Overcoming bias in Tasking Interfaces

Joseph Reed

Hydrosat, Inc.

jreed@hydrosat.com

https://www.linkedin.com/in/joseph-reed-46b88718b

A bit about me

• Currently with Hydrosat
• Developing a system to deliver daily, global high-resolution thermal imagery
• Formerly with Lockheed Martin Space
• 20 years working in earth-observing “mission planning” / “mission tasking” domains
• Numerous collection phenomenologies, various missions across all orbit regimes
• US Government + Commercial missions / use cases

Tasking is: a problem statement

• To a satellite system, tasking is the dynamic variable
• Satellite / ground constraints effectively constant
• Tasking is the external input
• Describes user needs to scheduler / optimizer
• Can be optimized and tuned

Collection system bias

• Basing design on collection systems bring bias
• Every system has unique drivers
• Quality: GSD vs GRD vs IPR vs NIIRS
• Geometry: whole image? center? edges?
• Time: LEO vs GEO/HEO collectors
• Repetition:
• Daily: does 23:59 and 00:05 satisfy 2 days?
• Separation: measured from start? end? center?
• Analogy: RCS vs iMessage

Image from know-your-meme

Overcome bias: Focus on users

• Remain focused on user, not satellite
• Common use cases
• Geospatial, temporal, geometric constraints
• Phenomenology extensions
• Focus on the common user, not the system expert
• Combine real-world user + satellite system experience to assess impacts and useability, avoid the “ill-defined phase 2”

Perhaps we’re really talking about a Geospatial/temporal data ordering service

Image from knowyourmeme.com/profit

Develop standard

Joy & Happiness