and the kinematics:
Pre multiply the dynamics by and use :
This gives the equation of motion in Cartesian coordinates:
| Symbol | Meaning |
|---|---|
| mass matrix in Cartesian space | |
| velocity terms in Cartesian space | |
| gravity terms in Cartesian space |
Controller
Use the same partitioning in Cartesian space:
This gives . The servo law is:
| Symbol | Meaning |
|---|---|
| input from the servo portion, in Cartesian space | |
| Cartesian error, (the same as ) | |
| , | diagonal matrices of position gains and velocity gains |
The joint torques are: