Joint torque sensor vs end-effector force sensor

Question

I am trying to understand the difference between joint torque sensors, and force sensors, in a robot arm. From what I understand, joint torque sensors estimate the torque at each joint, whereas a force sensor estimates the joint at the end-effector.

But isn't it true that you can use the joint torque sensors to estimate the force at the end-effector, using forward kinematics? And if so, then why would you need to use a force sensor, if you already have joint torque sensors? Or would you typically only have one, i.e. only a set of joint torque sensors, or only a single force sensor?

The reason I am questioning this is that at https://blog.robotiq.com/robotiq-ebook-summary-force-sensors-in-robotics-research, it says:

Force sensors are designed to detect forces applied between their base and sensing plate

But then at https://studywolf.wordpress.com/2013/09/02/robot-control-jacobians-velocity-and-force/, it says:

Jacobians allow us a direct way to calculate what the control signal is in the space that we control (torques), given a control signal in one we don’t (end-effector forces)

Therefore, it seems to me that joint torque sensors and force sensors do the same thing, and you would never need both of them. Is this correct? Or do they play complimentary roles?

Answer 1

If everything were ideal, then your thinking is correct. In practice, though, there are many accumulated errors both in the sensors and in the real vs modelled Jacobian. So if what you care about most is force applied at the end-effector then it is much more accurate to have a sensor located at the end effector. Unfortunately that would also add weight, and possible be more expensive, so for practical reasons you may only have joint torque sensors even though it's less accurate.