2012

2011

2010

2009

2008

2007

 Heerink, M., Kröse, B.J.A., Wielinga, B.J. & Evers, V. (2007). Observing conversational expressiveness of elderly users interacting with a robot and screen agent. In Proceedings of the International Conference on Rehabilitation Robotics . pages 154-157, Amsterdam: ACM.

 Heerink, M., Kröse, B.J.A., Wielinga, B.J. & Evers, V. (2007). iCat in Eldercare. In C Bartneck & T Kanda (Eds.), Proceedings of the 2nd ACM/IEEE International Conference on Human-Robot Interaction (pp. 177-184). Washington DC.

  Kasteren, T.L.M. van & and Ben J. A. Kröse (2007). Bayesian activity recognition in residence for elderly IE'07: Proceedings of the third international Intelligent Environments conference.

  Kasteren, T.L.M. van & and Ben J. A. Kröse (2007). Context awareness in residences for elders IEEE Pervasive Computing, 6(1) 59-60.

 Kasteren, T.L.M. van, Kröse, B.J.A. & Cemgil, A.T. (2007). Realtime Simultaneous Tempo Tracking and Rhythm Quantization in Music. In Demo in BNAIC 2007: The 19th Belgian-Dutch Conference on Artificial Intelligence (pp. 431-432).

 Kröse, B.J.A., Booij, O. & Zivkovic, Z. (2007). A geometrically constrained image similarity measure for visual mapping, localization and navigation. In Proceedings of the 3rd European Conference on Mobile Robots (pp. 168-174). Freiburg, Germany.

 Mensink, T., Kröse, B.J.A. & Zajdel, W.P. (2007). Distributed Appearance Based Tracking using the EM algorithm. In Proceedings of the 2007 First ACM/IEEE International Conference on Distributed Smart Cameras (pp. 178-184). Vienna, Austria: IEEE.

 Noulas, A. & Kröse, B.J.A. (2007). Learning in Multi-Modal Information Streams. In Proceedings of the 19th Belgian-Dutch Conference on Artificial Intelligence 2007 (pp. 245-252). Utrecht, The Netherlands.

 Noulas, A. & Kröse, B.J.A. (2007). On-line Multi-Modal Speaker Diarization. In Proceedings of International Conference on Multimodal Interfaces '07 (pp. 350-358). Nagoya, Japan.

 Noulas, A., Vlassis, N. & Kröse, B.J.A. (2007). Cross Entropy for learning in Multi-Modal Streams. In Proceeding of the Joint Workshop on MultiModal Interaction and Related Machine Learning Algorithms '07 . Brno, Czech Republic.

 Terwijn, B. & Noulas, A. (2007). BNAIC Demo: Online Speaker Detection by the iCat Robot. In BNAIC 2007: The 19th Belgian-Dutch Conference on Artificial Intelligence (pp. 451-452).

  Z. Zivkovic and Ben J. A. Kröse (2007). Part based people detection using 2D range data and images in: IEEE/RSJ International Conference on Intelligent Robots and Systems

 Zivkovic, Z. & Kröse, B.J.A. (2007). Part Based People Detection on a Mobile Robot. In Proceedings of IEEE ICRA2007 Workshop: From features to actions .

  Z. Zivkovic, O. Booij , and Ben J. A. Kröse (2007). From images to rooms Robotic and Autonomous Systems, vol.55, no.5, pages 411-418, 2007

2006

 Heerink, M., Kröse, B.J.A., Wielinga, B.J. & Evers, V. (2006). Studying the acceptance of a robotic agent by elderly users. International Journal of Assistive Robotics and Mechatronics, 7(3), 25-35.

  Wojciech Zajdel, A. Taylan Cemgil and Ben J. A. Kröse (2006). Dynamic Bayesian Networks for Visual Surveillance with Distributed Cameras in: Smart Sensing and Context 240-243.

 Heerink, M., Kröse, B.J.A., Wielinga, B.J., & Evers, V. (2006). Studying the acceptance of a robotic agent by elderly users International Journal of Assistive Robotics and Mechatronics, 7(3), 25-35.

Booij, O., Zivkovic, Z., & Kröse, B.J.A. (2006). From sensors to rooms. In Proc. IROS Workshop From Sensors to Human Spatial Concepts (pp. 53-58). IEEE.

Booij, O., Zivkovic, Z., & Kröse, B.J.A. (2006). Sparse appearance based modeling for robot localization. In Proc. IEEE/RSJ Int. Conf. on Intelligent Robots and Systems (pp. 1510-1515). ieee.

Spexard, T., Li, S., Wrede, B., Fritsch, J., Sagerer, G., Booij, O., Zivkovic, Z., Terwijn, B., & Kröse, B.J.A. (2006). BIRON, where are you? - Enabling a robot to learn new places in a real home environment by integrating spoken dialog and visual localization. In Proc. IEEE/RSJ Int. Conf. on Intelligent Robots and Systems (pp. 934-940). ieee.

M. Heerink, B.J.A. Kröse, B.J. Wielinga, and V. Evers. Human-robot user studies in eldercare: Lessons learned. In Proc. Int. Conf. on Smart Homes and Health Telematics, Belfast, Northern Ireland, June 2006(pp. 31-38)

Heerink, M., Kröse, B.J.A., Wielinga, B.J., & Evers, V. (2006). The Influence of a Robot's Social Abilities on Acceptance by Elderly Users In In Proceedings RO-MAN (pp. 521-526). Hertfordshire.

K. L. Koay, Z. Zivkovic, B. Kröse, K. Dautenhahn, M. L. Walters, N. R. Otero, and A. Alissandrakis. Methodological issues of annotating vision sensor data using subjects' own judgement of comfort in a robot human following experiment. In IEEE International Symposium on Robot and Human Interactive Communication, to appear, 2006.

Z. Zivkovic, B. Bakker, and B. Kröse. Hierarchical map building and planning based on graph partitioning. In IEEE International Conference on Robotics and Automation, pages 803-809, 2006. (PDF, 391 Kbytes)

2005

B. Bakker, Z. Zivkovic, and B.J.A. Kröse. Hierarchical dynamic programming for robot path planning. In Proc. IEEE/RSJ International Conference on Intelligent Robots and Systems, pages 3720-3725, 2005. (PDF, 243 Kbytes)

O. Booij, Z. Zivkovic, and B. Kröse. Pruning the image set for appearance based robot localization. In Proceedings of the Annual Conference of the Advanced School for Computing and Imaging, pages 57-64, June 2005. (PDF, 276 Kbytes)

A. T. Cemgil, W. Zajdel, and B. Kröse. A hybrid graphical model for robust feature extraction from video. In C. Schmid, S. Soatto, and C. Tomasi, editors, IEEE Computer Vision and Pattern Recognition (CVPR), pages 1158-1165, San Diego, June 2005. (PDF, 300 Kbytes)

G. Klaassen, W. Zajdel, and B.J.A. Kröse. Speech-based localization of multiple persons for an interface robot. In Proc. of IEEE Int. Conference on Computational Intelligence in Robotics and Automation (CIRA2005), pages 47-52, 2005. (PDF, 657 Kbytes)

B.J.A. Kröse. Digital life: de toegevoegde waarde van ICT in onze leefomgeving. HvA publicaties. Amsterdam University Press, 2005. in Dutch. (PDF, 1960 Kbytes)

B.J.A. Kröse. Digital life is extra hulp in zorgsector. de Automatiseringsgids, 34:13, 2005. in Dutch. (PDF, 27 Kbytes)

J.M. Porta, J.J. Verbeek, and B.J.A. Kröse.Active appearance-based robot localization using stereo vision.Autonomous Robots, 18(1):59-80, 2005.(PDF, 2262 Kbytes)

J.M. Porta and B. J. A Kröse. Appearance-based concurrent map building and localization. Robotics and Autonomous Systems, 54(2):159-164, 2005. ISBN 0921-8890. (PDF, 1120 Kbytes)

J. J. Verbeek, N. Vlassis, and B. J. A. Kröse.Self-organizing mixture models.Neurocomputing, 63:99-123, 2005.(PDF, 859 Kbytes)

W. Zajdel and B. J. A. Kröse. A sequential bayesian algorithm for surveillance with non-overlapping cameras. Int. Journal of Pattern Recognition and Artificial Intelligence, 19(8):977-996, 2005. (PDF, 568 Kbytes)

W. Zajdel, N. Vlassis, and B. J. A Kröse. Bayesian methods for tracking and localization. In E. Aarts, J. Korts, and W. Verhaegh, editors, Intelligent Algorithms, pages 243-258. Kluwer Academic Publishers, 2005. (PDF, 166 Kbytes)

W. Zajdel, Z. Zivkovic, and B.J.A. Kröse. Keeping track of humans: have I seen this person before?. In Proc. of Int. Conference on Robotics and Automation (ICRA), pages 2093-2098, 2005. (Gzipped PostScript, 6 pages, 1179 Kbytes) (PDF, 1517 Kbytes)

Z. Zivkovic and B.J.A. Kröse. On matching interest regions using local descriptors - can an information theoretic approach help?. In Proc. British Machine Vision Conference, pages 50-58, 2005. (PDF, 241 Kbytes)

Z. Zivkovic, B. Bakker, and B.J.A. Kröse. Hierarchical map building using visual landmarks and geometric constraints. In Proc. IEEE/RSJ International Conference on Intelligent Robots and Systems, pages 7-12, 2005. (PDF, 550 Kbytes)

2004

B. Kröse, R. Bunschoten, S. ten Hagen, B. Terwijn, and N. Vlassis. Household robots look and learn. IEEE Robotics and Automation Magazine, 11(4):45-52, December 2004.

B.J.A. Kröse, N. Vlassis, and W. Zajdel. Bayesian methods for tracking and localization. In Proc. of Philips Symposium On Intelligent Algorithms, (SOIA), pages 27-38, 2004. (Gzipped PostScript, 12 pages, 184 Kbytes) (PDF, 167 Kbytes)

J.M. Porta and B.J.A. Kröse. Appearance-based concurrent map building and localization. In F.C.A. Groen, editor, International Conference on Intelligent Autonomous Systems, IAS'04, pages 1022-1029. IOS Press, March 2004. ISBN 1-58603-414-6.

J.M. Porta and B.J.A. Kröse. Appearance-based concurrent map building and localization using a multi-hypotheses tracker.. In Proc.IEEE/RSJ International Conference on Intelligent Robots and Systems, pages 3424-3429, Sendai, Japan, 2004. IEEE Press. (PDF, 156 Kbytes)

Martijn Reuvers, Richard Kleihorst, Harry Broers, and Ben Kröse. A smart camera for face recognition. In Proceedings of SPS-2004, 2004. (PDF, 225 Kbytes)

S.H.G. ten Hagen and B.J.A. Kröse. Learning to understand tasks for mobile robots. In Proc. of the IEEE Int. Conf. on System, Man and Cybernetics, The Hague, The Netherlands, October 2004. To Appear. (Gzipped PostScript, 6 pages, 467 Kbytes) (PDF, 448 Kbytes)

J.M. Terwijn, B. Porta and B.J.A. Kröse. A particle filter to estimate non-markovian states. In F.C.A. Groen, editor, International Conference on Intelligent Autonomous Systems, IAS'04, pages 1062-1069. IOS Press, March 2004. ISBN 1-58603-414-6.

W. Zajdel, A.T. Cemgil, and B. Kröse. Online multicamera tracking with a switching state-space model. In Proc. of IEEE International Conference on Pattern Recognition (ICPR), pages IV:339-343, Cambridge, UK, 2004. (PDF, 175 Kbytes)

Z. Zivkovic and B. Kröse. An EM-like algorithm for color-histogram-based object tracking. In IEEE Conference on Computer Vision and Pattern Recognition, June 2004. To appear. (PDF, 372 Kbytes)

Z. Zivkovic and B. Kröse. A probabilistic model for an EM-like object tracking algorithm using color-histograms. In 6th IEEE International Workshop on Performance Evaluation of Tracking and Surveillance (in connection with ECCV2004), May 2004. To appear. (PDF, 171 Kbytes)

2003

R. Bunschoten and B. Kröse. Robust scene reconstruction from an omnidirectional vision system. IEEE Transactions on Robotics and Automation, 19(2):351-357, 2003. (PDF, 886 Kbytes)

Roland Bunschoten and Ben Kröse. Visual odometry from an omnidirectional vision system. In Proceedings of the International Conference on Robotics and Automation ICRA'03, pages 577-583, Taipei, Taiwan, 2003. ISBN 0-7803-7737-0.

B.J.A. Kröse, J.M. Porta, K. Crucq, A.J.N. van Breemen, M. Nuttin, and E. Demeester. Lino, the user-interface robot. In E. Aarts, R. Collier, E. van Loenen, and B.D. Ruyter, editors, Proceedings of the First European Symposium on Ambience Intelligence (EUSAI), pages 264-274, Eindhoven, The Netherlands, November 2003. Springer. ISBN 3-540-20418-0. (PDF, 7004 Kbytes)

J.M. Porta and B.J.A. Kröse. Vision-based localization for mobile platforms. In E. Aarts, R. Collier, E. van Loenen, and B.D. Ruyter, editors, Proceedings of the First European Symposium on Ambience Intelligence (EUSAI), pages 208-219, Eindhoven, The Netherlands, November 2003. Springer. ISBN 3-540-20418-0. (PDF, 2051 Kbytes)

Josep M. Porta and Ben Kröse. On the use of disparity maps for robust robot localization under different illumination conditions. In A.T. de Almeida and U. Nunes, editors, Proceedings of the 11th International Conference on Advanced Robotics, ICAR'03, pages 124-129, Coimbra, Portugal, June 30-July 3 2003. IEEE Press. ISBN 972-96889-9-0. (Gzipped PostScript, 6 pages, 345 Kbytes) (PDF, 988 Kbytes)

J.M. Porta, J.J. Verbeek, and B.J.A. Kröse. Enhancing appearance-based robot localization using sparse disparity maps. In C.S.G. Lee and J. Yuh, editors, Proc.IEEE/RSJ International Conference on Intelligent Robots and Systems, pages 980-985, Las Vegas, USA, October 2003. IEEE Press. ISBN 0-7803-7861-X. (PDF, 253 Kbytes)

Josep M. Porta, Bas Terwijn, and Ben Kröse. Efficient entropy-based action selection for appearance-based robot localization. In Proceedings of the International Conference on Robotics and Automation ICRA'03, pages 2842-2847, Taipei, Taiwan, 2003. ISBN 0-7803-7737-0. (PDF, 114 Kbytes)

Stephan ten Hagen and Ben Kröse. Neural Q-learning. Neural Computing & Applications, 12(2):81-88, November 2003. ISSN: 0941-0643 (Paper) 1433-3058 (Online). (Gzipped PostScript, 13 pages, 163 Kbytes) (PDF, 248 Kbytes)

S.H.G. ten Hagen and B.J.A. Kröse. Learning to navigate using a lazy map. In A.T. de Almeida and U. Nunes, editors, Proceedings of the 11th International Conference on Advanced Robotics, ICAR'03, pages 299-304, Coimbra, Portugal, June 30-July 3 2003. (Gzipped PostScript, 6 pages, 157 Kbytes) (PDF, 119 Kbytes)

A.J.N van Breemen, K. Crucq, B.J.A Kröse, M. Nuttin, J.M. Porta, and E. Demeester. A user-interface robot for ambient intelligent environments. In P. Fiorini, editor, Proceedings of the 1st International Workshop on Advances in Service Robotics, ASER'03, pages 132-139, Bardolino, Italy, 2003. Fraunhofer IRB Verlag. (Gzipped PostScript, 8 pages, 5288 Kbytes) (PDF, 3287 Kbytes)

J.J. Verbeek, N. Vlassis, and B.J.A. Kröse. Efficient greedy learning of Gaussian mixture models. Neural Computation, 15(2):469-485, 2003. (PDF, 505 Kbytes)

J.J. Verbeek, N. Vlassis, and B.J.A. Kröse. Non-linear feature extraction by the coordination of mixture models. In S. Vassiliadis, L.M.J. Florack, J.W.J. Heijnsdijk, and A. van der Steen, editors, Proc. 8th Ann. Conf. of the Advanced School for Computing and Imaging (ASCI), pages 287-293, Heijen, The Netherlands, June 2003. ASCI. (PDF, 1331 Kbytes)

J.J. Verbeek, N. Vlassis, and B.J.A. Kröse. Self-organization by optimizing free-energy. In M. Verleysen, editor, Proc. of European Symposium on Artificial Neural Networks, pages 125-130. D-side, Evere, Belgium, 2003. (PDF, 184 Kbytes)

A.H.G. Versluis, B.J.F Driessen, J.A. van Woerden, and B.J.A. Kröse. Enhancing the usability of the MANUS manipulator by using visual servoing. In Proceedings of International Conference on Rehabilitation Robotics, ICORR 2003, pages 43-46, KAIST, Daejon, South Korea, 22-25 April 2003.

Wojciech Zajdel and Ben Kröse. Approximate learning and inference for tracking with non-overlapping cameras. In M.H. Hamza, editor, Proc. of the IASTED Int. Conf. on Artificial Intelligence and Applications, pages 70-75. ACTA Press, Calgary, Canada, 2003. (PDF, 125 Kbytes)

Wojciech Zajdel and Ben Kröse. Gaussian mixture model for multi-sensor tracking. In T. Heskes, P. Lucas, L. Vuurpijl, and W. Wiegerinck, editors, Proceedings of the 15th Dutch-Belgian Artificial Intelligence Conference, BNAIC'03, pages 371-378, Nijmegen, The Netherlands, October 2003. Elsevier. (PDF, 200 Kbytes)

2002

R. Bunschoten and B. Kröse. 3-D scene reconstruction from cylindrical panoramic images. Robotics and Autonomous Systems (special issue), 41(2/3):111-118, November 2002. (PDF, 225 Kbytes)

B.J.A. Kröse, N. Vlassis, and R. Bunschoten. Omnidirectional vision for appearance-based robot localization. In G.D. Hagar, H.I. Cristensen, H. Bunke, and R. Klein, editors, Sensor Based Intelligent Robots: International Workshop, Dagstuhl Castle, Germany, October 2000, Selected Revised Papers, number 2238 in Lecture Notes in Computer Science, pages 39-50. Springer, 2002. (PDF, 730 Kbytes)

S.H.G. ten Hagen and B.J.A. Kröse. Towards global consistent pose estimation from images. In R. Siegwart and C. Laugier, editors, Proc.IEEE/RSJ International Conference on Intelligent Robots and Systems, pages 466-471, Lausanne,Switzerland, September 2002. Omnipress. (Gzipped PostScript, 6 pages, 317 Kbytes) (PDF, 176 Kbytes)

S.H.G. ten Hagen and B.J.A. Kröse. Trajectory reconstruction for self-localization and map building. In W.R. Hamel and A.A. Maciejewski, editors, Proc. IEEE Int. Conf. on Robotics and Automation, pages 1796-1801, Washington D.C., USA, May 2002. Omnipress. (Gzipped PostScript, 6 pages, 168 Kbytes) (PDF, 143 Kbytes)

J.J. Verbeek, N. Vlassis, and B. Kröse. A k-segments algorithm for finding principal curves. Pattern Recognition Letters, 23(8):1009-1017, 2002. (Gzipped PostScript, 12 pages, 97 Kbytes) (PDF, 149 Kbytes)

J.J. Verbeek, N. Vlassis, and B.J.A. Kröse. Coordinating Principal Component Analyzers. In J.R. Dorronsoro, editor, Proceedings of International Conference on Artificial Neural Networks, Lecture Notes in Computer Science, pages 914-919, Madrid, Spain, August 2002. Springer. (PDF, 251 Kbytes)

J.J. Verbeek, N. Vlassis, and B.J.A. Kröse. Fast non-linear dimensionality reduction using topology preserving networks. In M. Verleysen, editor, Proc. of European Symposium on Artificial Neural Networks, pages 193-198. D-side, Evere, Belgium, 2002. (PDF, 167 Kbytes)

J.J. Verbeek, N. Vlassis, and B.J.A. Kröse. Locally linear generative topographic mapping. In M. Wiering, editor, Benelearn 2002: Proceedings of the Twelfth Belgian-Dutch Conference on Machine Learning, Utrecht, The Netherlands, December 2002. (PDF, 158 Kbytes)

N. Vlassis, Y. Motomura, and B. Kröse. Supervised dimension reduction of intrinsically low-dimensional data. Neural Computation, 14(1):191-215, January 2002. (Gzipped PostScript, 22 pages, 331 Kbytes)

N. Vlassis, B. Terwijn, and B. Kröse. Auxiliary particle filter robot localization from high-dimensional sensor observations. In W.R. Hamel and A.A. Maciejewski, editors, Proc. IEEE Int. Conf. on Robotics and Automation, pages 7-12, Washington D.C., USA, May 2002. Omnipress. (Gzipped PostScript, 6 pages, 218 Kbytes) (PDF, 178 Kbytes)

W. Zajdel and B. Kröse. Bayesian network for multiple hypothesis tracking. In H. Blockeel and M. Denecker, editors, Proceedings of the 14th Dutch-Belgian Artificial Intelligence Conference, BNAIC'02, pages 379-386, Leuven, Belgium, October 2002. (Gzipped PostScript, 8 pages, 52 Kbytes) (PDF, 74 Kbytes)

H. Asoh, N. Vlassis, Y. Motomura, F. Asano, I. Hara, S. Hayamizu, K. Itou, T. Kurita, T. Matsui, R. Bunschoten, and Ben Kröse. Jijo-2: An office robot that communicates and learns. IEEE Intelligent Systems, 16(5):46-55, Sep/Oct 2001. (PDF, 1107 Kbytes)

R. Bunschoten and B. Kröse. 3-d scene reconstruction from cylindrical panoramic images. In Proceedings of the 9th International Symposium on Intelligent Robotic Systems (SIRS'2001), pages 199-205, LAAS-CNRS, Toulouse, France, July 2001. (PDF, 220 Kbytes)

R. Bunschoten and B. Kröse. 3-d scene reconstruction from multiple panoramic images. In Proceedings of 7th annual conference of the Advanced School for Computing and Imaging (ASCI 2001), pages 49-54, Heijen, The Netherlands, May 2001. ASCI. (PDF, 141 Kbytes)

R. Bunschoten and B. Kröse. Range estimation from a pair of omnidirectional images. In Proc. IEEE Int. Conf. on Robotics and Automation, pages 1174-1179, Seoul, Korea, May 2001.

F.C.A. Groen, W. van der Hoek, P. Jonker, B. Kröse, H. Spoelder, and S. Stramigioli. Robocup european championship: Report on the Amsterdam 2000 event. Robotics and Autonomous Systems, 36(2-3):59-66, August 2001.

B.J.A. Kröse, N. Vlassis, R. Bunschoten, and Y. Motomura. A probabilistic model for appearance-based robot localization. Image and Vision Computing, 19(6):381-391, April 2001. (Gzipped PostScript, 17 pages, 604 Kbytes) (PDF, 1074 Kbytes)

Stephan ten Hagen. Continuous State Space Q-Learning for Control of Nonlinear Systems. PhD thesis, Computer Science Institute, University of Amsterdam, The Netherlands, February 2001. (Gzipped PostScript, 128 pages, 1059 Kbytes) (PDF, 1729 Kbytes)

J.J. Verbeek, N. Vlassis, and B. Kröse. Efficient greedy learning of Gaussian mixtures. In Proc. 13th Belgian-Dutch Conf. on Artificial Intelligence, Amsterdam, The Netherlands, October 2001.

J.J. Verbeek, N. Vlassis, and B. Kröse. Greedy Gaussian mixture learning for texture segmentation. In A. Leonardis and H. Bischof, editors, ICANN'01, Workshop on Kernel and Subspace Methods for Computer Vision, pages 37-46, Vienna, Austria, August 2001.

J.J. Verbeek, N. Vlassis, and B. Kröse. A soft k-segments algorithm for principal curves. In Proc. Int. Conf. on Artificial Neural Networks, pages 450-456, Vienna, Austria, August 2001. (Gzipped PostScript, 7 pages, 80 Kbytes)

N. Vlassis, R. Bunschoten, and B. Kröse. Learning task-relevant features from robot data. In Proc. IEEE Int. Conf. on Robotics and Automation, pages 499-504, Seoul, Korea, May 2001. (Gzipped PostScript, 6 pages, 200 Kbytes)

N. Vlassis, Y. Motomura, and Ben Kröse. Supervised dimension reduction of intrinsically low-dimensional data. Neural Computation, 14:1-25, 2001. To appear. (Gzipped PostScript, 22 pages, 331 Kbytes)

``Feature selection for appearance-based robot localization'' Proceedings 2000 RWC Symposium, RWC Technical Report (TR-99-002)  Gzipped postscript 334Kb

---

Kröse,B.J.A. A. Dev and F.C.A. Groen (2000) 

`` Heading Direction of a Mobile Robot from the Optical~Flow'' Image and Vision Computing Journal, vol.18 nr. 5, pp. 415-424 Gzipped postscript 1.9Mb

---

Portegies Zwart, Joris and Kröse, Ben(2000)

`` Constrained Mixture Modeling of Intrinsically Low-Dimensional Distributions,'' 15th International Conference on Pattern Recognition, Volume 2: Pattern Recognition and Neural Networks, (Sanfeliu, A. and Villanueva, J.J. and Vanrell, M. and Alquézar, R. and Jain, A.K. and Kittler, J., ed.), IEEE,pp. 610-613 Postscript available from Joris' web page

---

Wiering, M., Kröse,B.J.A. and F.C.A. Groen (2000) 

``Learning in Multi-Agent Systems'' SubmittedGzipped postscript 134Kb

---

Vlassis, N., Motomura, Y. and Kröse,B.J.A. (2000) 

`` Appearance based robot localization'' IJCAI-99 Workshop Adaptive Spatial Representations of Dynamic Environments, Stockholm, Sweden   Gzipped postscript 0.5Mb

---

Kröse,B.J.A. and Bunschoten,R. (1999) 

`` Probabilistic localization by appearance models and active vision'' Proceedings of the IEEE International Conference on Robotics and Automation, pp 2255-2260   Gzipped postscript file: 300kb 

---

Y. Motomura, N. Vlassis, B. Kröse (1999) 

`` Probabilistic Robot Localization and Situated Feature Focusing'' Proc. SMC'99, IEEE Int. Conf. on Systems, Man, and Cybernetics, Tokyo, Japan, Oct 1999. 

---

Y. Motomura, N. Vlassis, B. Kröse (1999) 

`` Environment Modeling via PCA Regression and Situated Feature Focusing'' Special Interest Group on Mathematical modeling and Problem Solving of Information Processing Society of JAPAN, May 1999. 

---

N. Vlassis, Y. Motomura, B. Kröse (1999) 

`` An information-theoretic localization criterion for robot map building'' Proc. ACAI'99, Int. Conf. on Machine Learning and Applications Chania, Greece, Jul 1999

---

N. Vlassis, B. Kröse (1999) 

`` Mixture Conditional Density Estimation with the EM Algorithm'' Proc. ICANN'99, 9th Int. Conf. on Artificial Neural Networks, Edinburgh, Scotland, Sep 1999.

---

N. Vlassis, B. Kröse (1999) 

`` The design of OASIS: Open Architecture for Simulations with Intelligent Systems,'' Proc ESM'98, Manchester June 16-19 1998, SCS Publication, ISBN 1- 56555-148-6, (Zobel, R. and Moeller, D, ed.), pp. 455-459 

---

Corten, E. and Dorst, L. and Krose, B. (1998) 

`` OASIS: Open Architecture for Simulations with Intelligent Systems,'' Proc IAS-5, Sapporo June 2-4 1998, IOS press, ISBN 90 51993986, (Kakazu, Y. and Wada, M. and Sato, T., ed.), pp. 6-12 

---

Dev, A. and Kröse, B.J.A. and Groen, F.C.A. (1998) 

Postscript file: click here to get 396 Kb

Abstract: If a camera moves on a straight line, the optic flow field is a diverging vector field, of which the singularity is called "focus of expansion". An object which is seen in this FOE is located on the future path of the camera. If the camera is also rotating, the future path is no longer a point in the image domain, but a line. All objects which are on the future path (and thus will cause collisions) are projected on this line. However, not necessary the reverse is true: not all points on the line are collision points. In this paper we derive how the optic flow can be used to compute which points in the image are projections of collision points.

`` Predicting the future path from optic flow.,'' Proc. 1998 RWC Symposium, Tokyo June 9-10 1998, RWC Technical Report TR-98001, pp. 265-270 

---

Hagen, Stephan ten and Kröse, Ben (1998) 

Postscript file: click here to get 67 Kb

Abstract: This manuscript is a submission to the workshop ``Machine Learning and Reinforcement Learning for Manufacturing''. It introduces and positions our part in a project and motivates our approach with respect to reinforcement learning and manufacturing processes. In an ``extended appendix'' some additional information will be given about our problem domain and preliminary results. Our main issue is that advances in algorithms and theory should not be scaled up to bigger problems, but to more realistic problems. Realistic in the sense that the problem is formulated with the problems of existing manufacturing processes in mind.

Postscript file: click here to get 62 Kb 

Abstract: In this paper we focus on Q-learning in domains with continuous state and action spaces. We discuss how Q-learning relates to System Identification (SI) methods for Linear Quadratic Regulation (LQR) and show how the methods compare on linear systems. We also study the use of a feedforward network as a nonlinear function approximator for the Q-function and introduce the the concept of Pseudo-Parametric Q-Learning (PPQL). In the PPQL framework the feedforward network is implemented such, that the results can be interpreted in terms of LQR conditions. Experiments show that it performs well, but does not necessarily converge to a stable solution. The LQR interpretation indicates the origin of that problem. 

Postscript file: click here to get 82 Kb 

Abstract: In this paper we describe a possible way to mak e reinforcement learning more applicable in the context of industrial manufactur ing processes. We achieve this by formulating the optimization task in the linear quadratic reg ulation framework, for which a conventional control theoretic solution exist. By rewriting the Q-learning approach into a linear least squares approximation p roblem, we can make a fair comparison between the resulting approximation and th at of the conventional system identification approach. Our experiment shows that the conventional approach performs slightly better. Also we can show that the amount of exploration noise, added during the generati on of data, plays a crucial role in the outcome of both approaches. 

Postscript file: click here 

Abstract: For navigation to a desired state, a mobile rob ot needs some sort of global information about the environment it is operating i n. Usually this is provided in the form of a map, giving locations of objects and f ree space in the working space of the robot. Such a map can be provided by the programmer or learned by the system itself. In this paper an approach is described where the global information is not cast in a model of the geometry of the envir onment but in a model of all sensory data of the robot. Experimental results are presented. 

Postscript file: click here to get 295 Kb

Abstract: Estimation of image motion, also known as the optic flow, from a sequence of images is known to be difficult. This is due to: the sensitivity of the image motion model to noise (the derivative property), the limited observability of the image motion from the luminance (the aperture problem), and, the non-validity of the optic flow constraint (the assumption of intensity conservation). In this paper we analyze measures that assign a confidence value to the estimated image motion: the sensitivity of the model to noise, the validity of the model and the estimated variance of the image motion. Experiments show that selection of image motion vectors based on these measures dramatically improve the estimates of the image motion while keeping as much image motion vectors as possible. We conclude that the proposed estimated variance of the image motion optimizes this trade-off.

Postscript file: click here to get 450 Kb

Abstract: The robot navigation task presented in this paper is to drive through the center of a corridor, based on a sequence of images from an on-board camera. Our measurements of the system state, the distance to the wall and orientation of the wall, are derived from the optic flow. Whereas the structure of the environment is usually computed from the spatial derivatives of the optic flow, we use the structure contained in the temporal derivatives of the optic flow to compute the environment structure and hence the system state. The algorithm is used to control a `remote brain' robot and results on the accuracy of the state estimates are presented.

Postscript file: click here to get 52 Kb

Abstract: Convergence of TD($\lambda$) with radial base function network.

Postscript file: click here to get 69 Kb

Abstract: This introduction is meant for readers with no knowledge about reinforcement learning. It presents the basic framework and introduce the basic terminology. We hope that this will make it easier to read other reinforcement learning literature. Pointers to more tutorial sources will be given at the end.

Postscript file: click here to get 86 Kb

Abstract: In this paper we propose a reactive critic, that is able to respond to changing situations. We will explain why this is useful in reinforcement learning, where the critic is used to improve the control strategy. We take a problem for which we can derive the solution analytically. This enables us to investigate the relation between the parameters and the resulting approximations of the critic. We will also demonstrate how the reactive critic reponds to changing situations.

Postscript file: click here to get 76 Kb

Abstract: A review is given on the use of neural networks for mobile robots and autonomous vehicles. We focus on neural methods for navigation, making a distinction between sensor-based `reactive' navigation and planned navigation methods.

Postscript file: click here to get 450 Kb

Abstract: We describe a remote brain mobile robot based on off-the-shelve components. The navigation task presented in this paper is to drive through the center of a corridor, based on a sequence of images from an on-board camera. A simple control scheme is presented. Our measurements of the system state, the distance to the wall and orientation of the wall, are derived from the optic flow. Whereas this structure of the environment is usually computed from the spatial structure of the optic flow, i.e. the spatial derivatives of the optic flow, for robustness reasons we use the structure contained in the temporal derivatives of the optic flow to compute the environment structure and hence the system state.

Postscript file: click here to get 79 Kb

Abstract: Because of the huge amount of data which is available nowadays, the current manager or decision maker needs intelligent data analysis tools. Those tools must be able to visualize the data, to cluster the data or to make predictions based on the data. In this paper we describe how neural networks have been used for the analysis of flight booking profiles at KLM Royal Dutch Airlines.

Note: Presented at ``SNN'97, Europe's best neural networks practice', Amsterdam, 22 May 1997.

Postscript file: click here to get 100 Kb

Abstract: We present a method for vision-based model-free positioning of a 6-degree-of-freedom robot end-effector with respect to a planar target object using a feed-forward neural network. We investigate the necessary conditions under which a neural network can learn the mapping from feature domain to actuator domain. After satisfying these conditions, a neural network is used to learn this mapping. We consider only planar objects as target and their binary images. Moment-based image descriptors are used to represent the image in the feature domain. Simulation results are also presented.

Postscript file: click here to get 65 Kb

Abstract: For a unique characterization of the relative position between a 2-D planar object (target) and a camera, the following two mappings have to be single-valued: mapping from the relative position to the image plane and from the image plane to the feature domain. We consider only white planar targets located in a black background and designed a special target which allows a unique perspective from any relative position. From the image of this target, the 6 relative position and orientation parameters can be characterized by means of 6 independent features. We use moments to extract these features and choose the proper representation to make them independent.

Postscript file: click here to get 143 Kb

Keywords: learning sensor models, neural networks, sensor fusion, occupancy grids

Abstract: In this paper we consider the conversion of sensor data to a probabilistic representation of the environment (occupancy grid). We introduce a neural network which learns these conversions. The conversion of sensor data remains adaptive to changes in either the sensor or its environment. To place this in a broader context, we describe the architecture of our Sensor Data Fusion system in which these conversions are applied. We also introduce the PDOP: a rule for fusing occupancy grids in this system.

Postscript file: click here to get 112 Kb

Keywords: learning sensor models, neural networks, sensor fusion, occupancy grids

Abstract: Key issue in the design of a sensor data fusion system is the conversion of sensor measurements to an internal representation. In this article, we identify the problems with traditional conversion methods and we introduce a neural network which learns how to convert such measurements.

Postscript file: click here to get 103 Kb

Abstract: Reinforcement learning (RL) is a model-free tuning and adaptation method for control of dynamic systems. Contrary to supervised learning, based usually on gradient descent techniques, RL does not require any model or sensitivity function of the process. Hence, RL can be applied to systems that are poorly understood, uncertain, nonlinear or for other reasons untractable with conventional methods. In reinforcement learning, the overall controller performance is evaluated by a scalar measure, called reinforcement. Depending on the type of the control task, reinforcement may represent an evaluation of the most recent control action or, more often, of an entire sequence of past control moves. In the latter case, the RL system learns how to predict the outcome of each individual control action. This prediction is then used to adjust the parameters of the controller. The mathematical background of RL is closely related to optimal control and dynamic programming. This paper gives a comprehensive overview of the RL methods and presents an application to the attitude control of a satellite. Some well known applications from the literature are reviewed as well.

Postscript file: click here to get 114 Kb

Abstract: For a target tracking task, the hand-held camera of the anthropomorphic OSCAR-robot manipulator has to track an object which moves arbitrarily on a table. The desired camera-joint mapping is approximated by a feedforward neural network. Through the use of time derivatives of the position of the object and of the manipulator, the controller can inherently predict the next position of the moving target object. In this paper several `anticipative' controllers are described, and successfully applied to track a moving object.

Postscript file: click here to get 75 Kb

Abstract: The reconstruction of the observer motion and environment structure from the optic flow is considered for the case where the camera mapping is unknown. This mapping has therefore to be estimated from a given set of examples, the training set. Since this mapping is not a function in the sense of a $m$ to $ map , standard neural networks are unable to learn this mapping. We propose to represent these mappings from $\calX\rightarrow\calY$ as a manifold in the product space $\calX\times\calY$. We approximate a parameterization of the manifold from a given set of data points by using an auto association network with a \em bottleneck layer. A gradient descend algorithm is used on the trained network to find the approximation of the egomotion and scene structure for a given set of optic flow vectors.

Postscript file: click here to get 72 Kb

Keywords: Structure from Motion, time-to-contact, neural networks, Multi valued mappings

Abstract: The reconstruction of the observer motion and environment structure from optic flow is considered for the case where the camera mapping is unknown. This mapping has therefore to be estimated from a given set of examples, the training set. Since this mapping is not a function in the sense of a many to one mapping, standard neural networks are unable to learn this mapping. We propose to represent these mappings as a manifold in the product space. We approximate a parameterization of the manifold from a given set of data points by using an auto associative neural network with a bottleneck layer. A gradient descent algorithm is used on the parameterization of the learned manifold to find the approximation of the ego-motion for a given set of optic flow vectors.

---

 Kröse, B.J.A. (1995)

Postscript file: click here to get 31 Kb

Note: Editorial paper

---

 Smagt, P.P. van der and Groen, F.C.A. and Kröse, B.J.A. (1995)

Postscript file: click here to get 101 Kb

Keywords: time-to-contact, neural networks, hand-eye coordination, robot arm control, monocular vision

Abstract: In this paper we introduce a method for model-free monocular visual guidance of a robot arm. The robot arm, with a single camera in its end-effector, should be positioned above a visually observed target. It is shown that a trajectory can be planned in visual space by using components of the optic flow, and this trajectory can be translated to joint torques by a self-learning neural network. No model of the robot, camera, or environment is used. The method reaches a high grasping accuracy after only a few trials.

Postscript file: click here to get 97 Kb

Abstract: We propose a method for decreasing the computational complexity of self-organising maps. The method uses a partitioning of the neurons into disjoint clusters. Teaching of the neurons occurs on a cluster-basis instead of on a neuron-basis. For teaching an N-neuron network with N' samples, the computational complexity decreases from O(NN') to O(N log N'). Furthermore, we introduce a measure for the amount of order in a self-organising map, and show that the introduced algorithm behaves as well as the original algorithm.

Postscript file: click here to get 52 Kb

Abstract: Orthogonal incremental learning (OIL) is a new approach of incremental training for a feedforward network with a single hidden layer. OIL is based on the idea to describe the output weights (but not the hidden nodes) as a set of orthogonal basis functions. Hidden nodes are treated just as the orthogonal representation of the network in the output weights domain. We showed that the network training can be performed incrementally, one node at time, and there is no need to use an additional constraint to support a consistent optimization among the hidden nodes. An advantage of OIL over existing algorithms is extremely fast learning. This approach can be also easily extended to build-up incrementally an arbitrary function as a linear composition of adjustable functions which are not necessarily orthogonal. We tested this approach on a standard "two-spirals" benchmark problem to build incrementally a feedforward network with a single layer of Gaussian units.

---

 Dam, J.W.M. van and Kröse, B.J.A. and Groen, F.C.A. (May 1994)

Postscript file: click here to get 47 Kb

Postscript file: click here to get 42 Kb

Postscript file: click here to get 88 Kb

Postscript file: click here to get 110 Kb

Keywords: Structure from Motion, time-to-contact, Navigation, Mobile Robots, Curvature scaled depth

Abstract: The robot is equipped with monocular vision to sense its environment. Motion of the robot results in motion of the environment in the sensory domain. The optic flow equals the projection of the environment motion on the image plane. We show that under a continuity assumption, the collision points can be computed from the optic flow without deriving a model of the environment. We will mainly consider a mobile robot. We derive the collision points by introducing an invariant, the curvature scaled depth. This invariant couples the rotational velocity of the robot to its translational velocity and is closely related to the curvature of the mobile robot's path. We show that the spatial derivatives of the curvature scaled depth give the object surface orientation.

Postscript file: click here to get 140 Kb

---

 Kröse, B.J.A. and Eecen, M. (1994)

`` Self-learning maps for path planning in sensor space,'' ICANN'94, Proceedings of the International Conference on Artificial Neural Networks, (Marinaro, M. and Morasso, P.G., ed.), Springer-Verlag, pp. 1303-1306 

---

Kröse, B.J.A. and Eecen, M. (1994) 

Postscript file: click here to get 136 Kb

Keywords: mobile robot navigation, sensor-based planning, environment modelling, neural network techniques

Abstract: The paper describes a sensor based navigation scheme which makes use of a global representation of the environment by means of a self-organizing map or Kohonen network. In contrast to existing methods for self-organizing environment representation, this discrete map is not represented in the world domain or in the configuration space of the vehicle, but in the sensor domain. The map is built by exploration. A conventional path planning technique now gives a path from current state to a desired state in the sensor domain, which can be followed using sensor based control. Collisions with obstacles are detected and used in the path planning. Results from a simulation show that the learned representation gives correct paths from an arbitrary starting point to an arbitrary end point.

Postscript file: click here to get 438 Kb

Published as: lecture book

Postscript file: click here to get 100 Kb

Abstract: A pilot study is described on the practical application of artificial neural networks. The limit cycle of the attitude control of a satellite is selected as the test case. One of the sources of the limit cycle is a position dependent error in the observed attitude. A Reinforcement Learning method is selected, which is able to adapt a controller such that a cost function is optimised. An estimate of the cost function is learned by a neural critic. In our approach, the estimated cost function is directly represented as a function of the parameters of a linear controller. The critic is implemented as a CMAC network. Results from simulations show that the method is able to find optimal parameters without unstable behaviour. In particular in the case of large discontinuities in the attitude measurements, the method shows a clear improvement compared to the conventional approach: the RMS attitude error decreases approximately 30 procent.

Postscript file: click here to get 96 Kb

Abstract: Neural controllers are able to position the hand-held camera of the (3DOF) anthropomorphic OSCAR-robot manipulator above an object which is arbitrary placed on a table. The desired camera-joint mapping is approximated by feedforward neural networks. However, if the object is moving, the manipulator lags behind because of the required time to preprocess the visual information and to move the manipulator. Through the use of time derivatives of the position of the object and of the manipulator, the controller can inherently predict the next position of the object. In this paper several predictive controllers are proposed, and successfully applied to track a moving object.

Postscript file: click here to get 197 Kb

Postscript file: click here to get 52 Kb

Postscript file: click here to get 81 Kb

`` Neural Networks for robot eye-hand coordination,'' Artificial neural networks, (Gielen, S. and Kappen, B., ed.), Springer-Verlag, pp. 211-218 

---

Kröse, B.J.A. and Compagner, K. and Groen, F.C.A. (1993) 

`` Accurate estimation of environment parameters from ultrasonic data,'' Robotics and Autonomous Systems 11 (3/4), pp. 221-230 

---

Kröse, B.J.A. and Smagt, P.P. van der and Groen, F.C.A. (1993) 

Postscript file:click here to get 77 Kb

Keywords: neural networks, robot arm control, hand-eye coordination, monocular vision

Abstract: A self-learning, adaptive control system for a robot arm using a vision system in a feedback loop is described. The task of the control system is to position the end-effector as accurate as possible directly above a target object, so that it can be grasped. The camera of the vision system is positioned in the end-effector and the visual information is used directly to control the robot. Two strategies are presented to solve the problem of obtaining 3D information from a single camera: a) using the size of the target object and b) using information from a sequence of images from the moving camera. In both cases a neural network is trained to perform the desired mapping.

Postscript file: click here to get 493 Kb

Keywords: feed-forward neural networks, robot arm control, hand-eye coordination

Abstract: This paper focuses on static hand-eye coordination. The key issue that will be addressed is the construction of a controller that eliminates the need for calibration. Instead, the system should be self-learning and must be able to adapt itself to changes in the environment. In this application, only positional information in the system will be used; hence the above reference `static.' Three coordinate domains are used to describe the system: the Cartesian world-domain, the vision domain, and the robot domain. The task that is set out to be solved is the following. A robot manipulator has to be positioned directly above a pre-specified target, such that it can be grasped. The target is specified in terms of visual parameters. Only the (x,y,z) position of the end-effector relative to the target is taken into account; this suffices for many pick-and-place problems encountered in industry. (In a number of cases, also the rotation of the hand is of importance, but this rotation can be executed separate from the 3D positioning problem.) Thus the remaining problem is 3 degrees-of-freedom (DoF).

Keywords: Feedforward networks, function approximation, hidden units

Abstract: This paper presents a methodology to estimate the optimal number of learning samples and the number of hidden units needed to obtain a desired accuracy of a function approximation by a feedforward network. The representation error and the generalization error, components of the total approximation error are analyzed and the approximation accuracy of a feedforward network is investigated as a function of the number of hidden units and the number of learning samples. Based on the asymptotical behaviour of the approximation error, an asymptotical model of the error function (AMEF) is introduced of which the parameters can be determined experimentally. In combination with knowledge about the computational complexity of the learning rule an optimal learning set size and number of hidden units can be found resulting in a minimum computation time for a given desired precision of the approximation.

Postscript file: click here to get 137 Kb

Keywords: Feedforward networks, function approximation, continuous mapping, learning from examples, generalization, hidden units

Abstract: This paper presents a method to estimate the optimal number of learning samples and the number of hidden units for a function approximation by a feedforward network. The optimality is considered under the minimal learning time constraint for a given degree of accuracy which is an essential point for real-time learning. The approximation error is modeled as a function of the number of hidden units and the number of learning samples. Two models are presented: the first one is based on general bounds of approximation and the second one on an asymptotic expansion of the approximation error. This approach was applied to optimize the learning of the camera-robot mapping of a visually guided robot arm and a complex logarithm function approximation. The results of this investigation suggested that the actual approximation errors differ considerably from the theoretical upper bounds.

---

 Kröse, B.J.A. and Dam, J.W.M. van (Jun. 1992)

Postscript file: click here to get 67 Kb

Postscript file: click here to get 44 Kb

Postscript file: click here to get 47 Kb

`` Visually controlled manipulator movements: the SNN demo project,'' Proceedings of the 2nd Symposium on neural networks, Foundation for Neural Networks, Nijmegen, pp. 6-10 

---

 Smagt, P.P. van der and Kröse, B.J.A. and Groen, F.C.A. (Jun. 1992)

Postscript file: click here to get 68 Kb

Keywords: time-to-contact, neural networks, hand-eye coordination, robot arm control, monocular vision

Abstract: A method is presented to learn 3D grasping of objects with unknown dimensions using a monocular eye-in-hand manipulator. From a sequence of images a motion profile is generated to approach the object of unknown size. It is shown that monocular visual information suffices to control the deceleration of the robot manipulator. A strategy for generating learning samples is presented, and simulation results demonstrate the effectiveness of the method.

`` A Cyclops Learns to Grasp,'' Proceedings of the Second Symposium on Neural Networks, The Dutch Foundation for Neural Networks, pp. 88 

---

Verschure, P.F.M.J. and Pfeifer, R. and Kröse, B.J.A. (1992) 

`` Distributed Adaptive Control: the self organization of structured behavior,'' Robotics and Autonomous Systems 9 (2), pp. 181-196 

---

Smagt, P.P. van der and Kröse, B.J.A. (June 1991) 

Postscript file: click here to get 48 Kb

Keywords: neural networks, conjugate gradient learning, hand-eye coordination, robot arm control

Abstract: A neurally based adaptive controller for a 6 degrees of freedom (DOF) robot manipulator with only rotary joints and a hand-held camera is described. The task of the system is to place the manipulator directly above an object that is observed by the camera (i.e., 2D hand-eye coordination). The requirement of adaptivity results in a system which does not make use of any inverse kinematics formulas or other detailed knowledge of the plant; instead, it should be self-supervising and adapt on-line. The proposed neural system will directly translate the preprocessed sensory data to joint displacements. It controls the plant in a feedback loop. The robot arm may make a sequence of moves before the target is reached, when in the meantime the network learns from experience. The network is shown to adapt quickly (in only tens of trials) and form a correct mapping from input to output domain.

R.P.W. Duin and Kröse, B.J.A.  (1980) 
`` On the possibility of avoiding peaking.'' Proceedings 5th Int. Conf. on Pattern Recognition, 1980, Miami, U.S.A.

Kröse, B.J.A (1987) 
``Local structure analyzers as determinants of preattentive pattern discrimination,'' Biological Cybernetics 55, 286-298 (1987).

  G.J.F. Smets, P.J. Stappers and  Kröse, B.J.A (1988) 
 "Form detection: features or invariance". Perceptual and Motor Skills, 67, 311-317 (1988).

B. Julesz and  Kröse, B.J.A (1988) 
``Visual texture perception: features and spatial filters." Nature 333, 302-303 (1988).

Kröse, B.J.A and B. Julesz (1989) 
``The control and speed of shifts of attention". Vision Research 29 (11) 1607-1619 (1989).
. 
Kröse, B.J.A  and C.A. Burbeck (1989) 
``Spatial Interactions in rapid pattern discrimination. Spatial Vision 4 (4) 211-222 (1989)