XCUGAS / CONNECTED HARDWARE SYSTEMS Talk to our founder
Let's build together

01 / Indoor surveillance

Autonomous surveillance robot

Every route. A clearer picture.

A mobile observation platform for mapped indoor spaces. Connect a robot chassis, site-aware navigation and an operator workflow to turn patrol rounds into reviewable evidence.

Technical reference · Version 1.0 · 1 November 2026

Illustrative robot monitoring a data-center corridor with an operator alert display
Illustrative application · Configuration is project-specific
01 / PurposeIndoor surveillance
02 / ApproachObserve → review → respond
03 / Design statusConfigurable reference architecture

01 / THE PURPOSE

A useful view.
A reviewable record.

Fixed cameras provide a consistent view of selected locations. A mobile robot adds a changing viewpoint along a planned route. This reference design focuses on repeatable indoor inspection, recording where an observation occurred and routing it to an operator. It does not determine a person's identity or grant access to a controlled area.

Repeatable rounds

Visit agreed checkpoints and record missed locations rather than treating an incomplete route as a successful patrol.

Context for operators

Pair an observation with the mission, map location and capture time so a team can decide what to do next.

A platform to build on

Choose the body, cameras and application workflow around the site, using the chassis and software interfaces as a foundation.

Where to explore the platform

Data-center corridors

Observe agreed equipment-room approaches and route potential restricted-zone events for review.

Managed facilities

Check accessible corridors, doors and selected assets during scheduled rounds.

Indoor campuses

Repeat patrols through mapped, controlled spaces where pedestrian interaction can be assessed.

02 / FROM PLAN TO ACTION

An indoor patrol, explained.

Select a step or play the sequence. Change the scenario to explore a different outcome. This is an explanatory simulation, not a connected device.

Facility patrol / schematicIllustrative

Normal mission · Readiness checks assumed satisfied.

Facility patrol / schematicIllustrative route from the base through inspection checkpoints. The marker moves as you select mission steps; the text panel explains the current state.EquipmentroomRestrictedzoneService corridorBase / dockReview point!R
Planned routeRoute travelledReview point

STEP 01 / 06

Define the patrol

An operator selects the approved map, checkpoints and observation rules. The mission receives an identifier so every later event can be traced to its route.

System stateMission queued

A candidate event requires operator review. Mission success and alert acknowledgement are separate outcomes.

03 / SYSTEM ARCHITECTURE

Different layers.
Clear responsibilities.

Separate sensing, mission logic, local control and operator review. The diagram is a proposed allocation of responsibilities; component selection and interfaces are agreed during engineering.

01

Sense

LiDAR, camera, wheel feedback

Time-stamped measurements describe nearby geometry, visual context and motion. Sensor placement and calibration must be verified on the assembled robot.

02

Estimate & plan

Onboard computer / ROS 2 candidate

Maintain a map-relative pose and calculate an allowed path. A local controller translates the path into bounded motion requests.

03

Move & protect

Drive controller + local stop chain

Motor feedback closes the motion loop. A separately assessed stop path must remain available without a cloud connection.

04

Observe & review

Event service + operator panel

Observations become reviewable events. Operators receive health, mission state and evidence links through authenticated interfaces.

Local control boundary

Motion or flight responses must have a defined local policy. A web dashboard, cloud connection or language model is not a substitute for the platform's assessed control and stop / recovery systems.

01

Mapping is not the same as localization

SLAM builds or updates a map while estimating motion. A routine patrol may instead localize against an approved map. Door changes, moved equipment and reflective surfaces can reduce confidence; commissioning must define when to stop and relocalize.

02

Global route, local decisions

A candidate ROS 2/Nav2 integration separates route planning from the controller that follows it. Define the robot footprint, clearance and permitted zones using the actual body and payload. A camera-based alert should not override motion constraints.

03

Observation needs context

An event rule should consider zone, time and repeated observations. Keep evidence and operator decisions together. Tune thresholds against representative lighting and traffic, and record both false alerts and missed examples.

Technology background: Nav2: navigation servers · ROS 2: quality of service. These sources explain components, not Xcugas product certification or integration status.

04 / HARDWARE & CONFIGURATION

Specify the system
around the site.

Each layer has a job and an acceptance question. The final bill of materials, operating envelope and test evidence belong to the agreed project configuration.

SubsystemCandidate configurationWhat to verify
Mobile baseDrive motors, encoders and chassisFootprint, turning space, floor transitions, braking and payload stability.
Range sensing2D LiDAR and optional depth sensingCoverage near glass, dark surfaces, overhangs and low obstacles.
ObservationRGB camera; other payloads by scopeField of view, low-light quality, privacy masks and camera placement.
Compute & controlOnboard computer and drive controllerProcessing load, thermal limits, watchdogs and sensor time alignment.
Power & dockingBattery management and charging interfaceUsable mission energy, reserve policy, contact alignment and charging verification.
ConnectivitySite Wi-Fi or scoped network linkCoverage gaps, roaming behavior, buffering and reconnection handling.

No fixed speed, runtime, range, payload, ingress rating or accuracy figure is asserted in this reference design. Request a configuration-specific specification and supporting test results.

05 / SOFTWARE & DATA

Connect observations
to decisions.

A useful integration preserves data meaning: where the observation happened, when it was captured, which mission it belongs to and whether an operator has reviewed it.

InterfaceInformation exchangedIntegration requirement
Sensor → autonomyTime-stamped scans, images and odometryDefine frames, units, update expectations and stale-data behavior.
Mission → motionApproved waypoints and navigation goalsRequire preconditions, cancellation handling and state feedback.
Robot → dashboardHealth, mission state and observationsInclude capture time, receive time and a visible freshness indicator.
Dashboard → robotAuthorized mission / pause / resume requestsValidate ownership, expiry and acknowledgement before changing state.

One event, an explicit lifecycle.

  1. Captured locally
  2. Queued for transfer
  3. Received by the service
  4. Acknowledged by an operator
  5. Resolved with a recorded outcome

Retries should preserve the event identifier to avoid duplicate cases. A server receipt does not mean a human has acknowledged the event. Offline observations retain their original capture time.

ILLUSTRATIVE EVENT CONTRACT

{
  "event_id": "robot-example-001",
  "mission_id": "patrol-example",
  "platform": "robot",
  "captured_at": "2026-11-01T10:00:00Z",
  "location": {
    "frame": "site_map",
    "checkpoint": "review-zone"
  },
  "type": "observation_for_review",
  "evidence_ref": "access-controlled-reference",
  "delivery_state": "queued",
  "review_state": "unreviewed"
}

Example fields only. This is not a published API or live event.

Data handling is part of the design.

Access & authorization

Define operator roles, device identity and authenticated connections. Privileged commands need an audit trail and a clear authority model.

Retention & visibility

Agree which evidence is collected, who can see it and how long it is retained. Use restricted capture zones and masking where the site requires them.

Freshness & recovery

Mark stale telemetry visibly. Test reconnection, event de-duplication and expired-command rejection. Do not expose robot control interfaces directly to the public internet.

06 / OPERATING BOUNDARIES

Explain the exceptions.
Before they happen.

Normal operation is only part of the design. Agree what happens when the environment, position estimate, energy or communications no longer support the mission.

ConditionProposed responseAcceptance evidence
Obstacle blocks the routeStop, attempt only an allowed replan, then pause for assistance.No contact; an interrupted mission remains visibly incomplete.
Position confidence dropsSuspend motion and require a validated recovery or operator intervention.The robot does not continue along a pose estimate known to be unreliable.
Network connection dropsApply the configured local pause policy and buffer events with original timestamps.Dashboard shows stale data; queued and delivered events remain distinct.
Energy reaches reserveEnd the patrol and attempt the validated return procedure if feasible.Reserve is tested under the configured load; charging is confirmed separately.

Changing indoor geometry

A map is not a guarantee that a corridor remains clear. Reassess routes after layout changes.

Steps and inaccessible areas

Stairs, lifts and automatic doors require separate integration and validation; do not assume they are supported.

Shared spaces

Pedestrian behavior, visibility and stopping distance must be evaluated for the specific installation.

Unspecified capacity

Runtime, speed, payload, dimensions, detection accuracy and charging time remain configuration-specific.

07 / COMMISSIONING & VALIDATION

From a reference design
to a tested installation.

Set acceptance thresholds before the pilot. Measure the agreed configuration under representative conditions and document the cases where it needs help.

  1. 01

    Scope the site

    Document the task, operating area, environment, users and intervention responsibilities.

  2. 02

    Configure & integrate

    Select the hardware and software, calibrate sensors and define interfaces and exception policies.

  3. 03

    Run a controlled pilot

    Test normal missions, false events, interruptions and recovery. Retain logs and operator observations.

  4. 04

    Accept & maintain

    Sign off measured results, train operators and schedule checks. Revalidate after material changes.

MeasureHow to evaluate itTest conditions
Route completionCompleted checkpoints / scheduled checkpointsInclude blocked and interrupted runs in the report.
Obstacle responseStop clearance and intervention countTest representative people, objects and floor conditions.
Observation qualityPrecision and recall on a labelled test setInclude day/night scenes and difficult negative examples.
Alert lifecycleCapture-to-receipt and receipt-to-acknowledgement timeReport tail latency and connectivity failures, not just averages.
Return & chargeSuccessful docking and verified charge sessions / attemptsInclude low-energy and alignment-error cases.

Define pass / fail before testing.

For each measure, record the target, sample size, conditions, software version, hardware configuration and result. Targets are agreed with the customer; this document contains no benchmark results or promised performance levels.

What to bring to a scoping discussion

Provide floor plans, permitted routes, clearance constraints, observation examples, privacy zones, network coverage and a charging location. Define who reviews events and who can recover a paused robot.

08 / TECHNICAL WHITE PAPER

Take the design
into your next discussion.

Download the complete reference: purpose, mission flow, architecture, interfaces, operating boundaries, test plan and sources.

Version 1.0 · 1 November 2026 · PDF
Download white paper ↓Read PDF in a new tab ↗
Technical sources & document scope

The architecture is a proposed engineering design, informed by the existing Xcugas platform descriptions and the technical sources below. External documentation describes component behavior; it does not establish a deployed integration or a performance guarantee.

  1. Nav2: navigation servers ↗

    Explains the separation of planning, control and recovery in a ROS-based navigation system.

  2. ROS 2: quality of service ↗

    Describes message delivery policies and publisher/subscriber compatibility.

Source review: 1 November 2026. Links may describe evolving software; pin and validate versions during implementation.

A LITTLE MORE CLARITY

Questions worth asking.

Does it replace the security team?

This design supports observation and operator review. Response, escalation and access decisions stay with the designated team.

Is ROS 2 already a confirmed product specification?

ROS 2 and Nav2 are candidate components in this reference architecture. The delivered stack, supported versions and interfaces must be confirmed in the project scope.

Can it work without a network connection?

Local behavior must be defined and tested. The illustrated policy pauses the patrol and buffers events; it does not promise uninterrupted remote monitoring.

Start something extraordinary

Your challenge.
Our next build.

Tell us what you want to make possible. Let's find the right hardware, connectivity and path forward.

hello@xcugas.com