B. Kr ose, N. Vlassis, R. Bunschoten, and Y. Motomura. A probabilistic model for appearance-based robot localization. In Image and Vision Computing, volume 19, pages 381--391, 2001.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....map of the environment. When the robot is moving, the collected images are compressed using PCA and the resulting set of feature detectors is compared with those stored in the appearance based map. The position of the map points closer to the current observation are used within a Markov process [14] to improve the estimation of the position of the robot. We have obtained good results combining this approach with the use of a particle filter to estimate the probability distribution on the roboffs position [24] In our previous work, we used an omnidirectional camera. The advantage of such a ....

B.J.A. Kr6se, 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.

....J. M. Porta, B. Terwijn, and B. Kr ose IAS, Intelligent Autonomous Systems, Informatics Institute University of Amsterdam, Kruislaan 403, 1098 SJ, Amsterdam, The Netherlands porta,bterwijn,krose science.uva.nl Abstract In this paper, we extend our appearance based localization system [8] moving from a passive approach to an active one where the robot can execute actions with the only purpose of gaining information about its location in the environment. We present a general framework for entropy based action selection and we describe how this framework can be implemented in our ....

....(i.e. the subset of training points x i with an observation y i more similar to y) # j (y j ) a set of weights that favor closer nearest neighbors, and # a Gaussian. IV. ACTIVE LOCALIZATION The previous model offers good results for passive localization using an omnidirectional camera [8]. However, when using a camera mounted on a pan and tilt, as it is our case, the robot can decide by itself where to look to increase the certainty on its position estimation. Next, we describe a general criterion for action selection and how to implement it in our localization framework. A. ....

B.J.A. Kr ose, 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.

....of the hemisphere. Catadioptric omnidirectional vision sensors are quickly gaining popularity. These sensors consist of a camera and a carefully selected mirror lens combination. They have been proven to be useful for robot environment modelling, both in the sensory domain (appearance models) 1] [2], 3] 4] as well as in the geometric domain (Cartesian maps) 5] 4] A traditional approach to obtain a 3D reconstruction of a scene from image data is stereo vision. Stereo vision the process of recovering depth information from two or more calibrated images obtained from different but ....

B. Kr ose, N. Vlassis, R. Bunschoten, and Y. Motomura, "A probabilistic model for appearance-based robot localization," Image and Vision Computing, vol. 19, no. 6, pp. 381--391, Apr. 2001.

....in [4, Sec. 7] is to estimate it nonparametrically, that is, using a supervised training set of robot states fx k g and respective observations fy k g, for k = 1; K. A classical method is through kernel smoothing [9] an approach we have used in the past for similar modeling problems [10]. This method places multivariate kernels on each stateobservation pair, and then estimates the unknown conditional density by p(y; x) p(x) k=1 (xjx k ) yjy k ) k=1 (xjx k ) 9) where (xjx k ) and (yjy k ) are appropriate density kernels in the X and Y spaces, respectively. However, ....

B.J.A. Krose, N. Vlassis, R. Bunschoten, and Y. Motomura, \A probabilistic model for appearancebased robot localization," Image and Vision Computing, vol. 19, no. 6, pp. 381-391, Apr.

....y mentioned in [3, Sec. 7] is to estimate it nonparametrically, using a supervised training set of robot states fx k g and respective observations fy k g, for k = 1; K. A classical method is through kernel smoothing [7] an approach we have used in the past for similar modeling problems [8]. This method places multivariate kernels on each state observation pair, and then estimates the conditional density using p(yjx) p(y; x) p(x) P K k=1 (xjx k ) yjy k ) P K k=1 (xjx k ) 8) where (xjx k ) and (yjy k ) appropriate density kernels in the X and Y spaces, ....

B.J.A. Krose, N. Vlassis, R. Bunschoten, and Y. Motomura, \A probabilistic model for appearance-based robot localization," Image and Vision Computing, vol. 19, no. 6, pp. 381-391, Apr. 2001.

No context found.

B. Kr ose, N. Vlassis, R. Bunschoten, and Y. Motomura. A probabilistic model for appearance-based robot localization. In Image and Vision Computing, volume 19, pages 381--391, 2001.

No context found.

B. Kr ose, N. Vlassis, R. Bunschoten, and Y. Motomura. A probabilistic model for appearance-based robot localization. In Image and Vision Computing, volume 19, pages 381--391, 2001.

No context found.

B.J.A. Kr ose, 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.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC