Discussion - Robotized Transcranial Magnetic Stimulation - page 153

the robot's emergency stop. In this way, it acts as an independent safety layer for
the robotized TMS system.
We have shown that the required calibration of the accelerations to the force/
torque sensor coordinate frame can be done with a median error of roughly 3 : 5 .
However, there were some recordings with a larger fitting error due to noise in the
measurements. As the calibration is only required once for each FTA sensor, we
are able to repeat and extensively validate the calibration result. For instance, we
can use the fitting error to validate if the quality of the measurements is poor. In
case of noise, we will repeat the recordings to minimize the error. Therefore, it will
be possible to perform a final calibration of the FTA sensor with a calibration error
below 2 . Our evaluation further suggests that the presented calibration method
produces stable results. The median deviation between two calibration matrices
was 0 : 89 .
Beside these evaluations on the calibration itself, our practical test shows that
the gravity compensation based on accelerations is sufficient for the application of
robotized TMS. The median error was roughly 0.3 N for the force readings and
approximately 0.03-0.04 Nm for the torque readings. The maximum errors were
below 1.25 N and 0.13 Nm for forces and torques, respectively.
For the robotized TMS system, the used contact pressure is in the range of
2-4 N (cf. Sect. 5.2.3 ) . For user interaction with the robot using hand-assisted
positioning, only forces larger than 4 N and torques larger than 0.5 Nm are taken
into account to move the robot (see Sect. 5.2.2 ) . Therefore, the presented gravity
compensation is sufficient and applicable for the purpose of robotized TMS.
The most important benefit of the FTA sensor, beside its independence from
robot input, is the monitoring system in Real-time. The average maximum latency
of the FTA sensor is 1 ms as shown in our tests. In contrast, the average latency of
the robot (Adept s850) for a full emergency stop is 66 ms. Therefore, the FTA
sensor results only in an additional latency of roughly 1 : 5 %. In contrast, the
latency for the standard FT control implemented in software is in the range of
200 ms ( Sect. 5.3.3 ). The FTA sensor is therefore approximately 200 times faster
in triggering the emergency stop.
Without use of the FTA sensor or any other external emergency control, the
robot will move on until the hardware envelope of the robot is reached. This is
more than 400 N for the Adept robot as shown in our results. In the worst case this
can be a serious and dangerous situation for either the patient or the operator. For
the application of robotized TMS and most other medical robotics systems, the
robot speed is highly limited. The typical speed range is 3-10 % of the maximal
robot speed [ 10 ]. In this range, the robot will only move further for a very short
distance (less than 4 mm) after the impact when using the FTA sensor. Also, the
maximum force will be below 30 N. As the security limit was 10 N, this is an
additional force of less than 20 N. In the worst case, when the robot control is fully
lost and the robot moves with its maximum speed, the FTA sensor will ensure that
the robot stops as fast as possible to only expose patient or operator to a minimum
of force. Even at high speed, the maximum force will be less than 100 N and the
distance until the robot stops will be less than 55 mm. Even though this does not