Integration of Representation Into Goal-Driven Behavior-Based Robots By Maja J. Mataric Presented by Murali Kiran. M
About the Author Associate Professor Computer Science Department and Neuroscience Program Director, Center for Robotics and Embedded Systems (CRES) Co-Director, Robotics Research Lab President-Elect, Academic Senate Chair, VSoE Women in Science and Engineering (WiSE) Viterbi School of Engineering (VSoE), University of Southern California Associate Professor Computer Science Department and Neuroscience Program Director, Center for Robotics and Embedded Systems (CRES) Co-Director, Robotics Research Lab President-Elect, Academic Senate Chair, VSoE Women in Science and Engineering (WiSE) Viterbi School of Engineering (VSoE), University of Southern California Computer Science DepartmentNeuroscience ProgramCenter for Robotics and Embedded Systems (CRES)Robotics Research LabAcademic SenateVSoEWomen in Science and Engineering (WiSE) Viterbi School of Engineering (VSoE)University of Southern California Computer Science DepartmentNeuroscience ProgramCenter for Robotics and Embedded Systems (CRES)Robotics Research LabAcademic SenateVSoEWomen in Science and Engineering (WiSE) Viterbi School of Engineering (VSoE)University of Southern California Homepage: Homepage:
Contact Address Computer Science Department University of Southern California Office: Ronald Tutor Hall (RTH) 407 Mailing address: Henry Salvatori, Mailcode West 37th Place Los Angeles, CA USA Computer Science Department University of Southern California Office: Ronald Tutor Hall (RTH) 407 Mailing address: Henry Salvatori, Mailcode West 37th Place Los Angeles, CA USA Computer Science Department University of Southern California Computer Science Department University of Southern California
Task to perform Explore an office environment Explore an office environment Construct and maintain a map based on the landmarks it discovers. Construct and maintain a map based on the landmarks it discovers.
TOTO Omni directional three wheeled base. Omni directional three wheeled base. Twelve ultrasonic ranging sensors Twelve ultrasonic ranging sensors Flux gate compass Flux gate compass
TOTO
Competencies Basic navigation Basic navigation Landmark detection Landmark detection Map-related computation Map-related computation
Robot Behavior Stroll Stroll Avoid Avoid Align Align Correct Correct
Stroll
Avoid
Align
Correct
Schematic Diagram
Landmark Detection
Acquiring world model Acquiring world model Sensors Sensors Perceptual limitations Perceptual limitations Sensor noise Sensor noise Drift or slippage Drift or slippage Complexity and dynamics Complexity and dynamics
Fundamental paradigm Grid-based (metric) paradigm Grid-based (metric) paradigm Topological paradigm Topological paradigm
Grid-based paradigm Represents environment by evenly spaced grids. Represents environment by evenly spaced grids. Grid cell may represent an obstacle in the corresponding region of the environent. Grid cell may represent an obstacle in the corresponding region of the environent.
Topological paradigm Represents robots environments by graphs Represents robots environments by graphs Nodes in such graphs represent distinct places or landmarks. Nodes in such graphs represent distinct places or landmarks. They are connected by arcs if they have a direct path between them. They are connected by arcs if they have a direct path between them. Topological maps are build over Grid based maps Topological maps are build over Grid based maps We are concerned about the topological paradigm We are concerned about the topological paradigm
Differences
Constructing topological maps Thresholding Thresholding Voronoi Diagram Voronoi Diagram Critical points Critical points Critical lines Critical lines Topological graph Topological graph
Thresholding Cells whose occupancy value is below the threshold are considered as free space. Cells whose occupancy value is below the threshold are considered as free space. Free space denoted by C Free space denoted by C All other points are considered as occupied. All other points are considered as occupied. _ They are denoted by C. They are denoted by C.
Voronoi Diagram (x,y) is a point in C. (x,y) is a point in C. Nearest points to (x,y) in the occupied space are called basic points. Nearest points to (x,y) in the occupied space are called basic points. Clearence is the distance between the basic points and (x,y). Clearence is the distance between the basic points and (x,y). Voronoi diagram is the set of all points in the free space that have atleast two different basic points. Voronoi diagram is the set of all points in the free space that have atleast two different basic points.
Critical points Critical points (x,y) are points on the Voronoi diagram that minimize clearance locally. Each critical point (x,y) has the following two properties: (a) it is part of the Voronoi diagram. (b) the clearance of all points in an "neighborhood of (x,y) is not smaller.
Critical Lines Critical lines are obtained by connecting each critical point with its basis points.
Topological graph The partitioning is mapped into an isomorphic map. The partitioning is mapped into an isomorphic map.
The Tuple The Tuple
Graph Representation
Mapping Algorithm
Conclusion