GEOIDE Project Meeting

1
GEOIDE Project Meeting

• SPARCS Lab Projects
• M.L. Gavrilova

2
Research goals

• Applications of computational methods in spatial
  analysis
• GIS real-time terrain modeling and visualization
• Using methods for selection of Lp norms as
  alternative distance functions
• Pattern matching and point pattern analysis using
  computational geometry methods
• Clustering analysis using k-neighbor approach
• Application of the Voronoi diagram and Delaunay
  triangulation navigation problems

3
Students project participants

• 1. Russel Apu, Ph.D. program, Computer Science
  Department, University of Calgary, since Winter
  2005
• 2. Priyadarshi Bhattazhariya, M.Sc. program,
  Computer Science Department, University of
  Calgary, since Fall 2005
• 3. April Jensen, NSERC undergraduate research
  award recipient, Summer 2005.
• 4. Elia Palme, Patrick Terreaux, Technical
  University of Froeberg, Switzerland, Summer-Fall
  2005

4
Detailed Projects List

• M.Sc. student, Priyadarshi Bhattacharya is
  working on the Route intersection and planning
  using spatial neighborhood properties and
  Delaunay triangulation.
• Russel Apu, who has recently entered Ph.D.
  program, concentrates his research on adaptive
  memory subdivision and vision for strategic
  motion planning and navigation.
• April Jensen, NSERC undergraduate research award
  winner, was investigating topological approaches
  to risk analysis as well as graph based methods.
  She has produced 3 reports during her period of
  research on these two topics plus a literature
  review on normal vs. abnormal behavour patterns
  and their classifications.
• 4. Two visiting students from Switzerland in the
  last year of their thesis project Elia Palme and
  Patrick Terreaux have implemented the ship route
  visualization system and created corresponding
  documentation on their web site.

5
Route Intersection Project

• 1. The intersection points of ship routes are
  determined.
• 2. Delaunay Triangulation is formed on
  intersection and incident data.
• 3. Clusters are formed using edge properties of
  the Delaunay Triangulation.
• 4. Clusters are approximated as convex hulls.
• 5. Reduced Visibility graph is constructed for
  the convex and disjoint
• polygonal obstacles.
• Priyadarshi Bhattachariya next step is to
  integrate his software with MARIS data.

6
Program snapshot
7
Example of one of the queries
8
Adaptive Memory Subdivision Project (AMS)

• Russel Apu project concentrates on adaptive
  memory subdivision and computer vision techniques
  for strategic motion planning and navigation.
• Applications include
• Interactive path planning
• Obstacle avoidance
• Terrain /shore line reconstruction
• Mapping applications
• Real-time visualization
• Contour detection
• Optimization

9
AMS GIS Visualization
10
DEM Digital Elevation Model

• Contains only relative Height
• Regular interval
• Pixel color determine height
• Discrete resolution

X
Kluanne National Park
Y
11
Converting Height field data into 3D topological
mesh

• Pixel value (z) is used as Height Map
• Vertices are generated as points in 3D
• A Mesh is triangulated

X
200
255
150
100
Y
100
255
255
200
200
150
200
100
12
ROAM Real-time Optimally Adapting Mesh
Developed by Mark Duchaineau, Lawrence Livermore
National Laboratory
Quad Tree
Triangle BIN Tree

• No cracks
• Reversible
• Efficient
• Adaptive

13
ROAM Contd

• Can Subdivide when more details necessary
• Merge Split Queue
• Tree Structured

14
Navigation project

• Use Sensor Data
• Use Threat Assessment/stimuli
• Use Adaptive Spatial Memory
• Compute Path from KB
• KB is represented by ASM
• Fast (Real-time)
• Adaptive and evolutionary
• Learning, experience
• Natural behavior

15
Navigation Configuration space

• Convert Complex configuration of Agent
• Degree of freedom
• Joints
• Shape

Minkowski's Sum
16
Navigation Current Methods
Voronoi Diagram
Raster Method
Potential Method

• Maximum Clearance
• Slow
• No other Consideration

• Effective for low Res.
• Impractical for Hi.Res.

• Local Minima

17
Navigation the Agent

• ASM based method defines the agent at strategic
  and navigation stage

18
Image Distance Transform

• Smallest distance from obstacle
• Can be performed as a preprocessing
• Fast O(n2)
• Easy to obtain from range sensors
• Easy to compute c-space for agents with complex
  shape
• Algorithm also gives the coordinates of nearest
  obstacle

19
Runtime Analysis
20
Risk Analysis Project

• April Jensen, NSERC undergraduate research award
  winner, was investigating topological approaches
  to risk analysis as well as graph based methods.
• She has produced 3 reports during her period of
  research plus a literature review on normal vs.
  abnormal behavour patterns and their
  classifications.

21
Risk Analysis

• April Jensen research reports address the
  following topics
• studied background information in GIS,
  computational geometry and marine safety and
  security (report prepared)
• performed a literature review of methods used in
  outlier and abnormal behavior detection (report
  prepared).
• performed a literature review of methods used in
  modeling, storing and analyzing route data
  (review paper prepared)
• installed MARIS software for use in displaying
  and analyzing shipping track and incident data
• studied MARIS software including its capabilities
  and functionality and prepared a short summary of
  these (report)
• explored the ArcGIS features for displaying and
  analyzing shipping tracks and incident data.
• identified computational methodology for the
  analysis of data.

22
Route Visualization project

• Elia Palme and Patrick Terreaux have implemented
  the ship route visualization system and created
  corresponding documentation on their web site (a
  pilot project to assist better understand the
  route network).
• Web site http//www.elia.ch/VT3DA/

23
Goals of the project

• 1. To interlink realistic 3D visualization with
  MARIS data
• 2. To display ship routes (or any specified
  subset of them)
• 3. To compute and visualize intersections
• 4. To perform queries such as window query (which
  routes intersect given polygon), point query
  (which route is nearest to a given point),
  nearest-neighbor query (which subset of routes is
  closest to a given point/segment)
• To specify display options and different ways to
  compute point coordinates (longitude/latitude)

24
Snapshot of the system
25
System interface