Summary:

The RVCM team collaborated with MIT Lincoln Laboratory to optimize Puckboard’s scheduling tool through recommendations.

Schedulers are assisted by intelligent recommendations to calculate optimal aircrew for events while considering aircrew currency, availability and burden distribution across the unit.

Team:

Primary Audience:

Pilot, Scheduler and Aircrew members within Air Mobility Command’s C-17 Airframe.

Specific Units: 21st Airlift Squadron, Travis AFB | 8th Airlift Squadron, Joint Base Lewis-McChord | 15th Airlift Squadron, Hickam AFB

MIT Lincoln Laboratory

• Principal Investigator

• Algorithm Engineers

• Human Factors Researcher

• User Experience Designer

• Dept of Air Force / MIT AI Accelerator Director of AI Research

Discovery & Research

June 2022

• RVCM, MIT Lincoln Laboratory and Air Mobility Command stakeholders from the United States Air Force collaborated in Honolulu, HI to kick off the initiative. The main goal was to understand the algorithm capabilities, build trust and strategize the product development.

July 2022 - Present

• Continuous feedback from Aircrew to understand their needs and expectations for the “solver” recommendations.

• Soft Launches to specific USAF units who are interested in “solver” capabilities.

• User Working Group Feedback Sessions to include a more diverse pool of users.

January 2023 - Present:

• Continue to validate or invalidate if current users benefit from scheduling recommendations through 2 week iterations for the following scenarios:

  • Fuel savings

  • Optimal aircrew

  • Burden distribution

  • Upcoming events

  • Aircrew currency

Research Methodology:

• Feedback sessions via Zoom: 1:1 user interview, feedback sessions and larger group discussions

• In-person visits: observational research on USAF bases that include product demos, usability tests and focus groups

• Public Mattermost Channels: Track interest, post updates and understand user needs by chatting directly with participants.

Research Synthesis:

Affinity Mapping

Scheduler Persona & Workflow

Aircrew Member Persona & Workflow

Participant Feedback

Agile Ceremonies:

The design team works on a Scrum team and participants in

• Sprint Planning

• Design Collab

• Design Sync

• Sprint Review and Retrospectives

Additional Design Ceremonies:

• Vector Check: Check in with stakeholders to validate or invalidate design progress and pivot as needed

• Puckboard AI Working Session: Collaboration between MIT Lincoln Laboratory and Puckboard Design teams

• MIT/PB Sync: Check in to confirm design, development and solver teams are moving towards the same unified direction for integration of AI into Puckboard

• Design Audit: Confirmation that live product matches the designs

Problem Statements:

• As a Scheduler, I need to assign optimal aircrew members to events in Puckboard so I can save fuel and distribute burden effectively across my unit.

• As a Scheduler, I need to view recommendations so I can assign optimal aircrew members to events in Puckboard.

• As an Aircrew member, I need to view upcoming events and currency status so I can request events that will fulfill requirements to stay mission-ready.

Constraints & Opportunities:

• Building Trust: adjust to “Recommendation” vs. AI language. Communicate the solver as a tool enhance scheduling rather than a replacement of the scheduler role in Puckboard.

• Wait time: some participants were resistant to the solver tool because it took a few minutes due to limitations with ARMS and Platform 1.

• Design Tasks: The solver team was a resource for initial research and data. The design team was tasked with integrating the AI solver into the Puckboard product and workflow.