CS-417 Introduction to Robotics and Intelligent Systemsyiannis/417/2013/LectureSlides/16... ·...
Transcript of CS-417 Introduction to Robotics and Intelligent Systemsyiannis/417/2013/LectureSlides/16... ·...
![Page 1: CS-417 Introduction to Robotics and Intelligent Systemsyiannis/417/2013/LectureSlides/16... · 2013-12-03 · Mapping •What the world looks like •Improve the accuracy of the map](https://reader033.fdocuments.in/reader033/viewer/2022042804/5f5a436b577b057c37500572/html5/thumbnails/1.jpg)
Ioannis Rekleitis
Exploration
![Page 2: CS-417 Introduction to Robotics and Intelligent Systemsyiannis/417/2013/LectureSlides/16... · 2013-12-03 · Mapping •What the world looks like •Improve the accuracy of the map](https://reader033.fdocuments.in/reader033/viewer/2022042804/5f5a436b577b057c37500572/html5/thumbnails/2.jpg)
Three Main Challenges in Robotics
1. Where am I? (Localization)– Sense– relate sensor readings to a world model– compute location relative to model– assumes a perfect world model
2. What the world looks like? (Mapping)– sense from various positions– integrate measurements to produce map– assumes perfect knowledge of position
• Together 1 and 2 form the problem of Simultaneous Localization and Mapping (SLAM)
3. How do I go from A to B? (Path Planning)– More general: Which action should I pick next?
2CS-417 Introduction to Robotics and Intelligent Systems
![Page 3: CS-417 Introduction to Robotics and Intelligent Systemsyiannis/417/2013/LectureSlides/16... · 2013-12-03 · Mapping •What the world looks like •Improve the accuracy of the map](https://reader033.fdocuments.in/reader033/viewer/2022042804/5f5a436b577b057c37500572/html5/thumbnails/3.jpg)
Mapping
• What the world looks like
• Improve the accuracy of the map
• Ensure that all the important parts of the environment are mapped – Exploration!
3CS-417 Introduction to Robotics and Intelligent Systems
![Page 4: CS-417 Introduction to Robotics and Intelligent Systemsyiannis/417/2013/LectureSlides/16... · 2013-12-03 · Mapping •What the world looks like •Improve the accuracy of the map](https://reader033.fdocuments.in/reader033/viewer/2022042804/5f5a436b577b057c37500572/html5/thumbnails/4.jpg)
Environment Representation (Map)
• Grid Based Maps
• Feature Based Maps
• Topological Maps
• Hybrid Maps
4CS-417 Introduction to Robotics and Intelligent Systems
![Page 5: CS-417 Introduction to Robotics and Intelligent Systemsyiannis/417/2013/LectureSlides/16... · 2013-12-03 · Mapping •What the world looks like •Improve the accuracy of the map](https://reader033.fdocuments.in/reader033/viewer/2022042804/5f5a436b577b057c37500572/html5/thumbnails/5.jpg)
Consider this Environment:
5CS-417 Introduction to Robotics and Intelligent Systems
![Page 6: CS-417 Introduction to Robotics and Intelligent Systemsyiannis/417/2013/LectureSlides/16... · 2013-12-03 · Mapping •What the world looks like •Improve the accuracy of the map](https://reader033.fdocuments.in/reader033/viewer/2022042804/5f5a436b577b057c37500572/html5/thumbnails/6.jpg)
Three Basic Map Types
Topological:Collection of nodes and
their interconnections
Grid-Based:Collection of discretizedobstacle/free-space pixels
Feature-Based:Collection of landmark locations and correlated uncertainty
6CS-417 Introduction to Robotics and Intelligent Systems
![Page 7: CS-417 Introduction to Robotics and Intelligent Systemsyiannis/417/2013/LectureSlides/16... · 2013-12-03 · Mapping •What the world looks like •Improve the accuracy of the map](https://reader033.fdocuments.in/reader033/viewer/2022042804/5f5a436b577b057c37500572/html5/thumbnails/7.jpg)
Three Basic Map Types
Grid-Based Feature-Based Topological
Construction Occupancy grids Kalman Filter Navigation control laws
Complexity Grid size and resolution
Landmark covariance (N3)
Minimal complexity
Obstacles Discretizedobstacles
Only structured obstacles
GVG defined by the safest path
Localization Discrete localization
Arbitrary localization
Localize to nodes
Exploration Frontier-based exploration
No inherent exploration
Graph exploration
CS-417 Introduction to Robotics and Intelligent Systems 7
![Page 8: CS-417 Introduction to Robotics and Intelligent Systemsyiannis/417/2013/LectureSlides/16... · 2013-12-03 · Mapping •What the world looks like •Improve the accuracy of the map](https://reader033.fdocuments.in/reader033/viewer/2022042804/5f5a436b577b057c37500572/html5/thumbnails/8.jpg)
Grid Based Maps
• Occupied cells
• Free cells
• Unknown cells
8CS-417 Introduction to Robotics and Intelligent Systems
![Page 9: CS-417 Introduction to Robotics and Intelligent Systemsyiannis/417/2013/LectureSlides/16... · 2013-12-03 · Mapping •What the world looks like •Improve the accuracy of the map](https://reader033.fdocuments.in/reader033/viewer/2022042804/5f5a436b577b057c37500572/html5/thumbnails/9.jpg)
Frontier based Exploration (Grid Maps)
unknown
obstacle
emptyFrontier
CellsFrontier Targets
9CS-417 Introduction to Robotics and Intelligent Systems
![Page 10: CS-417 Introduction to Robotics and Intelligent Systemsyiannis/417/2013/LectureSlides/16... · 2013-12-03 · Mapping •What the world looks like •Improve the accuracy of the map](https://reader033.fdocuments.in/reader033/viewer/2022042804/5f5a436b577b057c37500572/html5/thumbnails/10.jpg)
Topological Representations
• B. J. Kuipers and Y.-T. Byun. “A robot exploration and mapping strategy based on a semantic hierarchy of spatial representations”. In Journal of Robotics and Autonomous Systems, 8: 47-63, 1991.
10CS-417 Introduction to Robotics and Intelligent Systems
![Page 11: CS-417 Introduction to Robotics and Intelligent Systemsyiannis/417/2013/LectureSlides/16... · 2013-12-03 · Mapping •What the world looks like •Improve the accuracy of the map](https://reader033.fdocuments.in/reader033/viewer/2022042804/5f5a436b577b057c37500572/html5/thumbnails/11.jpg)
Generalized Voronoi Graph (GVG)
H. Choset, J. Burdick, “Sensor based planning, part ii: Incremental construction of the generalized voronoi
graph”. In IEEE Conference on Robotics and Automation, pp. 1643 – 1648, 1995.
Free Space
11CS-417 Introduction to Robotics and Intelligent Systems
![Page 12: CS-417 Introduction to Robotics and Intelligent Systemsyiannis/417/2013/LectureSlides/16... · 2013-12-03 · Mapping •What the world looks like •Improve the accuracy of the map](https://reader033.fdocuments.in/reader033/viewer/2022042804/5f5a436b577b057c37500572/html5/thumbnails/12.jpg)
Generalized Voronoi Graph (GVG)
CS-417 Introduction to Robotics and Intelligent Systems 12
Free Space with Topological Map (GVG)
![Page 13: CS-417 Introduction to Robotics and Intelligent Systemsyiannis/417/2013/LectureSlides/16... · 2013-12-03 · Mapping •What the world looks like •Improve the accuracy of the map](https://reader033.fdocuments.in/reader033/viewer/2022042804/5f5a436b577b057c37500572/html5/thumbnails/13.jpg)
Generalized Voronoi Graph (GVG)
CS-417 Introduction to Robotics and Intelligent Systems 13
Free Space with Topological Map (GVG)
•Access GVG
![Page 14: CS-417 Introduction to Robotics and Intelligent Systemsyiannis/417/2013/LectureSlides/16... · 2013-12-03 · Mapping •What the world looks like •Improve the accuracy of the map](https://reader033.fdocuments.in/reader033/viewer/2022042804/5f5a436b577b057c37500572/html5/thumbnails/14.jpg)
Generalized Voronoi Graph (GVG)
CS-417 Introduction to Robotics and Intelligent Systems 14
Free Space with Topological Map (GVG)
•Access GVG•Follow Edge•Home to the MeetPoint•Select Edge
![Page 15: CS-417 Introduction to Robotics and Intelligent Systemsyiannis/417/2013/LectureSlides/16... · 2013-12-03 · Mapping •What the world looks like •Improve the accuracy of the map](https://reader033.fdocuments.in/reader033/viewer/2022042804/5f5a436b577b057c37500572/html5/thumbnails/15.jpg)
Exploration via Graph Search
• Exhaustive Depth First Search
• Breadth-First Search
• Heuristics
CS-417 Introduction to Robotics and Intelligent Systems 15
![Page 16: CS-417 Introduction to Robotics and Intelligent Systemsyiannis/417/2013/LectureSlides/16... · 2013-12-03 · Mapping •What the world looks like •Improve the accuracy of the map](https://reader033.fdocuments.in/reader033/viewer/2022042804/5f5a436b577b057c37500572/html5/thumbnails/16.jpg)
Irregular Triangular Mesh (ITM)
• Terrain Representation
• Underlying Topological Structure
• Path Planning and Exploration
CS-417 Introduction to Robotics and Intelligent Systems 16
![Page 17: CS-417 Introduction to Robotics and Intelligent Systemsyiannis/417/2013/LectureSlides/16... · 2013-12-03 · Mapping •What the world looks like •Improve the accuracy of the map](https://reader033.fdocuments.in/reader033/viewer/2022042804/5f5a436b577b057c37500572/html5/thumbnails/17.jpg)
• Convert ITM into Connected Graph
From 2.5D Representation to Topological
17CS-417 Introduction to Robotics and Intelligent Systems
![Page 18: CS-417 Introduction to Robotics and Intelligent Systemsyiannis/417/2013/LectureSlides/16... · 2013-12-03 · Mapping •What the world looks like •Improve the accuracy of the map](https://reader033.fdocuments.in/reader033/viewer/2022042804/5f5a436b577b057c37500572/html5/thumbnails/18.jpg)
Planning• Convert ITM into Connected Graph
• Planning using Graph Search Algorithms:
– Dijkstra, A* search algorithms
CS-417 Introduction to Robotics and Intelligent Systems 18
Start
Finish
![Page 19: CS-417 Introduction to Robotics and Intelligent Systemsyiannis/417/2013/LectureSlides/16... · 2013-12-03 · Mapping •What the world looks like •Improve the accuracy of the map](https://reader033.fdocuments.in/reader033/viewer/2022042804/5f5a436b577b057c37500572/html5/thumbnails/19.jpg)
Planning
• Convert ITM into Connected Graph
• Path Planning using Graph Search Algorithms:
– Dijkstra, A* search algorithms
• Different Cost Functions
– Number of triangles
CS-417 Introduction to Robotics and Intelligent Systems 19
1Q
Q
![Page 20: CS-417 Introduction to Robotics and Intelligent Systemsyiannis/417/2013/LectureSlides/16... · 2013-12-03 · Mapping •What the world looks like •Improve the accuracy of the map](https://reader033.fdocuments.in/reader033/viewer/2022042804/5f5a436b577b057c37500572/html5/thumbnails/20.jpg)
Planning
• Convert ITM into Connected Graph
• Path Planning using Graph Search Algorithms:
– Dijkstra, A*
• Different Cost Functions
– Number of triangles
– Euclidian distance
CS-417 Introduction to Robotics and Intelligent Systems 20
ji xxQ
Q
![Page 21: CS-417 Introduction to Robotics and Intelligent Systemsyiannis/417/2013/LectureSlides/16... · 2013-12-03 · Mapping •What the world looks like •Improve the accuracy of the map](https://reader033.fdocuments.in/reader033/viewer/2022042804/5f5a436b577b057c37500572/html5/thumbnails/21.jpg)
Planning• Convert ITM into Connected Graph
• Path Planning using Graph Search Algorithms:
– Dijkstra, A*
• Different Cost Functions
– Number of triangles
– Euclidian distance
– Slope of each triangle
CS-417 Introduction to Robotics and Intelligent Systems 21
21
21
jj
ij
jpp
ppv
jx
jv1
jp
2
jp
Q
![Page 22: CS-417 Introduction to Robotics and Intelligent Systemsyiannis/417/2013/LectureSlides/16... · 2013-12-03 · Mapping •What the world looks like •Improve the accuracy of the map](https://reader033.fdocuments.in/reader033/viewer/2022042804/5f5a436b577b057c37500572/html5/thumbnails/22.jpg)
Planning• Convert ITM into Connected Graph
• Path Planning using Graph Search Algorithms:
– Dijkstra, A*
• Different Cost Functions
– Number of triangles
– Euclidian distance
– Slope of each triangle
– Cross triangle slope
CS-417 Introduction to Robotics and Intelligent Systems 22
jx
jv
iv
ix
Q
![Page 23: CS-417 Introduction to Robotics and Intelligent Systemsyiannis/417/2013/LectureSlides/16... · 2013-12-03 · Mapping •What the world looks like •Improve the accuracy of the map](https://reader033.fdocuments.in/reader033/viewer/2022042804/5f5a436b577b057c37500572/html5/thumbnails/23.jpg)
Exploration Planning Problem
Two fundamental problems for path planning during exploration and mapping:
Planning for re-localization
23CS-417 Introduction to Robotics and Intelligent Systems
![Page 24: CS-417 Introduction to Robotics and Intelligent Systemsyiannis/417/2013/LectureSlides/16... · 2013-12-03 · Mapping •What the world looks like •Improve the accuracy of the map](https://reader033.fdocuments.in/reader033/viewer/2022042804/5f5a436b577b057c37500572/html5/thumbnails/24.jpg)
Exploration Planning Problem
– Planning for re-localization
– Planning the exploration of new territory
Two fundamental problems for path planning during exploration and mapping:
24CS-417 Introduction to Robotics and Intelligent Systems
![Page 25: CS-417 Introduction to Robotics and Intelligent Systemsyiannis/417/2013/LectureSlides/16... · 2013-12-03 · Mapping •What the world looks like •Improve the accuracy of the map](https://reader033.fdocuments.in/reader033/viewer/2022042804/5f5a436b577b057c37500572/html5/thumbnails/25.jpg)
Previous Localization Planning
• Reduce measure of map or position entropy
• Variety of graph search planning algorithms (breadth first, A*-search, RRT)
• Evaluate paths with simulation, or Cramer-Raobounds for expected uncertainty
• e.g. [Fox et al RAS 1998], [Sim and Roy ICRA 2005], [He et al ICRA 2008], [Censi et al ICRA 2008]
25CS-417 Introduction to Robotics and Intelligent Systems
![Page 26: CS-417 Introduction to Robotics and Intelligent Systemsyiannis/417/2013/LectureSlides/16... · 2013-12-03 · Mapping •What the world looks like •Improve the accuracy of the map](https://reader033.fdocuments.in/reader033/viewer/2022042804/5f5a436b577b057c37500572/html5/thumbnails/26.jpg)
Previous Exploration Planning
• Includes motion into unexplored regions
• Typically requires prior knowledge of environment properties or rough layout
• Computation of exploration effects is a challenge
• e.g. [Bourque and Dudek IROS 1999], [Bourgault et al IROS 2002], [Kollar and Roy IJRR 2008]
26CS-417 Introduction to Robotics and Intelligent Systems
![Page 27: CS-417 Introduction to Robotics and Intelligent Systemsyiannis/417/2013/LectureSlides/16... · 2013-12-03 · Mapping •What the world looks like •Improve the accuracy of the map](https://reader033.fdocuments.in/reader033/viewer/2022042804/5f5a436b577b057c37500572/html5/thumbnails/27.jpg)
Exploring a Camera Sensor Network
CS-417 Introduction to Robotics and Intelligent Systems 27
D. Meger, I. Rekleitis, and G. Dudek. “Heuristic Search Planning to Reduce Exploration Uncertainty”, IROS 2008.
![Page 28: CS-417 Introduction to Robotics and Intelligent Systemsyiannis/417/2013/LectureSlides/16... · 2013-12-03 · Mapping •What the world looks like •Improve the accuracy of the map](https://reader033.fdocuments.in/reader033/viewer/2022042804/5f5a436b577b057c37500572/html5/thumbnails/28.jpg)
Heuristic Search Planning Method
• Solution to exploration planning for camera sensor networks
– Composed of two alternated steps: exploration and re-localization
– Combined distance and uncertainty cost function
– Heuristic search for good paths
CS-417 Introduction to Robotics and Intelligent Systems 28
![Page 29: CS-417 Introduction to Robotics and Intelligent Systemsyiannis/417/2013/LectureSlides/16... · 2013-12-03 · Mapping •What the world looks like •Improve the accuracy of the map](https://reader033.fdocuments.in/reader033/viewer/2022042804/5f5a436b577b057c37500572/html5/thumbnails/29.jpg)
Exploration and Uncertainty Reduction
• Decision (exploration vs exploitation)
• Target Node
• Path Planning through the known graph
• Exploration Strategies
CS-417 Introduction to Robotics and Intelligent Systems 29
![Page 30: CS-417 Introduction to Robotics and Intelligent Systemsyiannis/417/2013/LectureSlides/16... · 2013-12-03 · Mapping •What the world looks like •Improve the accuracy of the map](https://reader033.fdocuments.in/reader033/viewer/2022042804/5f5a436b577b057c37500572/html5/thumbnails/30.jpg)
Exploration and Uncertainty Reduction
• Decision (exploration vs. exploitation)
– Epsilon-Greedy
– Epsilon-First
– Adaptive
– Bounded Uncertainty
• Target Node
• Path Planning through the known graph
• Exploration Strategies
CS-417 Introduction to Robotics and Intelligent Systems 30
![Page 31: CS-417 Introduction to Robotics and Intelligent Systemsyiannis/417/2013/LectureSlides/16... · 2013-12-03 · Mapping •What the world looks like •Improve the accuracy of the map](https://reader033.fdocuments.in/reader033/viewer/2022042804/5f5a436b577b057c37500572/html5/thumbnails/31.jpg)
Exploration and Uncertainty Reduction
• Decision (exploration vs. exploitation)
• Target Node (Exploration)
– Random
– Shortest distance
– Maximum Uncertainty
– Minimum Uncertainty
• Path Planning through the known graph
• Exploration Strategies
CS-417 Introduction to Robotics and Intelligent Systems 31
![Page 32: CS-417 Introduction to Robotics and Intelligent Systemsyiannis/417/2013/LectureSlides/16... · 2013-12-03 · Mapping •What the world looks like •Improve the accuracy of the map](https://reader033.fdocuments.in/reader033/viewer/2022042804/5f5a436b577b057c37500572/html5/thumbnails/32.jpg)
Exploration and Uncertainty Reduction
• Decision (exploration vs. exploitation)
• Target Node (Relocalization)
– Maximum Uncertainty
• Path Planning through the known graph
• Exploration Strategies
CS-417 Introduction to Robotics and Intelligent Systems 32
![Page 33: CS-417 Introduction to Robotics and Intelligent Systemsyiannis/417/2013/LectureSlides/16... · 2013-12-03 · Mapping •What the world looks like •Improve the accuracy of the map](https://reader033.fdocuments.in/reader033/viewer/2022042804/5f5a436b577b057c37500572/html5/thumbnails/33.jpg)
Exploration and Uncertainty Reduction
• Decision (exploration vs. exploitation)
• Target Node
• Path Planning through the known graph– Work with D. Meger and G. Dudek [IROS 2008]
– A* based strategy
– Cost:
– Distance-based “cost-to-go” heuristic function h used to compute estimated cost
• Exploration Strategies
CS-417 Introduction to Robotics and Intelligent Systems
))(()()( pPtraceplengthpC ud
Cost so far Estimated cost to goEstimated cost through n
33
![Page 34: CS-417 Introduction to Robotics and Intelligent Systemsyiannis/417/2013/LectureSlides/16... · 2013-12-03 · Mapping •What the world looks like •Improve the accuracy of the map](https://reader033.fdocuments.in/reader033/viewer/2022042804/5f5a436b577b057c37500572/html5/thumbnails/34.jpg)
Effect of α Parameter for Relocalization
CS-417 Introduction to Robotics and Intelligent Systems 34
![Page 35: CS-417 Introduction to Robotics and Intelligent Systemsyiannis/417/2013/LectureSlides/16... · 2013-12-03 · Mapping •What the world looks like •Improve the accuracy of the map](https://reader033.fdocuments.in/reader033/viewer/2022042804/5f5a436b577b057c37500572/html5/thumbnails/35.jpg)
• Graph search to optimize cost function
• Heuristic search allows considering only a fraction of the paths, ordered by expected cost
• Distance-based “cost-to-go” heuristic function hused to compute estimated cost
Heuristic Search
Cost so far Estimated cost to goEstimated cost through n
35CS-417 Introduction to Robotics and Intelligent Systems
))(()()( ptraceplengthpC ud
![Page 36: CS-417 Introduction to Robotics and Intelligent Systemsyiannis/417/2013/LectureSlides/16... · 2013-12-03 · Mapping •What the world looks like •Improve the accuracy of the map](https://reader033.fdocuments.in/reader033/viewer/2022042804/5f5a436b577b057c37500572/html5/thumbnails/36.jpg)
Exploration and Uncertainty Reduction
• Decision (exploration vs. exploitation)
• Target Node
• Path Planning through the known graph
• Exploration Strategies
– One Step Exploration
– Ear based exploration (submitted to IROS 2012)
CS-417 Introduction to Robotics and Intelligent Systems 36
![Page 37: CS-417 Introduction to Robotics and Intelligent Systemsyiannis/417/2013/LectureSlides/16... · 2013-12-03 · Mapping •What the world looks like •Improve the accuracy of the map](https://reader033.fdocuments.in/reader033/viewer/2022042804/5f5a436b577b057c37500572/html5/thumbnails/37.jpg)
Shortest NodeP(exploit)=0.3
CS-417 Introduction to Robotics and Intelligent Systems 37
![Page 38: CS-417 Introduction to Robotics and Intelligent Systemsyiannis/417/2013/LectureSlides/16... · 2013-12-03 · Mapping •What the world looks like •Improve the accuracy of the map](https://reader033.fdocuments.in/reader033/viewer/2022042804/5f5a436b577b057c37500572/html5/thumbnails/38.jpg)
Experimental ResultsBounded Uncertainty
CS-417 Introduction to Robotics and Intelligent Systems 38
![Page 39: CS-417 Introduction to Robotics and Intelligent Systemsyiannis/417/2013/LectureSlides/16... · 2013-12-03 · Mapping •What the world looks like •Improve the accuracy of the map](https://reader033.fdocuments.in/reader033/viewer/2022042804/5f5a436b577b057c37500572/html5/thumbnails/39.jpg)
Experimental ResultsDifferent Strategies
CS-417 Introduction to Robotics and Intelligent Systems 39
![Page 40: CS-417 Introduction to Robotics and Intelligent Systemsyiannis/417/2013/LectureSlides/16... · 2013-12-03 · Mapping •What the world looks like •Improve the accuracy of the map](https://reader033.fdocuments.in/reader033/viewer/2022042804/5f5a436b577b057c37500572/html5/thumbnails/40.jpg)
Planning Exploratory Steps• Choose motion in unexplored space to locate
additional camera nodes
• Planner cannot simulate these paths
• Evaluated 2 strategies: 1) nearest camera and 2) a randomly selected camera
CS-417 Introduction to Robotics and Intelligent Systems 40
![Page 41: CS-417 Introduction to Robotics and Intelligent Systemsyiannis/417/2013/LectureSlides/16... · 2013-12-03 · Mapping •What the world looks like •Improve the accuracy of the map](https://reader033.fdocuments.in/reader033/viewer/2022042804/5f5a436b577b057c37500572/html5/thumbnails/41.jpg)
Simulation Results
• Compared planners over many trials
• 3 realistic network types (2 shown)
• 3 methods for comparison:
– Depth-first
– Return to origin
– Return to nearest explored
41CS-417 Introduction to Robotics and Intelligent Systems
![Page 42: CS-417 Introduction to Robotics and Intelligent Systemsyiannis/417/2013/LectureSlides/16... · 2013-12-03 · Mapping •What the world looks like •Improve the accuracy of the map](https://reader033.fdocuments.in/reader033/viewer/2022042804/5f5a436b577b057c37500572/html5/thumbnails/42.jpg)
Simulated Relocalization Results
42CS-417 Introduction to Robotics and Intelligent Systems
![Page 43: CS-417 Introduction to Robotics and Intelligent Systemsyiannis/417/2013/LectureSlides/16... · 2013-12-03 · Mapping •What the world looks like •Improve the accuracy of the map](https://reader033.fdocuments.in/reader033/viewer/2022042804/5f5a436b577b057c37500572/html5/thumbnails/43.jpg)
Simulated Exploration Results
43CS-417 Introduction to Robotics and Intelligent Systems
![Page 44: CS-417 Introduction to Robotics and Intelligent Systemsyiannis/417/2013/LectureSlides/16... · 2013-12-03 · Mapping •What the world looks like •Improve the accuracy of the map](https://reader033.fdocuments.in/reader033/viewer/2022042804/5f5a436b577b057c37500572/html5/thumbnails/44.jpg)
Key Points
• Mapping requires exploration
• Exploration strategies depend on the representation
• Topological representations are the most convenient for exploration
• Two objectives:
– Explore new territory
– Improve the accuracy by relocalization
44CS-417 Introduction to Robotics and Intelligent Systems
![Page 45: CS-417 Introduction to Robotics and Intelligent Systemsyiannis/417/2013/LectureSlides/16... · 2013-12-03 · Mapping •What the world looks like •Improve the accuracy of the map](https://reader033.fdocuments.in/reader033/viewer/2022042804/5f5a436b577b057c37500572/html5/thumbnails/45.jpg)
References• B. J. Kuipers and Y.-T. Byun. “A robot exploration and mapping strategy based on a semantic hierarchy of spatial
representations”. In Journal of Robotics and Autonomous Systems, 8: 47-63, 1991.
• H. Choset, J. Burdick, “Sensor based planning, part ii: Incremental construction of the generalized voronoi graph”.
In IEEE Conference on Robotics and Automation, pp. 1643 – 1648, 1995.
• B. Yamauchi, “Frontier-based exploration using multiple robots”, In Second International Conference on Autonomous Agents, Minneapolis, MN, 1998, pp. 47–53.
• Makarenko, A.A. Williams, S.B. Bourgault, F. Durrant-Whyte, “An experiment in integrated exploration”, In IEEE/RSJ International Conference on Inte.lligent Robots and System, vol.1, pp 534-539, 2002.
• Stachniss, C. Hahnel, D. Burgard, W. , “Exploration with active loop-closing for FastSLAM”. In IEEE/RSJ International Conference on Intelligent Robots and Systems. vol.2, pp 1505-1510, 2004.
• R. Sim and N. Roy, “Global a-optimal robot exploration in slam”. In International Conference on Robotics and Automation, pp. 661– 666, 2005.
• T. Kollar and N. Roy, “Using reinforcement learning to improve exploration trajectories for error minimization”. In of the IEEE International Conference on Robotics and Automation, 2006.
• R. Martinez-Cantin, N. de Freitas, A. Doucet, and J. Castellanos, “Active policy learning for robot planning and exploration under Uncertainty”. In Robotics: Science and Systems, 2007.
• D. Meger, I. Rekleitis, and G. Dudek. “Heuristic Search Planning to Reduce Exploration Uncertainty”. In IEEE/RSJ International Conference on Intelligent Robots and Systems, pp 3382-3399, 2008.
• QUESTIONS?45CS-417 Introduction to Robotics and Intelligent Systems