![](bg8.png)
0
50 100 150 200 250
300 350
10
1
2
3
4
5
6
7
8
9
a
b
Q(s, a)
with costs
no costs
38
◦
Figure 9: Grippoint selection incorporating costs. The tar-
get is estimated to be at 38 degrees. The influence of the
cost function on the rating function is clearly visible.
15 were not acceptable, as the result of an incorrect
pose estimation by the classifier.
4 CONCLUSION AND OUTLOOK
In this paper, we have presented the combination of
two systems, an object tracker and an object classifier,
which are able to grip a non-trivial object using only
visual feedback.
An important aspect is that neither of these systems
require any explicit modeling, neither in the behavior
of the focal length adjustment, nor in the selection of
the gripping angle. Instead, the focal length adjust-
ment comes automatically from the information theo-
retic approach, while the correct angle is trained, al-
lowing the system to generate its own model.
Future work will focus on improving the individ-
ual components of this system, motivated by the goal
of tracking and gripping a moving target. This com-
prises prediction of the target’s position multiple steps
into the future, automatic adaptation of tracking fea-
tures to cope with visually changing objects, and eval-
uation of reinforcement learning techniques which al-
low learning an optimal sequence of actions.
REFERENCES
Bar-Shalom, Y. and Fortmann, T. (1988). Tracking and
Data Association. Academic Press, Boston, San
Diego, New York.
Bertsekas, D. P. (1995). Dynamic Programming and Op-
timal Control. Athena Scientific, Belmont, Mas-
sachusetts. Volumes 1 and 2.
Bicchi, A. and Kumar, V. (2000). Robotic grasping and con-
tact: A review. In Proceedings of the 2000 IEEE In-
ternational Conference on Robotics and Automation,
volume 1, pages 348–353, San Francisco.
Borotschnig, H., Paletta, L., Prantl, M., and Pinz, A. (2000).
Appearance-based active object recognition. Image
and Vision Computing, 18(9):715–727.
Deinzer, F., Denzler, J., and Niemann, H. (2003). Viewpoint
Selection – Planning Optimal Sequences of Views for
Object Recognition. In Computer Analysis of Images
and Patterns – CAIP 2003, LNCS 2756, pages 65–73,
Heidelberg. Springer.
Denzler, J. and Brown, C. (2002). Information Theoretic
Sensor Data Selection for Active Object Recognition
and State Estimation. IEEE Transactions on Pattern
Analysis and Machine Intelligence, 24(2):145–157.
Denzler, J., Zobel, M., and Niemann, H. (2003). Informa-
tion Theoretic Focal Length Selection for Real-Time
Active 3-D Object Tracking. In International Con-
ference on Computer Vision, pages 400–407, Nice,
France. IEEE Computer Society Press.
Grzegorzek, M., Deinzer, F., Reinhold, M., Denzler, J., and
Niemann, H. (2003). How Fusion of Multiple Views
Can Improve Object Recognition in Real-World En-
vironments. In Vision, Modeling, and Visualization
2003, pages 553–560, M
¨
unchen. Aka GmbH, Berlin.
Hager, G. and Belhumeur, P. (1998). Efficient region track-
ing with parametric models of geometry and illumi-
nation. IEEE Transactions on Pattern Analysis and
Machine Intelligence, 20(10):1025–1039.
Mason, M. (2001). Mechanics of Robotic Manipulation.
MIT Press. Intelligent Robotics and Autonomous
Agents Series, ISBN 0-262-13396-2.
Paletta, L. and Pinz, A. (2000). Active Object Recogni-
tion by View Integration and Reinforcement Learning.
Robotics and Autonomous Systems, 31(1–2):71–86.
Puckelsheim, F. (1993). Optimal Design of Experiments.
Wiley Series in Probability and Mathematical Statis-
tics. John Wiley & Sons, New York.
Shannon, C. (1948). A mathematical theory of communi-
cation. The Bell System Technical Journal, 27:379–
423,623–656.
Smith, C. and Papanikolopoulos, N. (1996). Vision-guided
robotic grasping: Issues and experiments. In Proceed-
ings of the 1996 IEEE International Conference on
Robotics and Automation, pages 3203–3208.
Sutton, R. and Barto, A. (1998). Reinforcement Learning.
A Bradford Book, Cambridge, London.
Tordoff, B. and Murray, D. (2001). Reactive Zoom Control
while Tracking Using an Affine Camera. In Proceed-
ings of the 12th British Machine Vision Conference,
volume 1, pages 53–62.
Zobel, M., Denzler, J., and Niemann, H. (2002). Binocu-
lar 3-D Object Tracking with Varying Focal Lengths.
In Proceedings of the IASTED International Confer-
ence on Signal Processing, Pattern Recognition, and
Application, Crete, Greece, pages 325–330, Anaheim,
Calgary, Zurich. ACTA Press.
ICINCO 2004 - ROBOTICS AND AUTOMATION
176