Repository files navigation

SORAL is theSARBayes Optimal Resource Allocation Library,a programmer's API of optimization algorithms tailored to Search And Rescue (SAR).

• Written in C++ (fast)
• Python & Node wrappers
• Charnes-Cooper algorithm for single resources
• Washburn'sJOTA algorithm for multiple resources
  • JOTA algorithm also in MATLAB
  • (MATLAB developed first, fromJOTA paper and FORTRAN code)
  • Many thanks to Prof. Alan Washburn
• Manual (user-specified) "algorithm"
• Clean object-oriented structure allows plug-in algorithms

SORAL is an abstracteffort allocator: it determines howmucheffort to apply to each region, but nothow to apply it. Thosetactics or path planning are left to the calling program, or thehumans. However, SORAL is fast, and can be used to provide a goodstarting point, or test variations.

SORAL is a black box with functions to best allocate a set of SARresources across areas, where "best" means "find the lost person fastest".Being open-source means the whole SAR and operations-researchcommunity can potentially review and improve the code.

Effort allocation requires an initial probability map for the lost person,and three key kinds of information about each search resource type:

• Travel Speed (can vary by region)
• Hours Available (total)
• Sweep Width (can vary by region)The first two should be known to search planners, or quickly estimated.Sweep Width ideally has to be measured experimentally, and isspecific to a <sensor, target, environment> combination.However,after analyzing 20+ experiments, empirically it seems Sweep Widthcan be estimated quite well from by applying a correction factor(between 1.0 and 2.0) to the "average maximum detection range"obtained from a "Northumbrian Rain Dance" or "AMDR" procedure.That fact opens up optimal allocation to practical use.

The SORAL API is a simple, clean, modular collection of algorithms,each of which is optimal in its own domain. By virtue of pluggingin to SORAL, each also provides important values for the end-user,such as probability of detection, coverage, and others. Also:

• Tested
• Documented
• Open-source
• Etc.

AGMSAR is a text-based resource-allocation program that uses SORAL.Perhaps it should get its own repository that refers to SORAL.

• 2017-11: Changed to MIT License: updated this file and LICENSE.md.
• 2017-04: Cleaned code uploaded to GitHub
  • Code cleaned by Sarah George atHeartfelt Tech
  • Using a 2003? snapshot, the 2008 delivery, andTom Russo's 2007 snapshot
  • Wrapped for Python and Node (using SWIG to wrap)
  • Need to update docs
• 2015: Tom Russo's ESCAPE 2015 presentation on hisSearch Theory Page includes detailed SORAL examples.
• 2011: Hosting Revolution vanished overnight taking down the SARBayes website. They were also my backup service, so I totally lost the CVStrac version history, as well as some non-SORAL content.
• 2008-02: Charles purchased rights from AGM and got a slightly updated version of the code & docs.
  • AGMSARpackage20080222.zip
• 2003 - 2011: Available on sarbayes.org
  • Self-hosted CVStrac and I believe CVSdude
  • Ported to Google Code, and failed to migrate therefrom
  • SARBayes' hosting company vanished, with hosted copy2003: Code Review and Release of vX.x under the Gnu GPL, copyright AGM.2001: SORAL begun as a summer project at Monash University.
  • Advisors: Charles Twardy, Ph.D. & David Albrecht, Ph.D.
  • AGM Team: Andre Oboler, Gareth Thompson, Michael Eldridge

SORAL is freelibre software available under the MITLicense. It was changed from Gnu GPL in November 2017. The MITlicense is more permissive: code using SORAL no longer needs tobe released under the Gnu GPL. You may incorporate SORAL intoproprietary code.

• Dr. Charles Twardy
• AGM: Andre Oboler, Gareth Thompson, Michael Eldridge
• Prof. David Albrecht at Monash U.
• Sarah George @wellbeingwebsites
• Tom Russo
• Extra thanks to Prof. Alan Washburn (NPS, emeritus), Jack Frost (USCG, ret'd), Dr. Bob Koester