S. Se, D. Lowe, and J. Little. Vision-based mobile robot localization and mapping using scaleinvariant features. In Proceedings of the IEEE International Conference on Robotics and Automation (ICRA), pages 2051.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....[14] 10] 6] 2] We extend this idea by measuring both azimuth and elevation angles of a landmark to represent its position on a sphere instead of a circle (Fig. 1) There have been a number of papers on the process of selecting useful features points or natural markers in image data [7], 8] 9] 11] 12] 13] 15] These approaches select optimal landmarks based on their appearance in the image. In this paper we analyze the problem of finding the optimal placement of tracked landmarks based on their 3D position in the world. In [3] we presented a system that allows ....

S. Se, D. Lowe, and J. Little. Vision-based mobile robot localization and mapping using scale-invariant features. In IEEE Conf on Robotics and Automation, pages 2051.

....from which the reference images were taken. 9] extract vertical lines from camera images and combine this information with data obtained from ultrasound sensors to estimate the position of the robot. 13, 21] consider trajectories in the Eigenspaces of features. A recent work presented in [15] uses scale invariant features to estimate the position of the robot within a small operational range. Furthermore, there are different approaches [10, 11, 19] that use techniques also applied for image retrieval to identify the current position of the robot. Whereas all these approaches use ....

S. Se, D. Lowe, and J. Little. Vision-based mobile robot localization and mapping using scale-invariant features. In Proc. of the International Conference on Robotics & Automation (ICRA), 2001.

....correction produces a consistent global 3D map. Our vision based mapping approach using sparse 3D data is di#erent from other existing approaches which use dense 2D range data from laser or sonar rangefinders. 1 Introduction We have proposed a vision based localization and mapping algorithm [10] by tracking Scale Invariant Feature Transform (SIFT) natural landmarks and building a 3D map simultaneously on our mobile robot equipped with Triclops, a trinocular stereo system. However, our algorithm builds a 3D map continuously without maintaining the local image data, and hence does not ....

....more accurate camera ego motion and hence better localization. This will also help adjust the 3D coordinates of the SIFT landmarks for map building. We build a 3D map when the robot moves around in our lab environment and a Kalman Filter is used to track each landmark with a 3x3 covariance matrix [10]. The system currently runs at 2Hz on a Pentium III 700MHz processor. 4 Map Alignment We would like to build submaps of the environment and then align them afterwards to obtain a global map. We consider the alignment of two maps based on the specificity of SIFT features. The algorithm is also ....

[Article contains additional citation context not shown here]

S. Se, D. Lowe, and J. Little. Vision-based mobile robot localization and mapping using scale-invariant features. In Proceedings of the IEEE International Conference on Robotics and Automation (ICRA), pages 2051.

....phase, for each grid position in pose space. Using a panoramic image based model for robot localization is proposed in [3] A panoramic model is constructed with depth and 3D planarity information. The matching is based on the planar patches. We have proposed a vision based SLAM algorithm [12] by tracking SIFT (Scale Invariant Feature Transform) landmarks and building a 3D map simultaneously on our mobile robot equipped with Triclops, a trinocular stereo system. In this paper, we consider global localization as a recognition problem, by matching the distinctive SIFT features detected ....

....for map building. We build a 3D map when the robot moves around in our lab environment. Figure 2 shows the bird s eye view of the map after 435 frames and there are 2783 SIFT landmarks in the database. The system currently runs at 2Hz on a Pentium III 700MHz processor. Readers are referred to [12] for further details. 2.3 Global Localization Global localization is similar to a recognition problem where the robot tries to match the current view to a previously built map. The SIFT features used here were originally designed for object recognition purposes, therefore these visual landmarks ....

[Article contains additional citation context not shown here]

S. Se, D. Lowe, and J. Little. Vision-based mobile robot localization and mapping using scale-invariant features. In Proceedings of the IEEE International Conference on Robotics and Automation (ICRA), pages 2051.

....phase, for each grid position in pose space. Using a panoramic image based model for robot lo calization is proposed in [3] A panoramic model is constructed with depth and 3D planarity information. The matching is based on the planar patches. We have proposed a vision based SLAM algo rithm [12] by tracking SIFT (Scale Invariant Feature Transform) landmarks and building a 3D map simul taneously on our mobile robot equipped with Triclops, a trinocular stereo system. In this paper, we consider global localization as a recognition problem, by matching the distinctive SIFT features ....

....for map building. We build a 3D map when the robot moves around in our lab environment. Figure 2 shows the bird s eye view of the map after 435 frames and there are 2783 SIFT landmarks in the database. The system curtenfly runs at 2Hz on a Pentlure III 700MHz processor. Readers are referred to [12] for further details. 2.3 Global Localization Global localization is similar to a recognition prob lem where the robot tries to match the current view to a previously built map. The SIFT features used here were originally designed for object recognition purposes, therefore these visual ....

[Article contains additional citation context not shown here]

S. Se, D. Lowe, and J. Little. Vision-based mobile robot localization and mapping using scale-invariant features. In Proceedings of the IEEE International Conference on Robotics and Automation (ICRA), pages 2051.

....correction produces a consistent global 3D map. Our vision based mapping approach using sparse 3D data is different from other existing approaches which use dense 2D range data from laser or sonar rangefinders. I Introduction We have proposed a vision based localization and mapping algorithm [10] by tracking Scale Invariant Feature Transform (SIFT) natural landmarks and building a 3D map simultaneously on our mobile robot equipped with Triclops, a trinocular stereo system. However, our algorithm builds a 3D map continuously without maintaining the local image data, and hence does not ....

....more accurate camera ego motion and hence better localization. This will also help adjust the 3D coordinates of the SIFT landmarks for map building. We build a 3D map when the robot moves around in our lab environment and a Kalman Filter is used to track each landmark with a 3x3 covariance matrix [10]. The system currently runs at 2Hz on a Pentium III 700MHz processor. 4 Map Alignment We would like to build submaps of the environment and then align them afterwards to obtain a global map. 154 We consider the alignment of two maps based on the specificity of SIFT features. The algorithm is ....

[Article contains additional citation context not shown here]

S. Se, D. Lowe, and J. Little. Vision-based mobile robot localization and mapping using scale-invariant features. In Proceedings of the IEEE International Conference on Robotics and Automation (ICRA), pages 2051.

....a tracked landmark is a set of image thumbnails detected in the learning phase, for each grid position in pose space. 3] selects image patches in terms of their uniqueness within the local region and dynamic reliability as landmarks for navigation. We have proposed a vision based SLAMB algorithm [17] by tracking SIFT (Scale Invariant Feature Transform) landmarks correcting odometry locally. As our robot is equipped with Triclops [14] a trinocular stereo system, the estimated 3D position of the landmarks can be obtained and hence a 3D map can be built and the robot can be localized ....

....landmarks and use it to match features found in subsequent views. The initial coordinate frame is used as the reference and all landmarks are relative to this fixed frame. Figure 1 show the SIFT detection, stereo matching and frame to frame matching for some typical scene. Readers are referred to [17] for further details. 3 Landmark Uncertainty There are various errors such as noisy sensors and quantization associated with the images and the SIFT features found. They introduce inaccuracy in both the landmarks position as well as the least squares estimation of the robot position. We would ....

S. Se, D. Lowe, and J. Little. Vision-based mobile robot localization and mapping using scale-invariant features. In Proceedings of the IEEE International Conference on Robotics and Automation (ICRA), pages

....shown in black. d) right image: corners (white) in planar regions 4 Feature based Localization Most existing mobile robot localization and mapping algorithms are based on laser or sonar sensors, as vision is more processor intensive and stable visual features are more difficult to extract. In [SLL00] Se, Lowe and Little propose a vision based Simultaneous Localization And Map Building (SLAMB) algorithm by tracking SIFT features, which can be detected at a variety of scales. Trinocular stereo obtains the 3D position of the landmarks, so a 3D map can be built and the robot can be localized ....

Stephen Se, David Lowe, and James J. Little. Vision-based mobile robot localization and mapping using scale-invariant features. submitted to ICRA-00, 2000.

No context found.

S. Se, D. Lowe, and J. Little. Vision-based mobile robot localization and mapping using scaleinvariant features. In Proceedings of the IEEE International Conference on Robotics and Automation (ICRA), pages 2051.

No context found.

S. Se, D. Lowe, and J. Little. Vision-Based Mobile Robot Localization and Mapping Using ScaleIinvariant Features. In Proceedings of the IEEE International Conference on Robotics and Automation (ICRA), pages 2051.

No context found.

S. Se, D. Lowe, and J. Little. Vision-based mobile robot localization and mapping using scale-invariantf3 ures. n Proc. of the International Conference on Robotics & Automation (ICRA), 2001. 109

No context found.

S. Se, D. Lowe, and J. Little. Vision-based mobile robot localization and mapping using scale-invariantf3 ures. n Proc. of the International Conference on Robotics & Automation (ICRA), 2001. 109

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