AI Energy Prediction for Autonomous Military Vehicles

Exergi Predictive is an AI-driven energy management system for autonomous military vehicles, selected for the Army’s Robotic Combat Vehicle (RCV) program. Robotic combat vehicles depend entirely on battery power — energy demands fluctuate unpredictably based on terrain, mission type, and operational tempo, creating a 3-second operator decision window where predictive UI is critical. As Lead Product Designer over ~18 months, I translated probabilistic AI outputs into a legible, field-tested Figma prototype for rugged tablet hardware operating in GPS-denied, contested environments. The design was adopted for React development by The Dreamers; the capabilities document featuring my graphics was the most-taken item at AUSA 2024.

This was an ~18-month engagement for Exergi Predictive, sourced through Visual Logic, where I served as Lead Product Designer alongside Andrew Kotz as PM. The platform targeted rugged tablet hardware running edge-computing software in GPS-denied, signal-contested environments. Deliverables included a Figma prototype, an illustration system, custom icon design, infographics, and sales materials for investor and defense-buyer audiences.

Military energy prediction presents inherent difficulties, particularly for autonomous systems. Robotic combat vehicles depend on battery power for movement, navigation, communication, and payload operation.

Fluctuating energy demands stem from terrain, soil composition, wind conditions, mission type, and operational tempo — creating inefficiencies, capability reduction, or mission failure.

The Army required an AI-powered planning, prediction, and management tool that operators could intuitively use under pressure in chaotic conditions.

The project followed a dynamic, iterative approach reflecting emerging technology complexities. Weekly meetings with the PM, four user visits, and ongoing stakeholder collaboration shaped the solution. No design was sacrosanct; continuous iteration drove refinement.

Key Human-Centered Activities

Stakeholder Collaboration

• In-depth discussions defining goals, technical limitations, and opportunities
• Ongoing collaboration with subject matter experts and technical teams

Research

• Analysis of public and private resources on autonomous vehicle energy management
• Proto-personas and journey maps capturing user needs and mission scenarios

Information Architecture

• IA diagrams and sitemaps defining scope and content organization

UI Design

• Comprehensive UI framework including widgets, graphs, and data visualizations
• Extensive iteration refining visual hierarchy and usability

Custom Visual Assets

• Vehicle illustrations and non-standard icons aligned with project requirements
• Infographics communicating Exergi’s value proposition

Interaction Design

• Interactions across widgets, sections, and screens ensuring seamless workflows

Development Feedback

• React prototype critique aligning technical implementation with user needs

The human-centered process articulated, refined, and amplified Exergi’s value across military echelons. Key solutions included:

Mission Input & Route Selection — Refined mission criteria and amplified route success probability visualization.

Energy Widget — Dynamic widget allowing users to conceptualize energy via time, kWh, or distance. Expandable for detailed breakdowns across all energy aspects.

Energy Dashboard — Comprehensive screen displaying all operator-critical energy information.

Prediction vs. Actual — Operators can toggle predictions on/off when missions deviate substantially, preventing prediction noise.

REx & Fan Controls — View-only tool permitting generator and fan control with status display, event schedules, and manual override toggles.

Map, Route, Flags, & Corridor — Dynamic map displaying energy-optimal corridors, event flags, and data layers (contested areas, etc.).

Timeline & Events — Context-adaptive timeline layered with event flags correlating to map data.

Fuel Savings Optimization — Special setting allowing generator parameter adjustment for fuel conservation.

Thermal Optimization for Stealth — Generator parameters reducing heat signature for stealth operations.

Location Estimation — Precision location algorithm using energy and physics models for signal-denied environments, displaying confidence ranges.

The design partnership uncovered a high-value solution for high-value users, recognizing that Exergi’s primary value targets strategic-level thinkers and planners.

“You’ve increased our level of professionalism by a large step.” — Will Northrop, Co-founder, Exergi Predictive

“The booth rep at NAMC said our capabilities doc was the #1 taken one [at AUSA 2024] and had great graphics!” — Andrew Kotz, PO, Exergi Predictive

The project attracted interest from other AAL cohort companies seeking similar design assistance.

This assignment began as an ambiguous request with minimal requirements — requiring ground-up product development. Key design challenges included:

1. User Clarity — Each screen required clear, actionable objectives avoiding information overload.
2. Data Prioritization — Sorting, grouping, and visualizing complex data into digestible, actionable insights.
3. Autonomy vs. Manual Control — Balancing autonomous RCV actions with seamless manual override capability.
4. Predictions vs. Actuals — Presenting historical data, real-time status, and future predictions simultaneously.
5. Energy Dynamics — Accounting for EREV drivetrain complexities with dual energy sources (electric and generator).
6. Unit Conversion — Bridging user needs across distance (off-road range), time (operational endurance), and kWh.

The program has since changed government oversight, making long-term outcomes uncertain. The design foundation provides a substantial starting point for future applications.