B. Schiele and J. Crowley. A comparison of position estimation techniques using occupancy grids. IEEE International Conference on Robotics and Automation, pages 1628-- 1634, 1994.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....from sensor data and matching them with a representation of the environment in terms of these features. Lines are often used for localization in polygonal environment. For instance, in [5] an algorithm for computing robot poses by line matching is presented, while the localization method in [6] is based on certainty grids and on the use of the Hough Transform only for extracting lines from these points. In this paper we present a method called Global Hough Localization, that extends the Hough Localization method described in [7] 8] in order to provide a solution for global ....

....have shown the e ectiveness and eciency of the approach for computing the absolute pose of the robot in polygonal environments. Similar previous works include the method in [5] that presents an algorithm for matching lines extracted by the sensors with reference lines, while the work in [6]isbasedontwo certainty grids (one containing data from the reference map, the other containing data acquired by a range sensor) and uses the Hough Transform for detecting segments from the two grids, while the association problem is addressed by comparing in the Cartesian space each pair of ....

B. Schiele and J. Crowley, \A comparison of position estimation techniques using occupancy grids," ######## ### ########## #######,vol. 12, pp. 163-171, 1994.

....such as localization, path planning, collision avoidance, and people finding. The basic occupancy grid map paradigm has been applied successfully in many different ways. For example, some systems use maps locally, to plan collision free paths or to identify environment features for localization [1, 10, 19, 20]. Others, such as many of the systems described in [11, 23] rely on global occupancy grid maps for global path planning and navigation. f Figure 1: A set of noise free sonar measurements that a robot may receive while passing an open door. While the measurements are perfectly consistent, ....

B. Schiele and J. Crowley. A comparison of position estimation techniques using occupancy grids. In Proceedings of the 1994.

....Each line indicates the range, as measured by the sensor. Also shown is an outline of the environment, which suggests that the range measurements are of high quality. The measurement model adopted in our implementation is a probabilistic generalization of the rich literature on scan matching [34, 54, 74]. It assumes that each scan induces a local map, which can be conceptually decomposed into three types of areas: free space, occupied space, and occluded space. The same conceptual decomposition applies to the map. Each of the measurements in a range scan z t can thus fall into three di erent ....

B. Schiele and J. Crowley. A comparison of position estimation techniques using occupancy grids. In Proceedings of the 1994.

....Each line indicates the range, as measured by the sensor. Also shown is an outline of the environment, which suggests that the range measurements are of high quality. The measurement model adopted in our implementation is a probabilistic generalization of the rich literature on scan matching [24, 39, 57]. It assumes that each scan induces a local map, which can be conceptually decomposed into three types of areas: free space, occupied space, and occluded space. The same conceptual decomposition applies to the map. Each of the measurements in a range scan z t can thus fall into three different ....

B. Schiele and J. Crowley. A comparison of position estimation techniques using occupancy grids. In Proceedings of the 1994.

....order of difficulty: 1. Position tracking. Here the initial robot pose is known, and the goal of localization is to compensate small odometry error as the robot moves. Typically, the uncertainty in position tracking is local, making unimodal state estimators such as Kalman filters applicable [2, 44, 61, 85]. 5 2. Global localization. If the robot does not know its initial pose, it faces a global localization problem. To localize itself from scratch, a robot must be able to cope with ambiguities and multiple beliefs during localization. 3. Robot kidnapping [35] This problem is a variant of the ....

B. Schiele and J. Crowley. A comparison of position estimation techniques using occupancy grids. In Proceedings of the 1994.

.... the most fundamental problem to providing a mobile robot with autonomous capabilities [8] The mobile robot localization problem comes in many different flavors [4,24] The most simple localization problem which has received by far the most attention in the literature is position tracking [4,64,74,75]. Here the initial robot pose is known, and Corresponding author. E mail address: thrun cs.cmu.edu (S. Thrun) 0004 3702 01 see front matter 2001 Published by Elsevier Science B.V. PII: S0004 3702(01)00069 8 the problem is to compensate incremental errors in a robot s odometry. ....

B. Schiele, J. Crowley, A comparison of position estimation techniques using occupancy grids, in: Proc. 1994.

....represent the beliefs p(l) and p(s) of the robot s position and the state of the object. Over the past years, different techniques have been used to represent the beliefs. Among them are piecewise constant approximations as applied in [3,4,12,15,19,25] A very popular approach is to use Gaussians [10,16,24,26] to represent the densities. In this paper we use particle filters to approximate the involved densities. The key idea of particle filters is to use a sample based representation for the densities. The updates are carried out using resampling techniques (see e.g. 6,11,21] To estimate the ....

B. Schiele, J.L. Crowley, A comparison of position estimation techniques using occupancy grids, in: Proceedings of the IEEE International Conference on Robotics and Automation (ICRA), San Diego, CA, 1994.

....distribution. Sensor readings, too, are assumed to map to Gaussian shaped distributions over the robot s position. For these assumptions, Kalman filters provide extremely efficient update rules that can be shown to be optimal (relative to the assumptions) 27] Kalman filter based techniques [28, 29, 30] have proven to be robust and accurate for keeping track of the robot s position. However, these techniques do not represent multi modal probability distributions, which frequently occur during global localization. In practice, localization approaches using Kalman filters typically require that ....

B. Schiele and J.L. Crowley. A comparison of position estimation techniques using occupancy grids. In Proc. of the IEEE International Conference on Robotics & Automation (ICRA), 1994.

....to questions regarding position esD R A F T November 30, 1999, 7:52pm D R A F T 6 timates. The search for the best position estimate can be conducted in two different spaces: 1) the space of correspondences between image and map features [2, 24, 25] or (2) the space of position estimates [7, 8, 21, 40, 49]. The search in the space of position estimates is often conducted by quantizing this space into a two dimensional array, an approach inspired by the occupancy grids developed for map making [13, 14, 34, 35] Position can also be estimated qualitatively: instead of precise geometric estimates, ....

B. Schiele and J. L. Crowley. A comparison of position estimation techniques using occupancy grids. IEEE International Conference on Robotics and Automation, pages 1628--1634, 1994.

....by comparing sensor data, and data obtained from a previously built representation of the environment. Several solutions have been proposed for environment representation. One of them consists of using grid cells, which allows a simple environment representation to be built from sensor data. [4,18,20]. Another solution consists of using topological maps, in which the environment is represented by means of objects and graphs; 6,7,9,14,16] path planning becomes easier, but learning and sensor integration becomes more difficult. Different neural network approaches have been used to maintain the ....

Schiele, B. and Crowley, J., 1994, "A Comparison of Position Estimation Techniques Using Occupancy Grids." Robotics and Autonomous Systems, Vol. 12, pp. 163-171.

....small odometric errors. Thus, they differ from the approach described here in that they require knowledge of the robot s initial position; and they are not able to recover from global localizing failures. Probably the most popular method for tracking a robot s position is Kalman filtering [28, 29, 46, 48, 59, 63], which represents uncertainty by the first and second moments of the density. These approaches are unable to localize robots under global uncertainty a problem which Engelson called the kidnapped robot problem [19] Recently, several researchers proposed Markov localization, which enables ....

B. Schiele and J.L. Crowley. A comparison of position estimation techniques using occupancy grids. In Proc. of the IEEE International Conference on Robotics & Automation (ICRA), 1994.

....specially regarding dynamic environments: i) they can be quickly updated; and ii) small changes in the environment tend to result in small changes in the grid. To minimise the problem instance, instead of working with complete grids most methods search for signi# cant structures in local ones [2][6] but they are usually limited to static environments. Thrun [7] uses evidence grids in dynamic environments, but it is assumed that walls are either parallel or perpendicular to each other and, therefore, the method is not valid for unstructured environments. Yamauchi [8] deals with dynamic ....

....A sonar reading typically provides information about the distance of the sensor to the closest obstacle in the direction of the beam. Since sonars present an arc of uncertainty, most systems rely on accumulating evidence from several ones, which are usually integrated into an evidence grid [6]. If a mobile robot builds an evidence grid while it knows its correct position, it can compare local grids adquired after losing its reference to the global one to locate itself. In order to achieve a grid accurate enough to represent most signi#cant features within a typical indoor environment, ....

Schiele, B. and Crowley, J. "A comparison of position estimation techniques using occupancy grids", Robotics and autonomous systems, 12, pp. 163171, 1994

....order of di#culty: 1. Position tracking. Here the initial robot pose is known, and the goal of localization is to compensate small odometry error as the robot moves. Typically, the uncertainty in position tracking is local, making unimodal state estimators such as Kalman filters applicable [2, 44, 61, 85]. 5 2. Global localization. If the robot does not know its initial pose, it faces a global localization problem. To localize itself from scratch, a robot must be able to cope with ambiguities and multiple beliefs during localization. 3. Robot kidnapping [35] This problem is a variant of the ....

B. Schiele and J. Crowley. A comparison of position estimation techniques using occupancy grids. In Proceedings of the 1994 IEEE International Conference on Robotics and Automation, pages 1628--1634, San Diego, CA, May 1994.

....between adjacent partitions. Then, whenever the robot is in the vicinity of the boundaries of one of the partitions, it uses the correlation of the local grid constructed on line as the robot moves , with the learned local model to correct possible errors of the dead reckoning system [10], 12] ....

B. Schiele and J. Crowley. "A comparison of position estimation techniques using occupancy grids". IEEE Int. Conf. on Robotics and Automation, pp. 1628-- 1634, 1994.

....order of difficulty: 1. Position tracking. Here the initial robot pose is known, and the goal of localization is to compensate small odometry error as the robot moves. Typically, the uncertainty in position tracking is bounded, making unimodal state estimators such as Kalman filters applicable [2, 57, 84, 120]. 5 2. Global localization. If the robot does not know its initial pose, it faces a global localization problem. To localize itself from scratch, a robot must be able to cope with ambiguities and multiple, competing hypotheses during localization. 3. Robot kidnapping [46] This problem is a ....

....the robot in a metric space, just like those methods proposed in [10, 141, 144] The vast majority of approaches is incapable of localizing a robot globally or to recover from robot kidnapping. Instead, they are designed to track the robot s position by compensating small odometric errors [56, 57, 89, 120, 134]. Recently, several researchers proposed the Markov localization used by Minerva, which enables robots to localize themselves under global uncertainty [20, 35, 48, 68, 101, 132] Minerva s and Rhino s localization algorithm goes beyond previous approaches in that it can cope with invisible hazards ....

B. Schiele and J. Crowley. A comparison of position estimation techniques using occupancy grids. In Proceedings of the 1994 IEEE International Conference on Robotics and Automation, pages 1628-- 1634, San Diego, CA, May 1994.

.... the most fundamental problem to providing a mobile robot with autonomous capabilities [6] The mobile robot localization problem comes in many different flavors [2,21] The most simple localization problem which has received by far the most attention in the literature is position tracking [2,62,52,61]. Here the initial robot pose is known, and the problem is to compensate small, incremental errors in a robot s Preprint submitted to Elsevier Preprint 3 December 2000 odometry. Algorithms for position tracking often make restrictive assumptions on the size of the error and the shape of the ....

B. Schiele and J. Crowley. A comparison of position estimation techniques using occupancy grids. In Proceedings of the 1994 IEEE International Conference on Robotics and Automation, pages 1628--1634, San Diego, CA, May 1994.

....and then merge these maps into a larger grid after applying some sort of position correction, perhaps derived from an attempt to match the data in the global grid to the data in the local grid in order to determine the best fit to what is already there. For example, Schiele and Crowley [43] take just such an approach. Their SONARbased robot maintains a global grid, which in effect maintains a picture of the environment as the robot believes it to be (including physical positions at which obstacles have been detected) The robot, of course, also maintains and updates a notion of ....

....terms for, e.g. change in wheel diameter (i.e. tire wear) however, they do, in practice manage to perform with pretty high accuracy. 2.6. 2 The Kalman filter applied to grids Having described the Kalman filter, let use return to the dual grid robot mapping system of Schiele and Crowley [43]. Their robot is a wheeled car like device with the sensors mounted on it. Now, a completely accurate state model for the robot is, of course, nonlinear, requiring the use of the EKF. The operation of the robot is, as described above, divided into discrete time intervals, each of which involves a ....

Bernt Schiele and James Crowley. A comparison of position estimation techniques using occupancy grids. Robotics and Autonomous Systems, 12:163--171, 1994.

....significant places. Fortunately, this drawback can be eliminated by introducing a two layered representation in which a global occupancy grid map is built from a set of local maps generated from short sequences of sensor data [Burgard et al. 1999] a technique also used in [Yamauchi, 1996; Schiele and Crowley, 1994] Examples of such local maps are shown in Figure 7. For such maps, which replace the landmarks used above, we assume that the odometric error can be neglected since the robot only moves a short distance. In the E step, the localization is not carried out with respect to a global map but rather ....

B. Schiele and J.L. Crowley. A comparison of position estimation techniques using occupancy grids. In Proc. of the IEEE International Conference on Robotics & Automation (ICRA), 1994.

....small odometric errors. Thus, they differ from the approach described here in that they require knowledge of the robot s initial position. Furthermore, they are not able to recover from global localizing failures. Probably the most popular method for tracking a robot s position is Kalman filtering [15, 20, 21, 26, 28], which represents the belief by a uni modal Gaussian distribution. These approaches are unable to localize robots under global uncertainty. Recently, several researchers proposed Markov localization, which enables robots to localize themselves under global uncertainty [6, 16, 23, 27] Global ....

B. Schiele and J.L. Crowley. A comparison of position estimation techniques using occupancy grids. In Proc. of the IEEE International Conference on Robotics & Automation (ICRA), 1994.

....for the representation of the robots belief ###(#) is based on Kalman ltering (Kalman, 1960; Smith et al. 1990) which rests on the restrictive assumption that the position of the robot can be modeled by a unimodal Gaussian distribution. Existing implementations (Leonard Durrant Whyte, 1992; Schiele Crowley, 1994; Gutmann Schlegel, 1996; Arras Vestli, 1998) have proven to be robust and accurate for keeping track of the robot s position. Because of the restrictive assumption of a Gaussian distribution these techniques lack the abilitytorepresent situations in which the position of the robot 397 Fox, ....

....sonar scans with beacons predicted from a geometric map of the environment. These beacons consist of planes, cylinders, and corners. To update the current estimate of the robot s position, Cox (1991) matches distances measured by infrared sensors with a line segment description of the environment. Schiele and Crowley (1994) compare di erent strategies to track the robot s position based on occupancy grid maps and ultrasonic sensors. They show that matching local occupancy grid maps with a global grid map results in a similar localization performance as if the matching is based on features that are extracted from ....

B. Schiele and J.L. Crowley. A comparison of position estimation techniques using occupancy grids. In Proc. of the IEEE International Conference on Robotics & Automation (ICRA), 1994. 426 Markov Localization for Mobile Robots in Dynamic Environments

....small odometric errors. Thus, they differ from the approach described here in that they require knowledge of the robot s initial position; and they are not able to recover from global localizing failures. Probably the most popular method for tracking a robot s position is Kalman filtering [28, 29, 46, 48, 59, 63], which represents uncertainty by the first and second moments of the density. These approaches are unable to localize robots under global uncertainty a problem which Engelson called the kidnapped robot problem [19] Recently, several researchers proposed Markov localization, which enables ....

B. Schiele and J.L. Crowley. A comparison of position estimation techniques using occupancy grids. In Proc. of the IEEE International Conference on Robotics & Automation (ICRA), 1994.

....and most walls have the same textures, anywhere in the environment. Random visual inspection of the occupancy data indicate that the 218 occupancy grids, used in generating the training data, is an encompassing representation of all possible situations. Simmons [52] Cassandra [22] and Schiele [53] use hard coded feature detectors. One advantage to using neural nets to compute high level features is that the robot can be easily trained to work in different environments, where walls have a different texture. This is facilitated by a short training time and the fact that data collection and ....

B. Schiele and J. Crowley, "A comparison of position estimation techniques using occupancy grids," IEEE Conference on Robotics and Autonomous Systems, pp. 1628--1634, May 1994.

....for the representation of the robots belief Bel(L) is based on Kalman ltering (Kalman, 1960; Smith et al. 1990) which rests on the restrictive assumption that the position of the robot can be modeled by a unimodal Gaussian distribution. Existing implementations (Leonard Durrant Whyte, 1992; Schiele Crowley, 1994; Gutmann Schlegel, 1996; Arras Vestli, 1998) have proven to be robust and accurate for keeping track of the robot s position. Because of the restrictive assumption of a Gaussian distribution these techniques lack the ability to represent situations in which the position of the robot 397 Fox, ....

....sonar scans with beacons predicted from a geometric map of the environment. These beacons consist of planes, cylinders, and corners. To update the current estimate of the robot s position, Cox (1991) matches distances measured by infrared sensors with a line segment description of the environment. Schiele and Crowley (1994) compare di erent strategies to track the robot s position based on occupancy grid maps and ultrasonic sensors. They show that matching local occupancy grid maps with a global grid map results in a similar localization performance as if the matching is based on features that are extracted from ....

B. Schiele and J.L. Crowley. A comparison of position estimation techniques using occupancy grids. In Proc. of the IEEE International Conference on Robotics & Automation (ICRA), 1994. 426 Markov Localization for Mobile Robots in Dynamic Environments

....compares two different and popular localization techniques for mobile robots: Markov localization, which represents arbitrary probability distributions across a grid of robot poses, and Kalman filtering which uses normal distributions together with scan matching. Previous work reported in [18, 16, 9] largely focuses on the 4 Recently, an extension of Markov localization has been described [8] which is designed to filter out those measurements that are reflected by obstacles not contained in the map and thus shows a better performance than the version used in the experiments described here. ....

B. Schiele and J. L. Crowley. A comparison of position estimation techniques using occupancy grids. In Proc. of the IEEE International Conference on Robotics and Automation, pages 1628--1634, 1994.

....of images of the interior space. By comparing these prerecorded images with the camera images taken during navigation, the robot is able to determine its location. Other previous research contributions that are relevant to mobile robot localization include [4] 5] 8] 12] 14] 16] and [17]. Manuscript received March 5, 1997; revised June 17, 1998. This paper was recommened for publication by Associate Editor R. Chatila and Editor V. Lumelsky upon evaluation of the reviewers comments. A. Ohya is with the Intelligent Robot Laboratory, Institute of Information Sciences and ....

B. Schiele and J. L. Crowley, "A comparison of position estimation techniques using occupancy grids," Robot. Auton. Syst., vol. 12, pp. 163--171, 1994.

....localization has been intensively studied in the past and a variety of systems and techniques have been developed (see [2] for a comprehensive overview) Recently there has been increasing interest in probabilistic methods. Several systems apply Kalman filters to keep track of the robot s position [10, 13, 7, 1]. Kalman filter techniques have been proven to be robust and accurate for keeping track of the robot s position. However, they cannot represent ambiguities and lack the ability to globally (re )localize the robot in the case of localization failures. To overcome these disadvantages, recently ....

B. Schiele and J. L. Crowley. A comparison of position estimation techniques using occupancy grids. In Proc. of the IEEE International Conference on Robotics and Automation, pages 1628--1634, 1994.

....Wei et al. 18] store angle histograms constructed out of range finder scans taken at different locations of the environment. The position and orientation of the robot is calculated by maximizing the correlation between histograms of new measurements with the stored histograms. Schiele and Crowley [14] compare different strategies to track the robot s position based on occupancy grid maps. They use two different maps: a local grid computed using the most recent sensor readings, and a global map built during a previous exploration of the environment or by an appropriate CADtool. The local map is ....

Bernt Schiele and James L. Crowley. A comparison of position estimation techniques using occupancy grids. In Proc. of the IEEE International Conference on Robotics and Automation, pages 1628--1634, 1994.

....can be described using a Gaussian density, and the initial state is also specified as a Gaussian, then the density p(x k jZ k ) will remain Gaussian at all times. In this case, 1) and (2) can be evaluated in closed form, yielding the classical Kalman filter [27] Kalman filter based techniques [24, 31, 14] have proven to be robust and accurate for keeping track of the robot s position. Because of its concise representation (the mean and covariance matrix suffice to describe the entire density) it is also a particularly efficient algorithm. However, it is clear that the basic assumption of Gaussian ....

B. Schiele and J. L. Crowley. A comparison of position estimation techniques using occupancy grids. In IEEE Conf. on Robotics and Automation (ICRA), volume 2, pages 1628--1634, 1994.

....described here in that they require knowledge of the robot s initial position; and they are not able to recover from global localizing failures. Probably the most popular method for tracking a robot s position is Kalman filtering [77,102] which represent uncertainty by single modal distributions [60,61,98, 125,140]. While these approaches are computationally extremely fast, they are unable to localize robots under global uncertainty a problem which Engelson called the kidnapped robot problem [44] Recently, several researchers proposed Markov localization,which enables robots to localize themselves ....

B. Schiele, J. Crowley, A comparison of position estimation techniques using occupancy grids, in: Proc. 1994 IEEE International Conference on Robotics and Automation, San Diego, CA, 1994, pp. 1628--1634.

....is a critical element for e ectively accomplishing complex tasks requiring autonomous navigation. The localization problem has been thus addressed in the past from many di erent perspectives. In particular, absolute positioning methods based on map matching have been extensively studied (see [4, 10] for occupancy grid matching strategies, 8] for the angle histogram method, 3] for a probabilistic approach, 7] for scan matching techniques, and [6] for experimental comparisons) They present di erent solutions that are generally robust to sensor noise, ambiguous situations, partial model ....

B. Schiele and J. Crowley. A comparison of position estimation techniques using occupancy grids. Robotics and Autonomous Systems, 12, 1994.

....element for e ectively accomplishing complex tasks requiring autonomous navigation. The localization problem has been thus addressed in the past from many di erent perspectives. In particular, absolute positioning methods based on map matching have been extensively studied (see [ Cox, 1991; Schiele and Crowley, 1994 ] for occupancy grid matching strategies, Hinkel and Knieriemen, 1988 ] for the angle histogram method, Burgard et al. 1996 ] for a probabilistic approach, Gutmann and Schlegel, 1996 ] for scan matching techniques, and [ Gutmann et al. 1998 ] for experimental comparisons) They present ....

B. Schiele and J. Crowley. A comparison of position estimation techniques using occupancy grids. Robotics and Autonomous Systems, 12, 1994.

....CAD map of the environment. He assigns scan points to line segments based on closest neighborhood and then searches for a translation and rotation that minimizes the total squared distance between scan points and their target lines. Scans can also be matched by correlating occupancy grids [21, 22, 15]. However accuracy and run time performance depend heavily on the grid resolution. Weiss et al. 27] use histograms for matching a pair of scans. They first compute a so called angle histogram for determining the rotation of the two scans and then use x and y histograms for computing the ....

B. Schiele and J. L. Crowley. A comparison of position estimation techniques using occupancy grids. In Proc. International Conference on Robotics and Automation (ICRA'94), pages 1628--1634, 1994.

....mapping, and navigation given the nature of the SSH framework. 1 D relations can simply be represented as attributes. A 2 D description can be represented in one of the following ways: ffl generalized cylinders for path segments [Brooks, 1981] ffl occupancy grids [Moravec and Elfes, 1985; Schiele and Crowley, 1994] ffl feature maps [Leonard and Durrant Whyte, 1992] Local metrical knowledge typically consists of headings and distances defined within local frames of references: local heading(view; place; heading) local heading of view at place is heading. path distance(path; p 1 ; p 2 ; distance) distance ....

Bernt Schiele & James L. Crowley, "A comparison of position estimation techniques using occupancy grids," in Proc. IEEE International Conference on Robotics and Automation, 1994, pp. 1628--1634.

....landmark based localization can be found [5,21,52,47,50,74,76,80,95,111] and various chapters in [53] Model matching. Model matching algorithms extract geometric features from the sensor readings and match those to a model of the environment in order to identify errors in the robot s odometry [9 12,15,22,85,90,99,101,109]. Among the earliest work in this field is that of Moravec, Elfes, and Chatila Laumond. Chatila and Laumond s approach [15] extracts geometric features such as line segments and polyhedral objects which are matched to a geometric map. Moravec and Elfes, who pioneered the development of occupancy ....

B. Schiele and J. Crowley. A comparison of position estimation techniques using occupancy grids. In Proceedings of the 1994 IEEE International Conference on Robotics and Automation, pages 1628--1634, San Diego, CA, May 1994.

....here in that they require knowl41 edge of the robot s initial position; and they are not able to recover from global localizing failures. Probably the most popular method for tracking a robot s position is Kalman filtering [77,102] which represent uncertainty by single modal distributions [60,61,98,125,140]. While these approaches are computationally extremely fast, they are unable to localize robots under global uncertainty a problem which Engelson called the kidnapped robot problem [44] Recently, several researchers proposed Markov localization, which enables robots to localize themselves ....

B. Schiele and J. Crowley. A comparison of position estimation techniques using occupancy grids. In Proceedings of the 1994 IEEE International Conference on Robotics and Automation, pages 1628--1634, San Diego, CA, May 1994.

....from a geometric map of the environment. These beacons consist of planes, cylinders, and corners. To update the current estimate of the robot s position, Cox [9] matches distances measured by infrared range finders against a line segment description of the environment. Schiele and Crowley [27] compare different strategies to track the robots position based on occupancy grid maps and ultrasonic sensors. They show that matching local occupancy grid maps against a global grid map results in a similar localization performance as if the matching is based on features that are extracted from ....

B. Schiele and J.L. Crowley. A comparison of position estimation techniques using occupancy grids. In Proc. of the IEEE International Conference on Robotics and Automation, 1994.

....Wetzler, von Puttkamer 1994) store angle histograms constructed out of range finder scans taken at different locations of the environment. The position and orientation of the robot is calculated by maximizing the correlation between histograms of new measurements and the stored histograms. (Schiele Crowley 1994) compare different strategies to track the robot s position based on occupancy grid maps. They use two different maps: a local grid computed using the most recent sensor readings, and a global map built during a previous exploration of the environment or by an appropriate CAD tool. The local map ....

Schiele, B., and Crowley, J. L. 1994. A comparison of position estimation techniques using occupancy grids. In Proc. of the IEEE International Conference on Robotics and Automation, 1628--1634.

....by the occupancy grid and the function approximated by the local bitmap translated by (x; y) and rotated by . Drawbacks of this approach are the computational complexity of correlation as well as the trade off between efficiency and precision, embodied in the choice of grid resolution. In [61] it is concluded that even for systems using an occupancy grid representation, more reliable position estimates are obtained by extracting segments from the grids and performing segment segment matching. Of course, extracting features such as segments from an occupancy grid can be computationally ....

B. Schiele and J. L. Crowley. A comparison of position estimation techniques using occupancy grids. In Proceedings of the IEEE International Conference on Robotics and Automation, pages 1628--1634, May 1994.

....small odometric errors. Thus, they differ from the approach described here in that they require knowledge of the robot s initial position; and they are not able to recover from global localizing failures. Probably the most popular method for tracking a robot s position is Kalman filtering [30, 31, 48, 50, 61, 65], which represent uncertainty by single modal 20 Dieter Fox, Wolfram Burgard, Hannes Kruppa, and Sebastian Thrun distributions. These approaches are unable to localize robots under global uncertainty a problem which Engelson called the kidnapped robot problem [20] Recently, several ....

B. Schiele and J.L. Crowley. A comparison of position estimation techniques using occupancy grids. In Proc. of the IEEE International Conference on Robotics and Automation, pages 1628--1634, 1994.

No context found.

B. Schiele and J.L. Crowley. A comparison of position estimation techniques using occupancy grids. Robotics and Autonomous Systems, 12:163--171, 1994. 15

No context found.

B. Schiele and J. Crowley. A comparison of position estimation techniques using occupancy grids. IEEE International Conference on Robotics and Automation, pages 1628-- 1634, 1994.

No context found.

B. Schiele and J. Crowley, "Comparison of position estimation techniques using occupancy grids," in Proc of International Conference on Robotics and Automation, pp. 1628--1634, May 1994.

No context found.

B. Schiele and J. Crowley. A comparison of position estimation techniques using occupancy grids. Robotics and Autonomous Systems, 12:163-171, 1994.

No context found.

B. Schiele and J. Crowley. A comparison of position estimation techniques using occupancy grids. Robotics and Autonomous Systems, 12, 1994.

No context found.

Bernt Schiele and James L. Crowley. A comparison of position estimation techniques using occupancy grids. In Proceedings of the 1994 IEEE International Conference on Robotics and Automation, pages 1628--1634, San Diego, CA, May 1994.

No context found.

Schiele, B. and Crowley, J. A Comparison of Position Estimation Techniques Using Occupancy Grids. in: Proceedings of the 1994.

No context found.

B. Schiele and J. L. Crowley, \A comparison of position estimation techniques using occupancy grids," in Proceedings of the International Conference on Robotics and Automation, vol. 2, pp. 1628{ 1634, IEEE, May 1994.

No context found.

Bernt Schiele and James L. Crowley. A comparison of position estimation techniques using occupancy grids. In Proc. of the IEEE International Conference on Robotics and Automation, pages 1628--1634, 1994.

No context found.

Schiele, B. and Crowley, J., 1994, "A Comparison of Position Estimation Techniques Using Occupancy Grids." Proceedings of IEEE International Conference on Robotics and Automation, San Diego, CA, May 8-13, pp. 1628-1634.

No context found.

B. Schiele and J. Crowley. A comparison of position estimation techniques using occupancy grids. In Proceedings of ICRA, 1994.

First 50 documents  Next 50

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