All packs

Defense & Critical Sites

Autonomous Perimeter Sweep

Persistent ground + air perimeter coverage. Units sweep, detect, and escalate intrusions in seconds — not minutes.

Guard forces can't watch every fence line 24/7. Static cameras flood operators with nuisance alerts until the one event that matters gets missed. A breach detected 10 minutes late is already a breach. Centertron runs a coordinated swarm against a live World Model of the site.

Deploy via outcome brief
78 teams run this pack 14,040 runs / 30 days 90% auto-resolution

Expected outcome

24/7 autonomous coverage. >90% of nuisance alerts auto-dismissed; real intrusions escalated to a human in <15 sec with live video + GPS fix.

Runs as a continuous outcome

A standing coverage SLA — gaps in autonomous perimeter coverage trip an Autopilot signal in real time and roll up into portfolio safety health.

SLAAutonomous perimeter coverage >= 99 %/ realtime

Install this pack, then graduate it into a continuous outcome from My Workflows — it earns autonomy (shadow → assisted → autonomous) as it proves itself on real telemetry.

Hours saved / wk

80

FTEs freed

2

Monthly value

$26,000

Triggers

Scheduled patrol route (cron / shift)
ready
Fence-line / seismic sensor fires
ready
Fixed-camera motion / object detection
ready
Operator manual dispatch
ready

Integrations

Video Management System (Genetec, Milestone)

Fixed-camera ingest + live escalation overlay

Access control / PSIM

Correlate breaches with door/gate events

Radio / dispatch console

Push alerts to responding officers

Slack / Teams (optional)

Command-channel notifications

Agent steps

3 credits / run
  1. 1

    Perceive

    Fuses ground rover + UAV feeds with fixed sensors into one World Model of the perimeter.

  2. 2

    Navigate

    Plans and flies/drives the sweep route, re-tasking units around obstacles in real time.

  3. 3

    Detect

    Flags humans, vehicles, and breaches against the site's known-actor baseline.

  4. 4

    Classify

    Friend / nuisance (wildlife, weather) / unknown intruder — with confidence + position.

  5. 5

    Decide

    Auto-dismiss nuisance, shadow-track unknowns, or escalate per the site's rules of engagement.

  6. 6

    Escalate & log

    Pushes live video + map pin to the guard desk and writes the event to the Decision Log.

See it in action

State Transformation

A formal contract: what this pack converts, with explicit uncertainty bounds and verification criteria.

Initial state

Unmonitored perimeter with unknown actor presence and stale sensor coverage

Target state

Continuously verified perimeter where every actor is detected, classified, and dispositioned (dismissed | tracked | escalated)

Invariants (must hold)

site geofence loadedevery unit position knownrules-of-engagement applied

Uncertainty bounds (max acceptable error)

max_missed_intrusion≤ 1.0%
max_false_escalation≤ 10.0%

Verification: Sweep coverage ≥ target %, every detection logged, escalations acknowledged by guard desk

Domain ontology — entities & states

The vocabulary this pack reasons over. Each entity has its own state machine.

Patrol5 states
queued→in_progress→rerouted→complete→aborted
Detection5 states
raised→classified→tracking→dismissed→escalated
Unit5 states
ready→deployed→returning→charging→offline
Zone3 states
clear→watch→breach

Relations

  • · Patrol-covers-Zone
  • · Detection-raised-in-Zone
  • · Unit-executes-Patrol

Audit & explainability

  • Full sweep replay: route, detections, and decisions on one timeline
  • Chain-of-custody export for every escalation (video + reasoning)
  • Nuisance-vs-real ratio trend to tune detection thresholds

AI assurance — Responsible AI + T&E

ResponsibleEquitableTraceableReliableGovernable
Autonomy gate: human-on-the-loop Model card: published

Test & evaluation record per run mode — shadow, assisted, supervised-auto — earned before autonomy is unlocked.

Aligned to DoD Responsible AI principles; every autonomous decision is explainable and reversible, with a human owning authority on mission-critical actions.

Contested operations — DDIL ready

  • Runs at the edge through Denied, Degraded, Intermittent, and Limited-comms (DDIL) conditions — decisions continue locally when the link drops.
  • On reconnect, the local decision log reconciles into the IG-ready ledger with conflict resolution — no lost accountability.

Pricing

~7× ROI — you keep ~85% of the value this pack creates. Priced at the market median, never as a race to the bottom.

Pilot

Prove the ROI on one bottleneck.

$3,900/mo

Onboarding · included free

  • 1 workflow deployment
  • Shadow + assisted run modes
  • White-glove setup call
  • Decision Log + audit export

Scale

Recommended

Run it as a continuous outcome.

$9,360/mo

Onboarding · included free

  • Up to 5 deployments
  • Autonomous mode (earned via self-grading)
  • Continuous SLA in the Outcome Console
  • Autopilot drift signals

Enterprise

Your operating model, contracted.

Custom

+ custom implementation

  • Unlimited deployments
  • Custom SLAs + integrations
  • Dedicated solutions engineer
  • SSO + compliance reviews
Talk to sales

Calculate my ROI

Drag the sliders. We'll email you a personalized report — no fluff, your numbers.

10 people
50

Hours/wk saved

80

FTEs freed

2

Monthly value

$26,000

$312,000/yr

We'll never sell your email. One report, no drip campaign.

Estimates use the same scaling formula sales reviews on the call. No marketing math.

Onboarding is included free — self-serve or white-glove

Pick Self-serve and our onboarding assistant walks you through connecting Gmail/Outlook/Slack/ERP, or choose White-glove and a Centertron specialist joins you for a live go-live call — both at no extra cost, live in <48 hours. Either way, the pack runs in your audit trail with every decision logged.

Talk to sales