Biomedical Engineering Reference
In-Depth Information
marker M is visible. Consequently, we can determine whether the robot and/or
tracking camera have moved in near real-time (in less than 200 ms) using the
position of the marker M, the predetermined transform
S
3
T
M
, and the position of
the robot's joints 1-4. This allows for quick adaptation of the robot/camera cali-
bration transform or, if the changes are too large, for an emergency shutdown of
the system. Furthermore, the new method also dramatically reduces the initial
setup time for the system. If the constant transform between the marker and the
robot's third link is known, initial calibration will only require one single mea-
surement of both the robot's position and the position of the marker M.
As a marker at the robot's base would be occluded during treatment (see
Fig.
4.2
), the marker on link three, however, can be assumed to be always visible
when the robot is operating in its ''elbow-up'' configuration. In case the marker on
link three is not visible for the tracking system, the tracking system is not posi-
tioned optimally as tracking the marker on the patient's head will be also difficult.
In the following sections, we present robust robot/camera calibration in detail
and evaluate its accuracy compared to the QR24 algorithm [
8
] which has been
used so far, as well as to the standard hand-eye calibration method proposed by
Tsai and Lenz [
25
]. But first, we will address the problem of robot/world cali-
bration, often named Hand-Eye Calibration, in some more detail.
Fig. 4.1 Mobile setups of the robotized TMS system. a Adept robot with mounted TMS coil and
Polaris tracking system. The robot is mounted on a pallet and a tripod supports the tracking
system for easy system assembly; b setup of the SmartMove
TM
by ANT (Advanced Neuro
Technology B.V., Enschede, The Netherlands). A Polaris Vicra tracking system is used on a
tripod and an Adept Viper s650 robot is mounted to a steel cart for stability and mobility
Search WWH ::
Custom Search