Levitt, T. S. and Lawton, D. T. (1990). Qualitative navigation for mobile robots. Artificial Intelligence, 44(3):305--360.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....demand very reliable and accurate sensor devices. In contrast, over the years, qualitative approaches to robotic navigation have been proposed to overcome problems regarding complexity and stability. Qualitative modelling for self localization and navigation includes the qualitative map approach [18, 20] and adaptive topological models [27] They are based on a semantic hierarchy of descriptions for representation, with the most abstract level being that of the topological neighbourhood . These approaches concern spatial learning and path planning in the absence of a single global coordinate ....

T S Levitt and D T Layto, `Qualitative navigation for mobile robots', Artificial Intelligence, 44, 305-- 360, (1990).

....robot can find its way to a target provided it is represented in the network. Prescott s approach is quantitative whereas our approach uses a fuzzy extension of the betacoefficient coding system in order to work with fuzzy qualitative information about distances and directions. Levitt and Lawton [10] also proposed a qualitative approach to the navigation problem but that assumes an unrealistically accurate distance and direction information between the robot and the landmarks. Another qualitative method for robot navigation was proposed by Escrig and Toledo [5] using constraint logic. ....

T.S. Levitt and D.T. Lawton. Qualitative navigation for mobile robots. Artificial Intelligence Journal, 44:305--360, 1990.

....robot can find its way to a target provided it is represented in the network. Prescott s approach is quantitative whereas our approach uses a fuzzy extension of the betacoefficient coding system in order to work with fuzzy qualitative information about distances and directions. Levitt and Lawton [13] also proposed a qualitative approach to the navigation problem but they also assume an unrealistically accurate distance and direction information between the robot and the landmarks. Another qualitative method for robot navigation was proposed by Escrig and Toledo [9] using constraint logic. ....

Levitt, T. S and Lawton, D. T, 1990. Qualitative navigation for mobile robots. Artificial Intelligence Journal, 44, 305--360.

....circular: Is the cognitive map rich enough to make localization possible Robotics research has tried considerably to provide solutions to the localization or positioning problem, as it is also known. Nevertheless, the different sensory modalities of the robots admit different Levitt and Lawton [LeLa90] define also navigation with a similar set of questions. solutions and there is no general method for all cases. The most significant work in the area is included in [BoEF95] Path planning is perhaps the most famous and the most important of the problems related to navigation, to the extent ....

Levitt, Tod S. and Daryl T. Lawton, "Qualitative Navigation for Mobile Robots," Artificial Intelligence 44, 3 (1990), 305--360.

.... batch estimation techniques [10] 5] 13] Still other approaches to the problem of map building and localisation have done away with the rigorous mathematical models of the vehicle and sensing properties and have relied instead on more qualitative knowledge of the nature of the environment [1] 6][9]. While all of these alternative approaches to the problem have their own particular strengths, this paper will be concerned primarily with a recursive, on line approach to the problem and will rely on the EKF as the primary means of simultaneously building a map while localising the vehicle. ....

T.S. Levitt and D.T. Lawton. Qualitative navigation for mobile robots. Artificial Intelligence Journal, 44(3):305--360, 1990.

....of the environment in which the vehicle operated. Still other approaches to the problem of map building and localisation have done away with the rigorous mathematical models of the vehicle and sensing properties and have relied instead on more qualitative knowledge of the nature of the environment [2, 15, 19]. These methods have a certain appeal in that they eliminate the need for accurate models of the vehicle motion and sensing processes, limit the computational requirements of map building and have a certain anthropomorphic appeal. While all of these alternative approaches to the problem have their ....

T.S. Levitt and D.T. Lawton. Qualitative navigation for mobile robots. Artificial Intelligence Journal, 44(3):305--360, 1990.

....defined by the intersections of the lines of sight between the objects in the environment. This tessellation is a map of the environment within which the robot can locate itself and plan paths through the world. This process is similar to the construction of orientation regions (Levitt and Lawton, [12]) A suitable data structure must be created in order to store the lines of sight and the tessellation of the environment, maintaining the order among the lines and the regions, so that path planning can be reduced to efficient search. This data structure might be similar to a quad tree, modified ....

Levitt, T. S. and Lawton, D. T. (1990) "Qualitative Navigation for Mobile Robots", Artificial Intelligence, Vol. 44, pages 305-360

.... instead, like brightness patterns or other image features only loosely related to the objects [20] 5] We believe that for an application like patrolling or remote surveillance the human readability is a must, so we selected features closely related to the objects in the environment, as in [9]. Vertical edges present a double advantage: in a man made environment like an o#ce or a building, they are di#usely present and they are Examples of vertical edges are doors, the sides of a cabinet, easy to extract from the image. In fact, as mentioned in Paragraph III, the vertical edges ....

T. S. Levitt and D. T. Lawton. Qualitative navigation for mobile robots. Artificial Intelligence Journal, 44(3):305-- 361, 1990.

.... Ap of Happens and a temporal ordering formulae such that This logical characterization of the event calculus planning is analgous to Green s logical characterization of situational calculus planning [Green, 1969] 10.3 Qualitative Representation of Space Qualnav model [Kuipers and Levitt, 1988, Levitt and Lawton, 1990] The Qualnav model provides a computable theory that integrates qualitative, topological representations of large scale space with quantitative, metric ones. A difference from the TOUR model, Qualnav considers unstructured environments, with significant perceptual events, called landmarks, ....

T. S. Levitt and D. T. Lawton. Qualitative navigation for mobile robots. Artificial intelligence, 44(3):305-360, 1990.

.... temporal ordering formulae such that # # ## # # # ###: This logical characterization of the event calculus planning is analgous to Green s logical characterization of situational calculus planning [Green, 1969] 10.3 Qualitative Representation of Space Qualnav model [Kuipers and Levitt, 1988, Levitt and Lawton, 1990] The Qualnav model provides a computable theory that integrates qualitative, topological representations of large scale space with quantitative, metric ones. A difference from the TOUR model, Qualnav considers unstructured environments, with significant perceptual events, called landmarks, ....

T. S. Levitt and D. T. Lawton. Qualitative navigation for mobile robots. Artificial intelligence, 44(3):305--360, 1990.

....robot can find its way to a target provided it is represented in the network. Prescott s approach is quantitative whereas our approach uses a fuzzy extension of the betacoefficient coding system in order to work with fuzzy qualitative information about distances and directions. Levitt and Lawton [9] also proposed a qualitative approach to the navigation problem but assumes an unrealistically accurate distance and direc2 tion information between the robot and the landmarks. Another qualitative method for robot navigation was proposed by Escrig and Toledo [4] using constraint logic. However, ....

T.S. Levitt and D.T. Lawton. Qualitative navigation for mobile robots. Artificial Intelligence Journal, 44:305--360, 1990.

....perceived landmark with the network, the robot can find its way to a target, provided it is represented in the network. Whereas Prescott s approach is quantitative, ours uses a fuzzy extension of his model to work with fuzzy qualitative information about distances and directions. Levitt and Lawton [10] also proposed a qualitative approach to the navigation problem, but assume unrealistically accurate distance and direction information between the robot and the landmarks. Another qualitative method for robot navigation was proposed by Escrig and Toledo [7] using constraint logic. However, they ....

T.S. Levitt and D.T. Lawton. Qualitative navigation for mobile robots. Artificial Intelligence Journal, 44:305--360, 1990.

....noise remains an open problem. The question is further complicated by the fact that the choice of an optimal mapping strategy is sensitive to the specific task at hand. A somewhat distinct research stream deals with the complexity issues in autonomous robot exploration of an unknown environment [5, 16, 21, 17]. In general, work on sensor fusion has tended to focus on issues of how best to combine measurements from di#erent sensors e.g. 24, 6] or how best to extract data with a single sensor and fuse the measurements over time e.g. 11, 12, 25] rather than how to selectively extract measurements from ....

T. Levitt and D. Lawton. Qualitative navigation for mobile robots. Artificial Intelligence, 44(3):305--360, August 1990.

.... landmark B is on the right, the viewpoint could be at either the upper star (top view) or the lower star (bottom view) 2 Background When three point landmarks in a view have been ordered and matched to a map, those landmarks can, in most cases, be used to uniquely determine viewpoint location [Levitt and Lawton, 1990, Krotkov, 1989, Sugihara, 1988] The visual angle from the viewpoint to one pair of landmark points is measured. This measurement constrains the viewpoint to the locus of a circle passing through the two points and the viewpoint. The third landmark is then paired with one of the others and an ....

....on angle measure can limit the viewpoint location. The LPB s divide the plane into orientation regions. Knowing the orientation region in which the viewpoint lies is, in this case, equivalent to knowing the order of LPB s (landmark pair boundaries) consist of the lines joining pairs of landmarks [Levitt and Lawton, 1990] . the landmarks in the view [Levitt and Lawton, 1990] Table 1 shows the possible orientation regions for the viewpoint location in Figure 3 if the position of only one landmark is known. Knowledge Landmark Position Visual Angle Orientation Region B in center er 38 2 2 or 5 ....

[Article contains additional citation context not shown here]

Tod S. Levitt and Daryl T. Lawton. Qualitative navigation for mobile robots. Artificial Intelligence, 44(3):305--360, August 1990.

....the robot can find its way to a target provided it is represented in the network. Whereas Prescott s approach is quantitative, ours uses a fuzzy extension of the beta coefficient coding system in order to work with fuzzy qualitative information about distances and directions. Levitt and Lawton [11] also proposed a qualitative approach to the navigation problem, but assume unrealistically accurate distance and direction information between the robot and the landmarks. Another qualitative method for robot navigation was proposed by Escrig and Toledo [8] using constraint logic. However, they ....

T.S. Levitt and D.T. Lawton. Qualitative navigation for mobile robots. Artificial Intelligence Journal, 44:305--360, 1990.

....the robot can find its way to a target provided it is represented in the network. Whereas Prescott s approach is quantitative, ours uses a fuzzy extension of the beta coefficient coding system in order to work with fuzzy qualitative information about distances and directions. Levitt and Lawton [11] also proposed a qualitative approach to the navigation problem, but assume unrealistically accurate distance and direction information between the robot and the landmarks. Another qualitative method for robot navigation was proposed by Escrig and Toledo [8] using constraint logic. However, they ....

T.S. Levitt and D.T. Lawton. Qualitative navigation for mobile robots. Artificial Intelligence Journal, 44:305--360, 1990.

....Often, the robot does not need a map to perform a simple displacement, especially if the environment it travels through is very simple or highly engineered. But, if the robot s task requires an understanding of the world, the robot has to answer the three questions posed by Levitt and Lawton [Levitt Lawton 90] Where am I . How do I get to other places from here . Where are other places relative to me In other words, it needs a map of its world. With the term map we do not refer only to the geographical maps humans are used to. There are a wide range of di#erent maps a robot can use. Two ....

Tod S. Levitt and Daryl T. Lawton. Qualitative navigation for mobile robots. Artificial Intelligence, 44(3):305--361, 1990.

....many authors assimilated 2 D objects to very elementary entities such as a point (barycenter) or a (bounding) rectangle. This process is extremely practical, therefore it has often been used, notably for spatial reasoning and representation and processing of qualitative spatial knowledge (see e.g. [3, 13, 14, 20]) However, a lot of morphological information on the considered objects is lost, and the procedure cannot be hoped to give a satisfactory modelling of the relationships. By introducing the notion of the histogram of angles, Miyajima and Ralescu [17] developed the idea that the relative position ....

T. S. Levitt, D. T. Lawton, "Qualitative Navigation for Mobile Robots", AI, vol. 44, no. 3, pp. 305-360, 1990.

....not suitable, and Freeman proposed that fuzzy relations be used. However, computers have not been able to effectively model these vital spatial concepts. For instance, many authors assimilated 2D objects to very elemen tary en tities such as a point (centroid) or a (bounding) rectangle [11] 12] [13] [14] The procedure is practical, but cannot be hoped to give a satisfactory modeling. 3 By in troducing the notion of the histogram of angles, Miyajima and Ralescu [15] developed the idea that the relative position between two objects can have a representation of its own and can thus be ....

T. S. Levitt and D. T. Lawton, "Qualitative Navigation for Mobile Robots", Artificial Intelligence, vol. 44, no. 3, pp. 305-360, 1990.

....sensor devices. In contrast, over the last few years, qualitative approaches to robotic navigation have been proposed to overcome problems regarding complexity and stability. Qualitative modelling for selflocalization and navigation include the qualitative map approach [Kuipers and Byun, 1991; Levitt and Layto, 1990] and the adaptive topological models introduced by [Prescott, 1994] Kurz s ultrasonic clustering techniques [Kurz, 1993] and the approach based on typical sequences of local sensorreadings rather than on explicit topology [Tani and Fukumura, 1994] as well as robust world modelling technique by ....

T S Levitt and D T Layto. Qualitative navigation for mobile robots. Artificial Intelligence, 44:305--360, 1990.

....possible application of such algorithms, consider the problem of a robot attempting to match an image of its current location to an image of a known landmark taken at some time in the past. For the sake of efficiency, one can use one dimensional visual data (of a 360 ffi view) to do the matching [48, 60, 70, 77]. Standard patternmatching approaches might encounter difficulties if the robot is translated and rotated slightly from the location and orientation where the first image was taken. To apply our learning algorithms, we pre process the images, placing points where there are significant changes in ....

....matching algorithm be performed in real time. To reduce the processing time required by the landmark matching algorithm, some have proposed the use of imaging systems that generate a one dimensional array of light intensities taken at eye level (see e.g. Hong et al. 48] Levitt and Lawton [60], Pinette [70] and Suzuki and Arimoto [77] We now briefly describe one such imaging system [48, 70] In their robot a spherical mirror is mounted above an upward pointing camera that enables it to instantaneously obtain a 360 ffi view of the world. See Figure 2.1 for a picture of such a ....

T. Levitt and D. Lawton. Qualitative navigation for mobile robots. Artificial Intelligence, 44(3):305--360, 1990.

....representation for location Approaches to robot navigation based on visually perceived landmarks have represented ordering information, but this has generally been done without recognizing it as such and has thus prevented its efficient use. The QUALNAV system described by Levitt and Lawton [6] distinguishes two non degenerate classes of robot positions with respect to two landmarks according to the angle under which the landmarks are perceived by the robots visual sensor: positions with angles smaller than for the robot approaching the landmarks and positons with angles bigger than for ....

Levitt, T. S., & Lawton, D. T. (1990) Qualitative navigation for mobile robots, Artificial Intelligence 44, 305-360

....error. In contrast, qualitative methods have been expected as methods which are not seriously affected by senor noise and enable us to navigate robots in a wide environment, for example. Levitt and Lawton reported a qualitative method for landmark based robot navigation in an outdoor environment [ Levitt and Lawton, 1990 ] The robot uses a map in which precise locations of landmarks are indicated. With the map, the robot can qualitatively know the location. Besides this method, several works have been reported which utilize pre defined qualitative maps and qualitatively utilize standard geometrical maps. On the ....

....3. This representation is one of the most fundamental representation and various methods have been proposed in the field of qualitative spatial reasoning [ Forbus et al. 1991; Freksa, 1992; Latecki and Rohrig, 1993 ] Especially, this representation can be used for map based robot navigation [ Levitt and Lawton, 1990 ] The qualitative positions as shown in the left of Figure 3 are formally represented based on relations among arbitrary three points as shown in the right of Figure 3. In general, the position of point p i is represented as follows [ Schlieder, 1995 ] p i p j p k = if p i p j p k lies in ....

Tod S. Levitt and Daryl T. Lawton. Qualitativenavigation for mobile robots. Int. J. Artificial Intelligence, 44:305--360, 1990.

....(b) On the chart, he relates his position to the destination, reference points and possible hazards; c) Based on this information, he sets the new course of the vessel. The nautical practice of navigation has entered almost unchanged into the domain of robotics. For instance, Levitt Lawton [37] define navigation as a process answering the following three questions: a) Where am I ; b) Where are other places with respect to me ; c) How do I get to other places from here . Applied to robot navigation, this means that the robot s sensory inputs are used to update a single global ....

....goal location. The minimal capabilities for navigation are thus to move in space, and to determine whether or not the goal has been found. To that end, the sensory features which identify the goal have to be stored in some form of longterm memory. In contrast to the definition of Levitt Lawton [37], this notion of navigation does not imply that the current location must be recognized, nor that a map like representation must be used to find the goal. The reference to a goal location distinguishes navigation from other forms of spatial behaviour such as exploration, foraging, obstacle ....

T. S. Levitt and D. T. Lawton. Qualitative navigation for mobile robots. Art. Intell., 44:305 -- 360, 1990.

....2000 27 The abduction consists of binding constants to the existentially quantified variables in the above axioms. It is implemented by rules that search for known places and paths with the required properties, creating new constants if existing ones cannot be found. Levitt and Lawton s QUALNAV [64, 58] provides an interesting comparison with the SSH. The traveller is assumed to have a panoramic visual sense for distant landmarks. It crosses a qualitative spatial boundary when it is colinear with two distant landmarks. The regions defined by these boundaries are the qualitative neighborhoods. ....

....as reactive responses to sensory input, prior to constructing a model of the environment. Other trajectory following control laws suitable for outdoor navigation include visual lane following for autonomous vehicles [36] crossing qualitative boundaries defined by pairs of distant landmarks [64], and following a fixed heading until a destination is reached [29, 99] Spatial Semantic Hierarchy DRAFT: February 18, 2000 42 Note that while control laws use continuous sensory input and issue continuous output signals, they embody very little knowledge about what the sensors are sensing or ....

[Article contains additional citation context not shown here]

T. S. Levitt and D. T. Lawton. Qualitative navigation for mobile robots. Artificial Intelligence, 44(3):305--360, 1990.

....Germany. Tel. 49 731 505 2120, fax: 49 731 505 4105. E mail address: matthias.franz daimlerchrysler.com (M.O. Franz) 1 From Latin: navis, ship; agere, to drive. The nautical practice of navigation has entered almost unchanged into the domain of robotics. For instance, Levitt and Lawton [37] define navigation as a process answering the following three questions: a) Where am I ; b) Where are other places with respect to me ; c) How do I get to other places from here . Applied to robot navigation, this means that the robot s sensory inputs are used to update a single global ....

....location. The minimal capabilities for navigation are thus to move in space, and to determine whether or not the goal has been found. To that end, the sensory features which identify the goal have to be stored in some form of long term memory. In contrast to the definition of Levitt and Lawton [37], this notion of navigation does not imply that the current location must be recognized, nor that a map like representation must be used to find the goal. The reference to a goal location distinguishes navigation from other forms of spatial behaviour such as exploration, foraging, obstacle ....

T.S. Levitt, D.T. Lawton, Qualitative navigation for mobile robots, Artificial Intelligence 44 (1990) 305--360.

....There is little work combining physical uncertainty with path planning. Latombe et al. [8] 18] identifies regions in the environment in terms of the likelihood that the robot can attain them and localise itself given inaccuracies in sensors and robot control. There have been some approaches [9] [17] in which the problem is ignored by describing the robot s position only in general terms relative to objects in the environment. Planning is then a simple task of moving from one region to another. When accurate localisation of the robot is necessary, however, these appraches are not ....

T. S. Levitt and D. T. Lawton. Qualitative navigation for mobile robots. Artificial Intelligence, 44:305--360, August 1990.

....robot can find its way to a target provided it is represented in the network. Prescott s approach is quantitative whereas our approach uses a fuzzy extension of the beta coefficient coding system in order to work with fuzzy qualitative information about distances and directions. Levitt and Lawton [10] also proposed a qualitative approach to the navigation problem but assumes an unrealistically accurate distance and direction information between the robot and the landmarks. Another qualitative method for robot navigation was proposed by Escrig and Toledo [5] using constraint logic. However, ....

T.S. Levitt and D.T. Lawton, `Qualitative navigation for mobile robots', Artificial Intelligence Journal, 44, 305--360, (1990).

....Pallottino, 1986; Pearl, 1988] However, limited research has been done in adapting these algorithms for route finding purposes. In AI, a number of studies have also been made. Their emphasis has been on using topographic and geographic knowledge and reasoning to solve the problem [Kuipers, 1978; Levitt, and Lawton, 1990; McDermott and Davis, 1984; Goel et al. 1991] The major concerns in AI are spatial knowledge representation and reasoning, rather than efficiency and optimization. Shapiro et al. 1992] which is the latest work we could find in network theory and its applications, propose a technique to use ....

T Levitt, and D. Lawton. Qualitative navigation for mobile robots. Artificial Intelligence, 44(3):305-360, 1990.

....L, it should check that it is really near L before heading to the next landmark. Then adjustments can be made if the robot is not at L by re homing to L and or updating its map. For efficiency s sake, one can use one dimensional visual data (of a 360 # view) to do the matching (Hong et al. 1992; Levitt Lawton, 1990; Pinette, 1993; Suzuki Arimoto, 1988) To apply our learning algorithms, we pre process the images, placing points on the real line where there are significant changes in light intensity, producing one dimensional geometric patterns. Then we take patterns produced from images taken near the ....

Levitt, T., & Lawton, D. (1990). Qualitative navigation for mobile robots. Artificial Intelligence, 44, 305--360.

....intelligence, neuroscience and robotics provide different tentative answers. Taken together as a whole, they are diverse and often conflicting. It is thus necessary to first define what navigation is and what the functions of a navigation system are. For instance, according to Levitt and Lawton [62], navigation is defined by the following three questions: a) Where am I , b) Where are other places relative to me , and (c) How do I get to other places from here Underlying the question (a) is the problem of recognizing and identifying the particular place the animal or the animat is ....

T. S. Levitt and D. T. Lawton. Qualitative navigation for mobile robots. Artificial Intelligence, 44:305--360, 1990.

....without any concept of a distance measure, and did so long before the introduction of the magnetic compass in 1100 A.D. The visual angle from a viewpoint to two point features will be defined as the angle formed by the rays from the viewpoint location to each point feature. It has been shown [Levitt and Lawton, 1990] that a perfect estimate of the visual angle between two points constrains the viewpoint to a surface of revolution somewhat resembling the torus shown in Figure 3.3a. If a two dimensional approximation of the environment is assumed, as in Figure 3.3b, the viewpoint is constrained to the ....

....such thickened rings. When a two dimensional approximation of the environment is assumed, an error in the visual angle estimate to two features, as shown in Figure 3. 5a, will constrain the viewpoint to a thickened ring, the thickness of the ring determined by the amount of error [Krotkov, 1989, Levitt and Lawton, 1990] . When three features are used, any given error in estimate constrains the viewpoint to the intersection of two such rings [Kuipers and Levitt, 1988, Sutherland, 1992] 1 See Appendix A for 1 A third ring passing through the two features lying at greatest distance from each other can be ....

[Article contains additional citation context not shown here]

Tod S. Levitt and Daryl T. Lawton. Qualitative navigation for mobile robots. Artificial Intelligence, 44(3):305--360, August 1990.

....(19) The abduction consists of binding constants to the existentially quantified variables in the above axioms. It is implemented by rules that search for known places and paths with the required properties, creating new constants if existing ones cannot be found. Levitt and Lawton s QUALNAV [47, 41] provides an interesting comparison with the SSH. The traveller is assumed to have a panoramic visual sense for distant landmarks. It crosses a qualitative spatial boundary when it is colinear with two distant landmarks. The regions defined by these boundaries are the qualitative neighborhoods. ....

T. S. Levitt and D. T. Lawton. Qualitative navigation for mobile robots. Artificial Intelligence, 44(3):305--360, 1990.

....for continuous sensorimotor interaction with the world and performs perceptual processing of real sensory input to obtain views. The TOUR model thus corresponds to the SSH causal, topological and metrical levels. 2. 3 Levitt et al. Qualnav) The Qualnav model developed by Levitt et al. 1987; Levitt and Lawton, 1990; Kuipers and Levitt, 1988] is another implemented model of the SSH approach besides the TOUR model, the NX model and Spot model. Qualnav uses the multi level spatial semantic representation that supports topological representation of relative spatial locations, and smooth integration of available ....

Tod S. Levitt & Daryl T. Lawton, "Qualitative navigation for mobile robots," in Artificial Intelligence, 44, 1990, pp. 303--360.

....X (b) visibility boundary surface for edge in scale space (a) instances of visibility boundary curves in viewpoint space v2 = 1, 1) v1 = 1, 1) s Figure 9: Visibility ranges of an edge for varying visual arc angles. and oe is the visual arc angle. This curve is merely a circular arc segment [19], the size and center of which vary with oe for a particular edge as shown in Figure 9.a. The shape of the surface in scale space can be seen in Figure 9.b. The nature of this surface is such that in the ideal case of infinite camera resolution (oe = 0 ffi ) the curve in the plane degenerates to ....

....remain constant, the viewpoint dependent nature of limbs on curved surface objects will likely render their solution computationally intractable. For polyhedra, the visibility surface for an edge feature is the toroidal shape generated by rotating the circular curves about the edge in question [19]. Similar to the pseudo edges in 2 D, the visibility of chains of edges around the boundary of a face must be examined, as the nature of the face s shape changes. The visibility of a face as a whole is governed by the size of its projected area. If any two points along the boundary are closer ....

Levitt, T. S. and Lawton, D. T. "Qualitative navigation for mobile robots", Artificial Intelligence, vol. 44, pp. 305-360, 1990.

.... (such as a compass) or because the navigating agent has no global map (so that even the availability of global coordinates is not useful) Navigation in a large unstructured environment requires different information and tools than that of navigation in a small structured environment [SVW85] KB88][LT90][ST93] The information that is useful is mostly qualitative rather than quantitative. High level reasoning is necessary for the navigator to be able to handle ambiguities and errors; by high level we mean symbolic direction specifying invariants that can be used to capture the appropriate ....

....environments [KB88] Metric information becomes inaccurate, due to the low mechanical accuracy and sensory errors [ST93] errors accumulate. Qualitative approaches to robot navigation include: the TOUR model of Kuipers[Kui78] the NX Robot by Kuipers et al. KB88] Qualnav by Levitt et al. [LT90], inexact navigation by Sutherland et al. ST93] Dai et al. s range free navigation [DL93] and the PV (Panoramic View) representation used by Zheng et al. [ZT92] Most of the work in qualitative navigation emphasizes the importance of landmarks. However, none provides a formal answer to the ....

Todd Levitt and Lawton Daryl T. Qualitative navigation for mobile robots. Artificial Intelligence, 44(3):305--360, 1990.

....suffer from cumulative errors introduced by encoder imprecision and by wheels and gears slippage. Other approaches have focused on developing robust perceptual based inference techniques, and have used perceptual based schemes that emphasize the integration of distinct sensor modality data [1, 6, 9, 8]. Perception based approaches are faced with a fundamental tradeoff between using the perceptual clues that occur naturally in the environment, or instrumenting the environment with artificial clues. While physical alteration of the environment is not always feasible or desirable, the use of ....

T. S. Levitt and D. T. Lawton. Qualitative navigation for mobile robots. Artificial Intelligence, 44(3), 1990, 305--361.

....and difficult problem how to utilize the map and match it with the enormous amount of observed geometric information for robot self localization and path planning. Alternatively, qualitative maps, such as topological graphs, are proposed to model robot environments which are characterized by a small set of distinct locations and the routes between them [DJMW1,DM,DP,KB,LL,RS] Kuipers and Levitt have proposed a hierarchical spatial representation consisting of four levels: sensorimotor level (a robot uses sensors to detect local features of the environment) procedural level where the robot ....

....nor purely qualitative, leave certain features of the environment out but keep necessary information helpful for robot motion planning. The robot s perception of the world can vary according to the sensors it carries. In many situations, it is assumed that nodes or edges traversed previously can all be distinguished. In contrast, it is assumed in [RS] that nodes are divided into a small number of classes, for example, white and black colors, and can only be recognized as such. Dudek et al. DJMW1] apply the world model introduced by Kuipers and Levitt to a specific situation in which no ....

[Article contains additional citation context not shown here]

T.S. Levitt, and D. T. Lawton, "Qualitative Navigation for Mobile Robots," Artificial Intelligence 44 (1990), 305-360.

....that the landmark matching algorithm can be performed in realtime. To reduce the processing time required by the landmark matching algorithm, some are proposing the use of imaging systems that generate a one dimensional array of light intensities taken at eye level (see e.g. Hong et al. 1992) Levitt and Lawton (1990), Pinette (1993) Suzuki and Arimoto (1988) We now briefly de PAC LEARNING OF ONE DIMENSIONAL PATTERNS 3 Not available electronically. Figure 1. The imaging system on the robot. This figure comes directly from Pinnette s (1993) thesis. scribe one such imaging system (Hong et al. 1992) and ....

Levitt, T.S. & Lawton, D.T. (1990). Qualitative navigation for mobile robots. Artificial Intelligence, 44(3): pp. 305-360.

No context found.

Levitt, T. S. and Lawton, D. T. (1990). Qualitative navigation for mobile robots. Artificial Intelligence, 44(3):305--360.

No context found.

Tod S. Levitt, Daryl T. Lawton. Qualitative Navigation for Mobile Robots. Artif. Intell. 44(3): 305-360 (1990)

No context found.

T. S. Levitt and D. T. Lawton. Qualitative Navigation for Mobile Robots. Artificial Intelligence, 44:305--360, 1990.

No context found.

T. S. Levitt and D. T. Lawton. Qualitative Navigation for Mobile Robots. Artificial Intelligence, 44:305--360, 1990.

No context found.

T. Levitt and D. Lawton. Qualitative navigation for mobile robots. Artificial Intelligence, 44(3):305, 361 1990.

No context found.

T. Levitt and D. Lawton. Qualitative navigation for mobile robots. Artificial Intelligence, 44(3):305, 361 1990.

No context found.

T. S. Levitt and D. T. Lawton. Qualitative Navigation for Mobile Robots. Artificial Intelligence, 44:305--360, 1990.

No context found.

T. S. Levitt and D. T. Lawton. Qualitativenavigation for mobile robots. Artificial Intelligence, 44(3):305--360, August 1990.

No context found.

T. S. Levitt and D. T. Lawton. Qualitative Navigation for Mobile Robots. Artificial Intelligence Journal, 44:305 -- 360, 1990.

No context found.

Levitt, T., & Lawton, D. (1990). Qualitative Navigation for Mobile Robots. Artificial Intelligence,

No context found.

T. Levitt and D. Lawton. Qualitative navigation for mobile robots. Artificial Intelligence, 44(3):305--360, 1990.

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