1 / 24

A Robot Exploration and Mapping Strategy Based on a Semantic Hierarchy of Spatial Representations Benjamine Kuipers and

A Robot Exploration and Mapping Strategy Based on a Semantic Hierarchy of Spatial Representations Benjamine Kuipers and Yung-Tai Byun. Tai S. Jang MAE University of Missouri-Columbia ECE 7340 : Building Intelligent Robots Winter, 2005. PRESENTATION OUTLINE. INTRODUCTION BACKGROUND

beata
Download Presentation

A Robot Exploration and Mapping Strategy Based on a Semantic Hierarchy of Spatial Representations Benjamine Kuipers and

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. A Robot Exploration and Mapping Strategy Based on a Semantic Hierarchy of Spatial RepresentationsBenjamine Kuipers and Yung-Tai Byun Tai S. Jang MAE University of Missouri-Columbia ECE 7340 : Building Intelligent Robots Winter, 2005

  2. PRESENTATION OUTLINE • INTRODUCTION • BACKGROUND • THE HIERARCHICAL MAP • A ROBOT INSTANCE: NX • EXPLORATION RESULTS • SUMMARY

  3. INTRODUCTION • A robust qualitative method is presented for robot exploration, mapping, and navigation in large-scale spatial environments.

  4. INTRODUCTION • Experiments with a simulated robot in a variety of 2-D environments have demonstrated that the proposed method can build an accurate map of an unknown environment in spite of substantial random and systematic sensorimotor error • In this qualitative method, location-specific control algorithms are dynamically selected to control the robot’s interaction with its environment (large - scale). These algorithms define distinctive places and paths, which are linked to form a topological network description. Finally the geometry is obtained from it.

  5. INTRODUCTION • The Control Level: Distinctive places and path are defined in terms of the control strategies and sensory measures (called d-measures) which support convergence to them from anywhere within a local neighborhood. • A distinctive place is defined as the local maximum found by a hill – climbing control strategy. • A distinctive path is defined by the d-measure and control strategy (e.g. follow – the – midline) • The Topological Level: A topological network description of the global environment is created before the global geometric map, by identifying and linking distinctive places and distinctive paths in the environment. • The Geometric Level: The geometric map is created by accumulating, first, local geometric information about places and paths, then global metrical relations among these elements.

  6. INTRODUCTION

  7. INTRODUCTION Definition: Topology • We do not ask: how big is it? but rather: does it have any holes in it? is it all connected together, or can it be separated into parts? A commonly cited example is the London Underground map. • This will not reliably tell you how far it is from one station to other, or even the compass direction from one to the other; but it will tell you how the lines connect up between them.

  8. PRESENTATION OUTLINE • INTRODUCTION • BACKGROUND • THE HIERARCHICAL MAP • A ROBOT INSTANCE: NX • EXPLORATION RESULTS • SUMMARY

  9. BACKGROUND • Humans perform well at spatial learning and spatial problems solving in spite of sensory and processing limitations and frequently-incomplete knowledge. • It is a qualitative approach rather than a metrical or quantitative one. • Robot exploration and map-learning problem is a enormous challenge because low mechanical accuracy and sensory errors make difficult to obtain purely metrically accurate map in large-scale space. • The layered structure of the cognitive map or topological description is responsible for humans’ robust performance in large-scale space. • So here the authors attempt to apply qualitative method to the problem of robot exploration and map-learning.

  10. PRESENTATION OUTLINE • INTRODUCTION • BACKGROUND • THE HIERARCHICAL MAP • A ROBOT INSTANCE: NX • EXPLORATION RESULTS • SUMMARY

  11. PRESENTATION OUTLINE • INTRODUCTION • BACKGROUND • THE HIERARCHICAL MAP • A ROBOT INSTANCE: NX • EXPLORATION RESULTS • SUMMARY

  12. THE HIERARCHICAL MAP • The central element of the authors’ hierarchical model is the topological network description, in which nodes correspond to distinctive places and arcs correspond to travel paths. • Here how to define distinctive places and travel paths, and their descriptions at the control and metrical levels will be discussed. • Distinctive places • Distinctive Travel Paths • The Basic Exploration Strategy • Robust against Errors • The Position Referencing Problem • The Exploration Agenda • The Rehearsal Procedure

  13. THE HIERARCHICAL MAP • Distinctive places • A Place corresponding to a node must be locally distinctive within its immediate neighborhood by some criterion definable in terms of sensory input. • The following distinctiveness measures (d-measures) are defined for example • Extent of distant differences to near objects • Extent and quality of symmetry across the center of the robot or a line • Temporal discontinuity in one or more sensors, experienced over a small step • Number of directions of reasonable motion into open spaces around the robot. • Temporal change in number of directions of motion provided by the distinct open spaces, experienced over a small step. • The point along a path that minimizes or maximizes lateral distance readings. • Once the robot recognizes that it is in the neighborhood of a distinctive place, it applies a hill-climbing control strategy to move to the point where some distinctiveness measure ahs its local maximum value.

  14. THE HIERARCHICAL MAP • 2. Distinctive Travel Paths • The paths followd during exploration are defined by some distinctiveness criterion that is sufficient to specify a one-dimentsional set of points • The travel paths connecting two distinctive places are defined in terms of local control strategies (LCS). • For example, • Follow-the-midline • Move-along-Object-on Right • Move-along-Object-on-Left • Blind Step • The knowledge used for selecting and performing the proper LCS is dependent on the robot’s sensorimotor system. • Besides connectivity information, locally-observable metrical information is accumulated to describe the geometric features of an path, such as length, lateral width, curvature, net change in orientation, etc.

  15. THE HIERARCHICAL MAP

  16. THE HIERARCHICAL MAP 3. The Basic Exploration Strategy • From a place, move into an open direction • Select a control strategy and follow a path • Detect a neighborhood, select a d-measure, and begin hill-climbing • Reach a local maximum that defines being at another distinctive place. • The topological model is built as a side-effect of motion through this transition graph.

  17. THE HIERARCHICAL MAP 4. Robust against Errors Although the robot is subject to sensory and movement errors, hill-climbing search based on continuous sensory feedback will bring it very near to a distinctive place. The Hill-Climbing search always moves towards the goal. Using heuristics it finds which direction will take it closest to the goal. The name "hill-climbing" comes from an analogy: A hiker is lost halfway up/down (depends on if you are an optimist) a mountain at night. His camp is at the top of the mountain. Even though it is dark, the hiker knows that every step he takes up the mountain is a step towards his goal. So a hill-climbing search always goes to the node closest to the goal.

  18. THE HIERARCHICAL MAP 5. The Position Referencing Problem The current position is described at two levels: topological and metrical. At the topological level, the current position is described by either a distinctive place, or by a pair representing a path and a direction. At the metrical level, then the robot its at a distinctive place, the current local sensory information and its current orientation are given. When it is on a path, the robot’s current position may be describe in terms of the place it is coming from, the distance it has traveled, lateral distance information, and its current orientation.

  19. THE HIERARCHICAL MAP 6. The Rehearsal Procedure When a robot reaches a place during the exploration, the identification of the palce is the most important task. If the place has been visited before and the robot comes back to that place, the robot should recognize it. A new place must be recognized as new. Place matching is done using global topological constraints as well as local metrical comparison

  20. PRESENTATION OUTLINE • INTRODUCTION • BACKGROUND • THE HIERARCHICAL MAP • A ROBOT INSTANCE: NX • EXPLORATION RESULTS • SUMMARY

  21. THE Robot Instance: NX The NX Robot Simulator The robot NX with 16 sensors covering 360 degrees with equal angle differences and an absolute compass for global orientation has used to test this qualitative method. EXPLORATION RESULTS To demonstrate the effect of sensory errors, the authors showed the exploration result for three different random error rates: the error-free, five percent error, and then percent error It showed that the method used is quite robust to sensory errors.

  22. SUMMARY • Its is very difficult to build a metrically accurate map within a global coordinate frame, through exploration in an unknown unstructured environment. • The authors used a hierarchical description of the spatial environment, in which a topological network description mediates between a control and a metrical level. • Distinctive places and paths are defined by their properties at the control level, and serve as the nodes and arcs of the topological model. • Each place and path can then accumulate local metrical information • Cumulative location error minimized: due to alternating between the path-following and hill-climbing control algorithms. • Successful navigation : It doesn’t depend on geometric accuracy. • Geometric sensor fusion methods: can be naturally incorporated as methods for acquiring local geometric descriptions of places and paths • Indistinguishable places: i.e. places with identical local sensory characteristics can be identified correctly, except in the most pathological environments.

  23. SUMMARY Limitations • The robot is currently considered low-speed: A more realistic robot would be high speed requiring a more sophisticated set of control strategies to move through the environment. • It requires a improved obstacle-avoidance capabilities to deal with moving agents • The environment itself may change, as doors are opened or closed, or parked cars move around requiring diagnosis to discriminate between fixed and changeable aspect of the environment.

  24. Thanks...

More Related