1 / 16

MINEFIELD MODELING ISSUES

MINEFIELD MODELING ISSUES. MINWARA 6 9-13 May 2004 Alan Washburn Naval Postgraduate School Operations Research Department (831) 656-3127 awashburn@nps.navy.edu. MY CONTENTION…. Clearing a minefield is a complicated process that should be aided by computers.

Download Presentation

MINEFIELD MODELING ISSUES

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. MINEFIELD MODELING ISSUES MINWARA 6 9-13 May 2004 Alan Washburn Naval Postgraduate School Operations Research Department (831) 656-3127 awashburn@nps.navy.edu

  2. MY CONTENTION… • Clearing a minefield is a complicated process that should be aided by computers. • The design of a clearance TDA will be heavily influenced by available data and the concept of the clearance process. • It is important to face certain issues early in development, rather than late.

  3. ABOUT MOWING THE LAWN • Why don’t we just “mow the lawn” and go home? • Radius of effects is not definite • Environmental variations • Buried mines • Mixed mines • Inherent randomness • Mine counters, probability actuators, sensitivity • Distractions

  4. NINE ISSUES • What will be in the database? • Sweeper casualties? • Mixed mine types? • Mixed sweep types? • Optimization or evaluation? • Input estimated mine numbers? • Geometry rectangular? • Sequential clearance? • Is it a game?

  5. pd Lateral range 1. SWEEP/HUNT DATABASE • Use A and B, where AB = area under lateral range curve • Or use the lateral range curve itself • Or avoid lateral range curves, as in • SL(sweeper)-TL(environment)>DT(mine)+noise • Or use a detailed simulation such as TMSS

  6. 2. SWEEPER CASUALTIES • Assume no casualties • Tremendous conceptual simplification • UMPM, NUCEVL, UCPLN, MEDAL,... • Or assume casualties have only economic implications • Replacements available immediately, at a cost • Decouples mine types • COGNIT, MIXER(opt) • Or permit casualties to potentially spoil the plan • Clearance plan only partially completed • Mine types not decoupled • BREAKTHRU, MIXER(sim)

  7. 3. MIXED MINE TYPES • Good tactics for the miner, commonly encountered (IRAQ) • Cheap generality if mine types are decoupled • Clearance is mainly a search problem • NUCEVL, UCPLN • Potential couplings between mines • Sweeper kills • Sweeper inefficiencies (time delays) • Threat to traffic (all mine types contribute)

  8. 4. MULTIPLE SWEEP TYPES • Order of entry important if sweepers are vulnerable • 7! = 5040 possible orders with seven sweep types • Can multiple types sweep at the same time? • Fratricide danger • Helicopters usually precede ships • Aid tactical choice of clearance type • Sweep or hunt? • Mechanical gear or sled? • Remote vehicles?

  9. 5. TACTICS OPTIMIZATION • Tactical questions • Track spacing • Time on task by sweep type • Equipment settings • Measures of effectiveness • Total clearance time (T) • Clearance casualties (C) • Target traffic casualties (H) • MIXER (opt) minimizes C + H subject to constraint on T • COGNIT minimizes C subject to constraints on T and H, etc.

  10. 6. ESTIMATED MINE NUMBERS • Number of mines may need to be an input • Required for most optimization issues • It is not true that SIT + clearance level = 1 • Number of mines notoriously random • Initial guesses will be WAGs • Need mine inventory database by country • Updating by evidence (Bayes) • MIXER requires mean and standard deviation by type • COGNIT assumes Poisson distribution

  11. 7. GEOMETRY Conventional rectangular (all but MEDAL) MEDAL allows arbitrary path orientations

  12. 8. SEQUENTIAL CLEARANCE • Conventional method has one plan, no feedback • Sequential method is to clear, observe, clear, … • Clearance achieved in stages • Plan for stage n + 1 depends on results in stage n • Summary statistics passed from stage to stage • MIXER optimizes within a stage, but not between • MEDAL passes summary statistics between stages • COGNIT and most other clearance programs are conventionally oriented

  13. 9. GAME THEORY • Minefield clearance is a competition • Mother Nature doesn’t make minefields • Why not use the zero-sum theory? • Clear opposition of interest • Mine sensitivity settings, for example, are an enemy choice not observable by the sweeper

  14. 10. ADVANTAGES OF GAME THEORY • Lack of requirement to guess things that cannot possibly be known, such as mine counter settings • Robustness of resulting tactics

  15. 11. DISADVANTAGES • Sensitivity to objective function • Optimal tactics may be mixed • “Flip a coin to decide whether to sweep or hunt first” • For a mine, “flip a coin to decide whether to detonate” • Computation is still an issue • But computers are getting faster…

  16. SUMMARY • Crucial decisions about clearance model should be faced early in development • Resulting software strongly affected • Effects are interrelated • ………………………….……..Questions?

More Related