AURORA INTELLIGENT SCHEDULING FRAMEWORK 13 M IN PREVIOUS

AURORA INTELLIGENT SCHEDULING FRAMEWORK ($13 M IN PREVIOUS NON-DOD SALES) • A major component of planning is scheduling (assigning resources and times to tasks) • Human experts are very successful at building highly optimal schedules. • • Aurora is a general-purpose scheduling framework that applies experts’ processes and knowledge to automate the scheduling process. • • BUT this requires lots of training and experience, and building a good schedule manually requires a great deal of time, effort, and expertise. Quickly produces effective schedules that respect constraints and maximize objectives: typical 10 to 40% more tasks accomplished with the same time and resources compared to other automated solutions. Applied to many dozens of shockingly diverse applications, including scheduling equipment and personnel Aurora has won every single one of the dozens of scheduling competitions it has participated in. Designed as an extensible framework to build domain-specific scheduling tools. • Each domain requires integration with existing tools and processes to access and transform data 1

SCHEDULING APPLICATIONS • • • Intelligent Paring Assistant: Assigning tasks to aircraft in Polaris Alpha’s MAAPTK • • • Smart COordination of Unmanned Teams (SCOUT): Autonomously assigning tasks to UAVs • • Plus Viper Naval Planning Systems (Air, Surface, Submarine) OPIR Scheduling: Assigning ISR collection tasks to satellite-based OPIR sensors Space Surveillance Network (SSN) Scheduling: Assigning SSA tasks to SSA sensors Ballistic Missile Defense: Assigning Interceptors and Sensors to suborbital targets AFSCN Scheduling: Satellite Support Communication Scheduling (Air Force Launch Success Story) Naval A/C & UAV Scheduling: Assigning naval ID tasks to manned and unmanned A/C Many very tangled/complex scheduling applications (e. g. 787 Dreamliner Assembly, Nuclear Sub Refurbishment) Plus SHERLOC Space Situational Awareness (SSA) system for NSDC 2

Automatic AF Satellite Control Network Scheduling Stottler Henke’s MIDAS: Automatic AF Satellite Control Network Scheduling 10 Previous Failed Attempts; Stottler Henke was the Only Success $10 s of Millions Previously Wasted on Failed Attempts vs $1. 5 M for MIDAS 20 Man-hours for humans to make 24 -hour schedule reduced to 30 minutes 3

Scheduling Comparison Example: Micro. Soft Project Versus Aurora 4

5

OPIR Scheduling: Assigning ISR Collection Tasks To Satellite-Based OPIR Sensors 6

• OPIR Video goes in place of this slide 7

Space Surveillance Network (SSN) Scheduling • Assigning SSA tasks to SSA sensors • 3 x reduction in orbit parameter errors with same sensor resources and time • 50, 000 observation tasks allocated to 600 sensors at 100 sites while considering 1. 5 Million visibilities in 30 seconds on a standard desktop computer • Quick reaction: 30 ms (100 in 3 seconds) to schedule high priority target on next visibility/sensor 8

Ballistic Missile Defense • Assigning Interceptors and Sensors to suborbital targets 9

Many Very Tangled/Complex Scheduling Applications • 787 Dreamliner Assembly • • • $40 B Annual Assembly Lane completely scheduled by Aurora Thousands of Tasks/Plane Many planes in process simultaneously, competing for resources Many many resources and constraints per task Unique issues, e. g. tasks creating resources (e. g. floor space) • Nuclear Sub Refurbishment) • [need impressive numbers from Rob] 10