Applications to Safe Human–Robot Interaction - 3D Computer Vision: Efficient Methods and Applications - page 293

Fig. 7.9 Overview of the
fusion stage

Fig. 7.10 Convergence
behaviour of the
ICP-MOCCD fusion stage.
The first column illustrates
the ICP results for each
iteration and the second
column the MOCCD results.
In the last column the fusion
result is shown for each
iteration

The two pose updates T ICP and T MOCCD of the pose estimation algorithms are
then used together with the calculated weighting factor to compute the fusion pose
parameter vector T fusion according to

w ICP
w ICP +

w MOCCD
w ICP +

T fusion = T init + T ICP + T MOCCD . (7.3)

w MOCCD w MOCCD

This procedure is performed several times using the fusion result as a new ini-
tialisation for the next pose estimation stages, until the pose parameters converge
(cf. Fig. 7.10 ).