Skip to content
    Collaborative autonomy ecosystem / active programs

    Sense. Decide.
    Move as one.

    Arc connects operators, sensors, autonomous vehicles, deployable model operations, and resilient mission networks in one collaborative system. Perception, planning, swarm coordination, and decisions stay close to the mission—with full communications PACE across program-selected radios, mesh, LTE/5G, and SATCOM.

    One runtime / eight systems
    Arc-Edge v1Arc-UASArc-UIArc-RFArc-FusionArc-UxVArc-SentryArc-Forge
    Multi-INT sensingVehicle autonomyFull comms PACEModel operations
    Explore the family

    Rownd program credentials

    Program deliveryActive customer contractsSupporting current defense and dual-use work
    Government entityCAGE 8KHL2Rownd, Inc. · UEI K5LLASNQNPG3
    Arc product family

    One platform.
    Many edges.

    Air, spectrum, sensors, vehicles, and the operator layer share one sovereign runtime. Each system has a distinct job; together they coordinate through program-selected PACE communications and continue when a bearer is degraded or unavailable.

    ARC / E1

    Arc-Edge v1

    Mission-ready carrier + PACE

    One flight-optimized carrier board with compute, acceleration, modular radios, and the Arc mission stack provisioned for the target vehicle. Change the link plan without rebuilding the autonomy layer.

    • ■Three universal radio slots
    • ■Radio · mesh · LTE / 5G · SATCOM
    • ■Software and models preloaded
    • ■MAVLink vehicle integration
    Open system brief
    ARC / UAS

    Arc-UAS

    Autonomous aerial intelligence

    On-aircraft perception, planning, and mission execution for 10-inch quadrotors through 5-foot fixed-wing platforms. The aircraft closes the loop from validated intent through recovery.

    • ■Quadrotor + fixed-wing platforms
    • ■Up to 2-hour fixed-wing endurance
    • ■PX4 · ArduPilot · MAVLink
    • ■Collaborative mission behaviors
    Open system brief
    ARC / UI

    Arc-UI

    Mission-native operator control

    A field-first replacement for fragmented ground-control workflows. Plan, fly, monitor video and detections, and intervene from one responsive web, Android, phone, tablet, or desktop surface.

    • ■Map · split · stream modes
    • ■Voice and touch tasking
    • ■Routes · zones · targets
    • ■Telemetry · video · detections
    Open system brief
    ARC / RF

    Arc-RF

    Signal intelligence and triangulation

    Man-portable RF intelligence that treats spectrum observations as a vision problem, classifies labeled signatures at the edge, combines bearing evidence, and publishes mission-ready events.

    • ■Controlled 37-class benchmark
    • ■Phase-interferometry direction finding
    • ■Cooperative bearing fusion
    • ■ATAK / WinTAK output
    Open system brief
    ARC / FUS

    Arc-Fusion

    Agentic edge sensor fusion

    A provenance-aware evidence layer that correlates program-selected vision, RF, LiDAR, acoustic, telemetry, and TAK observations without hiding uncertainty or contradictory sources.

    • ■Program-selected sensor adapters
    • ■Source + confidence retained
    • ■Policy-bounded mission tools
    • ■Reviewable evidence picture
    Open system brief
    ARC / UXV

    Arc-UxV

    Autonomy for any vehicle

    A vehicle-independent autonomy layer for air, ground, surface, subsurface, and mission-specific platforms. Run connected, through a degraded link, or in configured low-signature profiles.

    • ■Common autonomy core
    • ■Platform and payload adapters
    • ■Link-optional execution
    • ■Multi-vehicle coordination
    Open system brief
    ARC / SEN

    Arc-Sentry

    Persistent autonomous sensor mesh

    A distributed small-UAS-class sensing system designed to perch, watch, classify local events, reposition, recharge, and share concise conclusions when the selected platform and program support each behavior.

    • ■EO/IR, RF, and acoustic options
    • ■Tile-aware aerial perception
    • ■Configured swarm tasking
    • ■ATAK / CoT integration
    Open system brief
    ARC / FRG

    Arc-Forge

    In-theater model operations

    A deployable model-operations node that turns reviewed mission evidence into target-aware candidates, comparison results, and approved edge artifacts without requiring a continuous cloud pipeline.

    • ■Observe-to-export model loop
    • ■Adapt + quantize + evaluate
    • ■Champion / challenger gate
    • ■Identity + evidence + rollback
    Open system brief
    System architecture / shared runtime

    Different missions.
    One operational brain.

    Arc normalizes sensor evidence into a shared local world model. Mission-tuned models interpret the scene, propose or execute policy-bounded actions, and synchronize tasking and conclusions across the available mission network—without depending on a cloud round trip.

    Conclusions instead of raw feedsShared swarm stateBearer-aware transportFaster decisions
    Operational applications

    One autonomy core.
    Different mission sets.

    Each deployment is bounded by the approved aircraft, payload, communications loadout, autonomy behaviors, and mission authority.

    01

    ISR & Area Search

    Execute validated routes and search patterns, correlate onboard detections with mission geography, and return concise observations instead of requiring constant raw-feed monitoring.

    02

    Force Protection

    Run bounded patrol, overwatch, and alerting missions with explicit geofences, recovery criteria, and operator intervention paths.

    03

    Search & Rescue

    Task compact quadrotors or longer-endurance fixed-wing aircraft to search named areas, identify mission-relevant observations, and report locations over the available PACE path.

    04

    RF & EW Mapping

    When paired with the appropriate Arc-RF payload, collect and correlate spectrum observations while the aircraft executes the approved flight plan.

    05

    Collaborative Multi-UAS

    Use program-qualified swarm behaviors to share mission context, distribute bounded tasks, and preserve local aircraft safety authority across the team.

    06

    Critical Infrastructure

    Inspect linear and distributed assets with onboard processing, configurable endurance, and store-and-forward reporting when connectivity is constrained.

    Mission logic / representative playback

    See how the mission closes the loop.

    Mission feed A / ISRSimulated logic
    Autonomous intelligence, surveillance, and reconnaissance area searchAn aircraft follows a bounded search pattern, identifies observations locally, and sends a concise report to the operator.AOI-14 / SEARCH BOUNDSOBSERVATION / 0.91LOCAL CLASSIFICATIONROUTE ACTIVEONBOARD PERCEPTIONCONCISE REPORT

    Search, understand, report

    Arc-UAS executes the approved search geometry, evaluates observations onboard, and forwards mission-relevant conclusions instead of demanding a continuous raw-feed watch.

    1. Step 1Bound the search
    2. Step 2Classify locally
    3. Step 3Report with context
    Mission feed B / ProtectionSimulated logic
    Policy-bounded force protection missionAutonomous nodes watch a defined perimeter, detect an observation at the boundary, and cue the operator without exceeding configured authority.ASSETWATCH-2BOUNDARY OBSERVATIONCUE / OPERATOR REVIEWCONFIGURED AUTHORITYPATROL · ALERT · HOLD · RETURNINNER ZONEPATROL RINGHUMAN AUTHORITY

    Watch within authority

    Configured nodes patrol a defined perimeter, distinguish a boundary observation, and cue the operator while hold, return, and intervention paths remain explicit.

    1. Step 1Define the perimeter
    2. Step 2Watch and detect
    3. Step 3Cue the operator
    Mission feed C / RFSimulated logic
    Cooperative radio-frequency mappingThree configured receiver nodes combine bearing evidence into a location estimate with visible uncertainty and provenance.NODE ANODE BNODE CLOCATION ESTIMATEUNCERTAINTY RETAINEDCOOPERATIVE RF EVIDENCESOURCE · BEARING · TIME · CONFIDENCECAPTURECLASSIFYCORRELATEPUBLISH

    Turn spectrum into evidence

    Arc-RF can combine classified spectrum observations and calibrated bearing inputs from multiple nodes into a location estimate that retains source, confidence, and uncertainty.

    1. Step 1Capture and classify
    2. Step 2Combine bearings
    3. Step 3Publish uncertainty
    Mission feed D / CollaborativeSimulated logic
    Collaborative multi-vehicle task distributionThree autonomous aircraft divide a search area, share a mission-relevant observation, and reassign work while preserving local safety authority.SECTOR ALPHASECTOR BRAVOSECTOR CHARLIEUAS-1UAS-2UAS-3SHARED OBSERVATIONTASKING UPDATEDSHARED MISSION STATEDISTRIBUTED TASKS · LOCAL SAFETYPARTITIONOBSERVESHAREREASSIGN

    Distribute the mission

    Program-qualified collaborative behaviors partition work, share observations, and reassign bounded tasks while each vehicle keeps local safety and recovery authority.

    1. Step 1Partition the area
    2. Step 2Share local findings
    3. Step 3Reassign bounded work

    Representative mission logic — aircraft, sensors, models, communications, authority, and performance are configured and validated for each program.

    Arc-UI / operator layer

    One interface.
    Every mission edge.

    Plan routes, direct vehicles, inspect RF tracks, fuse detections, and intervene from one coherent operating picture. Arc-UI is the mission-native alternative to fragmented QGroundControl and Mission Planner workflows.

    Explore Arc-UI
    Arc-UI // Operator consoleLive mission state
    Arc operator console showing live map, detections, telemetry, and flight controls
    Voice + touch / Android + webMap · split · streamOptional TAK / CoT output
    EO node / airborneTargets classified
    Airborne ISR view with edge-classified vehicles
    Edge-to-operator workflow

    The platform reasons before it reports.

    On-device models turn video, RF, acoustic, LiDAR, and telemetry into concise, confidence-aware events before transmission—preserving scarce bandwidth and keeping the operator focused on decisions.

    Sense → perceive → decide → act → report
    Core competencies / shared stack

    The capability
    beneath every system.

    A sovereign neural runtime, modular hardware, collaborative mission services, and one assumption: the mission must continue when any single network path does not.

    01

    Synthetic Vision Generation

    Controlled scene and asset generation expands mission-approved training sets across viewpoint, lighting, weather, clutter, and occlusion—then feeds those cases into the same evaluation workflow as field evidence.

    02

    Synthetic SLM Pipelines

    Program-reviewed synthetic examples can supplement scarce task data for narrow command-routing, planning, and tool-use models without exposing sensitive source material to a public service.

    03

    Controlled Model Improvement

    Review evidence, analyze failure modes, build a candidate, evaluate it against mission thresholds, and promote only an approved artifact. Arc-Forge keeps the improvement loop close and accountable.

    04

    Mission-Tuned SLMs

    Small language models tuned for narrow mission tasks interpret objectives, constraints, and tool schemas locally—supporting planning, operator interaction, and policy-bounded action.

    05

    Full Communications PACE

    Arc separates the mission from the bearer. Program-selected radios, mesh, LTE/5G, and SATCOM can be integrated as primary, alternate, contingency, and emergency paths for the target platform.

    06

    Collaborative Autonomy

    Vehicles and sensor nodes can share bounded tasking, mission state, tracks, and conclusions across the configured network. Swarm behavior is integrated and qualified for each program and safety envelope.

    07

    Rapid Carrier Board Iteration

    A modular carrier architecture keeps compute, acceleration, sensor, vehicle-bus, and radio upgrades from forcing a complete platform redesign.

    08

    Multi-Modal Sensor Fusion

    EO/IR, LiDAR, RF, acoustic, telemetry, and TAK observations can be normalized into a confidence-aware local world model with source provenance preserved.

    09

    Agentic Tool-Calling

    On-device agents can use approved tools to query mission state, propose or issue bounded tasking, publish CoT events, and chain controlled workflows at the edge.

    10

    Disconnected by Design

    Local model, data, and inference workflows support disconnected deployments. External access, release controls, and approval boundaries are configured with the program.

    11

    Mission-Critical Work Local

    Perception, reasoning, voice, fusion, and mission actions run on the platform or local mission network. DDIL is the design assumption; connectivity expands the mission without owning it.

    Data rights / governance

    Sovereign by design.
    Open by choice.

    Arc builds on open-source foundations, including LlamaFarm and Atmosphere. Programs can deploy core workflows inside customer-controlled infrastructure, define data boundaries, and negotiate model, sensor-data, and mission-log rights in the acquisition terms. Open interfaces preserve an exit path without promising that every mission dependency is open source.

    LlamaFarm on GitHubDocumentation
    ■Disconnected Training Option■Open-Source Foundations■Program-Defined Data Rights■Modular Acquisition Path
    Program office / contact

    Put Arc
    on the mission.

    Defense inquiries

    [email protected]

    Partnerships & commercial

    [email protected]

    Powered by Rownd