Path Finding In Hazard Terrain

Post on 26-Jan-2017

165 views 0 download

Transcript of Path Finding In Hazard Terrain

By

Oren Koler

Bar Ilan University

1. Introduction

2. Battle lab

3. Terrain Analysis

4. Path Finding (with Terrain Analysis)

5. My project highlights

Topics

Unmanned Vehicle patrol

Guardium

Unmanned Vehicles

1. Path finding

• Building a graph

2. Path planning

• Where to go on the graph at any given moment

Patrol

1. Find shortest route between points

2. Good for solving mazes

3. Field based heavily on Dijkstra's algorithm

Path finding

• Creating the graph

• Find observation points

• Create routes between observation points

• Take hazard forces into consideration

• Base on Real achievable types of data

My Contribution

Find Shortest Route

Dijkstra's Algorithm A* Algorithm

Battle Lab

1. Part of the IDF ground forces weapons development department

2. Assists in evaluating

1. New weapons

2. New combat doctrines

3. Aided in human engineering for new systems

3. The lab can simulate a virtual battle space

4. It’s techniques are similar to the gaming industry

5. The lab is located outside “Tel Hashomer”

6. Come visit

Battle Lab

Battle Lab on the Map

1. I am in-charged of the CGF team

2. The team is responsible of all virtual autonomous entities

3. Virtual entities types are vehicles & humans

4. The battle space can be in open & urban areas

5. Combat can be close & far range

6. Entities use COTS product called B-Have for path finding

Computer Generated Forces

1. Creates Mesh Navigation maps

2. Uses online A* path finding

B-Have

Navigation Mesh

1. An abstract data structure used in AI applications

2. Aids agents in path-finding through large spaces

Navigation Mesh

1. Done offline, no dynamic path finding

2. No collision avoidance algorithms

3. A lot of needed starting materials

I. DTM

II. vector data

4. Large terrain

• there are many paths from one point to another

Project Highlights

1. Responsible for intel gathering by using visual means

1. Satellites

2. Unmanned aerial vehicle

3. Ground Sensors

2. The map unit is responsible for data distribution

3. Route planning is based on the maps received from them

9900 division

• The collection, analysis, evaluation, and interpretation of geographic information on the natural and man-made features of the terrain, to predict the effect of terrain on military operations

3D Terrain Analysis

Terrain Analysis Tools

1. Also known as

1. digital Terrain model (DTM)

2. digital surface model (DSM)

2. A 3D representation of a terrain's surface

3. Represented as

1. A raster (a grid of squares)

2. A vector-based triangular irregular network (TIN).

4. Produced by types of radar

Digital elevation model (DEM)

TIN

1. Hypsometric tinting

2. Contour map

3. Hill shading

4. View shed

5. Slope

DEM Products

1. Applies different color symbols to represent elevation or depth zones

Hypsometric Tinting

1. Contour lines connect points of equal elevation

2. They are very intuitive for humans

Contour Lines

1. Setting a hypothetical light source

2. Calculating the illumination values for each cell

3. It can greatly enhance the visualization of a surface for analysis

Hill shade

1. Identifies the cells in an input raster that can be seen

2. It is useful for finding the visibility

3. For instance, finding a well-exposed places for communication towers

Viewshed

• Slope measures the rate of change of elevation at a surface location

Slope

1. A coordinate-based data model

2. Represents geographic features as points, lines, and polygons

Vector Data

1. Create a cost distance layer

I. Data analysis

II. Reclassification

III. Accumulate

IV. Distance cost + Direction

2. Find path

Path Finding with Terrain Analysis

• Make all data sets simpler and in the same range

Reclassification

• Sum data sets with weights

Accumulated Cost layer

Cost Distance layer • For each observation point create a cost distance layer

Direction Layer • For any point show where to go

Procedure Example

Input

1. Terrain data

I. Elevation map (DTM)

II. Vector data

1) Roads

2) Rivers

3) Houses

4) Forests

5) Fields

6) Walls

7) Man made obstacles

8) Soil types

2. Vehicle configuration

The Project

Output

1. Observation points

• An array of geographical points ( x,y,z)

2. A graph of Routes between observation points

• For each route:

I. An array of geographical points

II. The cost of each route

The Project

• How to find observation points

• Find points that are higher than all there surrounding points

Finding Peaks

1. For each observation point

1. Create a cost layer

2. Find paths to all other observation points

2. Dissolve graph

Creating Graph

Example

A

E

G

F

C

D

B

• Observation A routes to all other observation points

• Full Graph

• Some routes are similar to other routes

Example

A

E

G

F

C

D

B

• Removing all unnecessary routes

Graph Dissolving

A

E

G

F

C

D

B

1. Threat types

I. Short range AT units

II. Improvised explosive device

2. Patrol routes are known to enemy

• No point in hiding !!!!

3. Avoid Areas with high probability of ambush

• Unseen areas from most angles

I. Forests

II. Ravines

Path planning in Threatened Area

• Data resolution isn’t that high

• Cant analysis small cracks in the mountain

• Prefer high ground terrain

• Stay away from forests

• Select routes that are watched by our observers

Path planning in Threatened Area

• Is the path from A to B going to be the path from B to A

• Are threats based on direction ?

• Should the cost be different ?

• How to decide the weights for each type of layer ?

• Is shorter path preferred than safer ?

• Is a steep short path preferred that a long moderate path ?

• Should the weights be set according to vehicle fuel consumption ?

• How should routes be dissolved ?

Things to think about

1. For real world path finding- use terrain analysis

2. Battle lab also works in this field and can assist others

3. Problem with data not being precise enough

4. For creating a real word graph

I. Create data set layers from DEM & vector data

II. Find peaks

III. Reclassify & accumulate data into a cost layer

IV. Create a cost distance layer for each point

V. Create paths/routes from point to al other points

VI. Merging all routes and dissolve similar routes

5. Prefer high ground routes that are watched

6. Problem in deciding weights and costs

Summary

The End