P1254156865ldqNS

1
ERC Enhancements OneSAF Users Conference March
2009
2
ERC Overview

• Acronmys and definitions
• ERC Past Where we came from
• ERC Present Where we are now
• ERC Future Where we are going
• Questions??

3
ERC Acronyms and Definitions

• DVED Database Virtual Environment Development
  component of SE Core tasked with database
  generation.
• ESS (Earth Skin State) Returns elevation and
  trafficability data at a location.
• ITIN (Integrated Triangular Irregular Network)
  Triangle mesh used to represent the terrain data
  in the OTF.
• LOS Line of Sight
• OSI OneSAF Integration component of SE Core
  that is integrating OneSAF into CCTT and AVCATT
• OTF (Objective Terrain Format) Stores the
  terrain and feature data for OneSAF.
• - OneSAF 1.0 and 1.1 use OTF 5.0
• - OneSAF 1.5, 2.0, and 2.1 use OTF 6.0
• - OneSAF 3.0 will use OTF 7.0
• Repoly Process of updating the ITIN with new
  triangles that represent dynamic features such as
  craters.
• SE Core Synthetic Environment Core
• UHRB (Ultra High Resolution Building) Buildings
  with doors, windows, and interiors.

4
ERC Past

• Directed re-use of WARSIM Synthetic Natural
  Environment (SNE) code
• Over 700k sloc providing foundation
• Updated, modified, enhanced to meet the needs of
  OneSAF environment

5
ERC Present

• Supporting the program requirements as directed
  by TPO
• Supporting the integration of OneSAF into SE Core
• Supporting the generation of OTF databases by
  DVED
• Supporting the LVC capability
• Supporting external customers

6
ERC Design Constraints

• Various resolutions for source data
• Low (DTED0, DTED1, etc.)
• Medium (DTED2, etc.)
• High (LIDAR, IFSAR, etc.)
• Various sizes of geographic areas (degrees)
• Small (Less than 2 x 2)
• Medium (2 x 2 to 5 x 5)
• Large (Larger than 5 x 5)
• Scalability
• CPU Usage (ERC interfaces are called often)
• Storage (RAM and Disk Usage)
• Composability
• Various fidelity of models running on a single
  terrain database with various resolutions of
  terrain source data (Needed to support LVC
  exercises).

7
ERC in LVC

• Live
• OneTESS
• High resolution data (DTED 5)
• 10 centimeter accuracy
• OTF and ERC stored on Player Unit
• 1 millisecond allocation for LOS
• 5880 km2

• Constructive
• OneSAF
• BDE and below
• 30k entities
• Low, medium, high resource entities
• Multi-level resolution terrain databases
• Low/medium/high size terrain database
• WARSIM
• Division level
• Thousands of units
• Low resolution terrain databases
• Medium/high size terrain database

• Virtual
• SE Core/AVCATT/CCTT
• Support forest canopies and individual trees
• Medium size areas
• OTF viewable in current h/w

8
2.1 to 3.0 Enhancements

• OTF Enhancements
• OTF API Enhancements
• Scalability
• Storage
• CPU Usage
• Threading
• Enhancements from SE Core/DVED
• Terrain Databases
• SWA
• JRTC
• Robotics
• Fort Stewart

9
7.0 OTF Enhancements

• Improved spatial organization of feature data -
  improved CPU usage for ERC interfaces(LOS, ESS,
  etc.)
• Better road network support
• Improved CPU usage for road route planning
• Smaller road network files in OTF
• Local coordinates - Use 32-bit floats vice 64-bit
  floats to store terrain and feature data yielding
  smaller OTF
• Shrunk UHRB storage - Used 32-bit floats instead
  of 64-bit floats to store vertices

10
7.0 OTF API Enhancements

• Added support for 64-bit Linux
• Allows access to larger memory heap for building
  larger OTFs
• Decreases time to build OTFs
• Stressed OTF API by building OTFs with
• 1,048,000,000 ITIN triangles
• 25 million trees
• Various PTR fixes to support OneSAF, TerrSim,
  DVED, and RUGUD

11
7.0 OTF Storage Improvements
12
32-bit LOS CPU Usage
32-bit Linux
Intel(R) Pentium(R) 4 CPU 3.40GHz 3 GBs RAM
13
64-bit LOS CPU Usage
64-bit Linux
Dual Intel(R) Xeon(R) 4 CPU X5482 _at_ 3.20 GHz 64
GBs RAM
14
Overall OneSAF Performance

• Performance comparison of OneSAF v3.0 and
  previous releases on Dell 670 CHP

15
Threading Enhancements

• Threading Improvements
• Changed to spin locks for Linux mutexes
• Improved performance of ERC software in critical
  sections
• Improved threading performance of Terrain Tile
  Generator

16
SE Core Enhancments

• OSI
• AVCATT Terrains converted to OTF
• I1
• P2
• CCTT Terrains converted to OTF
• Baghdad
• Fort Hood
• Improved following functionality
• LOS checking for individual trees
• Collision detection
• Bridge models
• Steep slope calculation
• RWA Support
• PTR Fixes

• DVED
• Terrain Databases
• Fort Stewart
• Fort Irwin
• Fort Hood
• Fort Riley
• Pinon Canyon
• OTF API Bug Fixes

17
SWA OTF

• Built to support the 2.0 BDE and 3.0 BDE
  exercises
• Contains the following data
• 9 GeoTiles built with DTED1 and 112.5k map scale
  data from TEC
• 8,240,220 ITIN triangles
• 194,934 features
• 36,389 areal buildings
• 3,498 areal built up regions
• 113,803 linear roads
• 23,994 linear rivers
• 876 linear bridges
• 318 UHRBs
• 2 UHRB tunnels

18
SWA - Zoom Out of Buildings
19
SWA - Zoom In of Buildings
20
SWA UHRBs in ARES
21
JRTC MOUT Site in ARES
22
JRTC UHRBs in ARES
23
ERC and Robotics

• Baghdad geo-typical database
• Will be used to train operators using robots to
  remove IEDs
• 1 GeoTile with high resolution inset
• Contains the following data
• 2,797,892 ITIN triangles
• 106,106 features
• 29,044 areal buildings
• 65,497 linear roads
• Urban clutter
• 10 dumpsters
• 8 oil drums
• 22 point rubble
• 61 vehicle barriers
• 22 wire obstacles
• Curbs in high resolution inset

24
Urban Clutter in Robotics OTF
25
Fort Stewart OTF from DVED

• Built to support AVCATT and CCTT training
  requirements
• Built to run on AVCATT and CCTT Visuals
• Contains the following data
• 8,978,035 ITIN triangles
• 7,784,762 Features
• 7,680,599 individual trees
• 47,851 linear roads
• 29,598 point buildings
• 2,797 areal buildings
• 6,774 linear rivers

26
Fort Stewart in MCT
27
Fort Stewart in ARES
28
ERC Future

• Supporting future TPO and domain requirements
• Supporting the integration of OneSAF into SE Core
• Supporting the generation of OTF databases by
  DVED
• Supporting the LVC capability
• Supporting external customers

29
3.0 to 4.0 Enhancements

• RIB 45a
• 8.0 OTF Format (GeoTile Constraint)
• OTF format for OneSAF releases 3.1 thru 4.0
• Added ability to build OTFs that are smaller than
  a GeoTile (1 x 1)
• Will decrease OTF storage
• Needed by OneSAF, SE Core, DVED, and OneTESS
• Obstacle Avoidance Improvements
• Work includes
• Moving OneSAF obstacle avoidance software from
  Java to C
• Doing extensive testing on OneSAF built OTFs
• Doing extensive testing on DVED built OTFs
• Benefits
• Better OneSAF BDE performance
• Better performance on dense terrains (ERC is
  currently building a SWA OTF with approximately
  550,000 areal buildings)
• SE Core needs this functionality to deal with
  millions of individual trees in DVED built OTFs

30
Obstacle Avoidance in Fort Stewart
31
3.0 to 4.0 Enhancements

• RIB 41c
• Dynamic Terrain
• Supports a higher fidelity dynamic terrain
  solution thru Repoly of the ITIN
• Examples of supported features are
• Craters
• Tank defilades
• Infantry trenches
• Walls
• Breach holes in walls
• Uses server to resolve dynamic terrain requests
  that happen in the same location at the same time
• Includes support to make placement of dynamic
  terrain easier
• Needed by SE Core to satisfy virtual dynamic
  terrain requirements

32
Crater Example in MCT
33
3.0 to 4.0 Enhancements

• RIB 125 (Scalability)
• Collision Detection
• Affects entity and feature collision detection
• Includes improving accuracy and performance of
  collision calculations
• Needed to support OneSAF BDE scalability
• Needed to support SE Core virtual requirements
• UHRB
• UHRBs use KD trees for storing geometric data
  (surfaces for walls, doors, apertures, furniture)
• Need to improve the following areas in KD trees
• KD tree organization
• KD tree runtime software
• Better KD tree performance will improve
  performance for the following ERC interfaces in
  UHRBs
• LOS
• ESS
• Route planning
• Smoke
• Higher fidelity smoke model
• Will support smoke clouds and plumes
• Needed to satisfy SE Core requirements

34
3.0 to 4.0 Enhancements

• SE Core RWA Path Planning Interface
• Needed by SE Core to satisfy AVCATT requirements
• Supports NOE and Flight Profile
• Full implementation in C allows for fast direct
  access to terrain data
• Path corners are smoothed so that RWA physical
  model can follow the path

35
RWA NOE Path Planning Example
36
LVC OTFs

• Must support Live, Virtual, and Constructive
  entities on 1 OTF
• ERC is building SWA
• Stresses flat terrain with lots of features
• 9 GeoTiles (3 x 3)
• Center GeoTile is high resolution Live inset
• 8 surrounding GeoTiles are medium resolution for
  Virtual and Constructive entities
• ARA is building NTC
• Stresses mountainous terrain
• Being built with LIDAR and IFSAR terrain data
• Will investigate using 40 centimeter terrain data
  for Fort Irwin Main Post

37
LVC SWA Statistics

• Live inset currently contains
• 10,233,195 ITIN triangles
• 560,809 features
• 538,520 areal buildings
• 3,487 point buildings
• 73,719 linear roads
• 11,329 linear trails
• 17,791 individual trees
• 209 linear fences
• 161 linear walls
• Still need to add UHRBs and UHRB tunnels to Live
  inset
• Still need to build Virtual and Constructive
  GeoTiles

38
LVC SWA in ARES
39
LVC NTC in MCT
40
LVC NTC in ARES