Unicycle-like Robots with Eye-in-Hand Monocular Cameras: From PBVS towards IBVS - Visual Servoing via Advanced Numerical Methods - page 337

(a) (b)

Fig. 18.6 Map navigation: (a) desired image; and (b) initial image

Fig. 18.7 Desired images from each topological map node on the top row. Images grabbed
from the camera after each visually servoed path, on the bottom row

web-cams. Although even low cost cameras ensure about 20 to 30 frames per sec-
ond, the rate changes depending on ambient illumination variations.
Figure 18.5 reports an image graph created during the topological mapping phase.
The visual servoing controller has been used to travel the distance between the
mapped images, parking the vehicle in the position x T (position I 3 in the topolog-
ical map, see Figure 18.6(a)). The initial robot position is unknown, but the archi-
tecture solves the kidnapped robot problem identifying the topological position x T
(see Figure 18.6(b)). Hence, the visual servoing path corresponds to a travel between
image node I 5 to I 3 .
In Figure 18.7, the nodes crossed by the robot during the parking are represented.
In the top row, the images stored in the topological map ( i.e. desired images for the
visual servoing) are represented, while in the bottom row, the images grabbed from
the camera after each path are depicted. Approaching an intermediate node ends
once it is possible to localize and to track features of the next node to be reached.
A wide movement in the mapped environment comprises several limited move-
ments between each pair of images (Figure 18.7). Nevertheless, the visual servoed
motion between successive images is still quite small. Indeed, it is well known in