FT-Control - Robotized Transcranial Magnetic Stimulation - page 96

coil using hand-assisted positioning was performed in less than one minute without
collision. Thus, hot-spot search was achieved in less than four minutes with hand-
assisted positioning.
5.3.3 Latency of Contact Pressure Control
With our setup using two robots, we have been able to measure the latency of the
pure contact pressure control. We have therefore compared the delay between start
of head motion and start of coil motion to maintain the contact to the head. We
have found that the latency for the contact pressure control is 203 ms on average.
As the robot's inertia already results in a latency of almost 100 ms [ 8 ], a latency of
approximately 200 ms is convincing and satisfactory. This shows that measure-
ment of forces and torques, computation of user applied forces and torques, and the
transfer into robot movements are done in roughly 100 ms. Subsequently, the
control software can send the stop command to the robot within roughly 100 ms
after a collision or error has occurred.
Additionally, we have estimated the maximum control frequency of the contact
pressure control. We have found that the presented control cycle runs with up to
40 Hz, which also includes the robot movements. Thus, we measure the occurrent
forces and torques roughly every 25 ms, which is suitable to detect even fast or
short-term impacts.
5.4 FT-Control in the Context of Robotized TMS
The FT-control implements two key features of robotized TMS. First, it enables
the operator to position the TMS coil in an intuitive fashion. The robot moves
accordingly to the user applied forces and torques to the TMS coil. Second, the
contact pressure control allows for an optimal coil placement on the head as it
gently moves the coil on the head. Furthermore, it maintains the contact of the coil
to the head throughout the application and monitors the forces and torques con-
tinuously to stop the robot in case of an error or collision.
We have shown that the presented method for hand-assisted positioning com-
bined with the coil calibration method is sufficient for use with different standard
TMS coils. The mean errors for forces and torques are 1.89 N and 0.31 Nm,
respectively. These errors are mostly due to the heavy TMS supply cable that is
connected to the stimulator. As the weight of the cable is approximately 0.5-1 kg
and the weight of the coil without cable is roughly 0.5-1.9 kg, depending on the
coil type, the observed results are satisfactory. Due to the flexibility of the cable,
the errors are related to the gravity compensation of the cable's weight. The
maximum errors are below the threshold for contact pressure control.
Most important, our practical test has shown that the hand-assisted positioning
method allows even unexperienced users to effectively position the coil with the