Skip to main content
Heex follows a Smart-Data philosophy: instead of recording everything and filtering later, we filter intelligently and capture only relevant data. This reduces storage by 70-90% while ensuring every captured event is meaningful. Start risk-free: Deploy Heex on a server to analyze your existing ROS bags and historical data—no changes to your robots, no risk of missing critical information. Once your triggers are refined and validated, deploy the same configuration to production robots for real-time capture.

​
High-Level Flow

Heex Data Flow Architecture

​
The Smart-Data Approach

Think of Heex like a smart security camera for robots:
Traditional ApproachHeex Smart-Data
Record 24/7 continuouslyRecord only when conditions are met
100GB+ per day per robot500MB-2GB per day per robot
Hard to find relevant momentsEvery recording is a tagged, searchable event
Review hours of footageJump directly to the incident

​
Three-Layer Architecture

​
1. Edge Layer (Agent Deployment)

The Heex Agent can be deployed in two environments with the same workflow:
DeploymentUse CaseData Source
Edge (Robot)Real-time monitoring on physical systemsLive sensor data, ROS topics
ServerHistorical data analysis and replayROS bags, recorded datasets
Why this matters: Switching to Smart-Data capture can feel risky—what if you miss something important? Heex eliminates this concern:
  1. Start on a server — Analyze your existing ROS bags and historical datasets with zero impact on operations
  2. Refine your triggers — Iterate until you’re confident you’re capturing the right events
  3. Deploy to production — Push the same validated configuration to your robots via OTA
No workflow changes required until you’re ready. No fear of being blind to critical data. The agent runs with minimal footprint:
  • < 6% CPU usage on Jetson Orin NX
  • < 1% RAM overhead
ComponentRole
DataSourceConnection to data streams (ROS topics, sensors, APIs, or ROS bags)
SignalIndividual data point from a DataSource (speed, temperature, GPS)
AgentOrchestrates monitoring and recording
MonitorWatches signals for trigger conditions
RecorderCaptures data when events occur
ROS 2 Integration: With RDA (Resource and Data Automation), Heex offers plug-and-play integration for ROS 2 systems—automatic discovery of topics and signals, no manual configuration required.

​
2. Cloud Layer (Heex Platform)

The cloud layer manages configuration and stores events:
ComponentRole
WorkspaceIsolated environment for your team/project
SystemRepresents a physical robot in the platform
TriggerDefines conditions and recording rules
EventGenerated when trigger conditions are met

​
Privacy by Design

Heex separates metadata from raw data to protect your operational information:
Data TypeStorage LocationContent
MetadataHeex CloudEvent timestamps, trigger info, system status, labels
RecordingsYour Cloud StorageRaw sensor data, ROS bags, video files
This architecture ensures:
  • Your sensitive operational data never leaves your infrastructure
  • Only lightweight metadata is transmitted to Heex for indexing and search
  • You maintain full control over your recordings storage and retention policies

​
Over-the-Air (OTA) Updates

Deploy configuration changes across your entire fleet in minutes:
  • Push new triggers, monitors, and recording rules remotely
  • Update all systems simultaneously or target specific groups
  • No physical access required—manage robots worldwide from a single dashboard
  • Changes propagate automatically when systems connect

​
3. Analysis Layer (Insights)

Access and analyze your Smart-Data:
ComponentRole
Event SearchFilter events by system, trigger, time, labels
RecordingsDownload or stream captured data
DashboardsVisualize trends and analytics
StudioReplay and analyze recordings

​
How It Works: A Practical Example

Imagine monitoring a warehouse robot for near-collision events: 
T+0s    Robot operating normally, speed = 5 km/h
T+10s   Robot accelerates, speed = 22 km/h
T+10s   Agent monitor detects: speed > 20 km/h ✓
T+10s   Agent monitor detects: obstacle_distance < 1m ✓
T+10.1s Both conditions met → Trigger fires
T+10.1s Event created with metadata (system, location, values)
T+10.2s Recorder captures: 5 seconds before + 10 seconds after
T+11s   Recording uploaded to cloud with event tags
T+12s   Notification sent to operations team
Result: Your team receives an alert with a 15-second recording showing exactly what happened, tagged and searchable.

​
Ways to Interact with Heex

Heex provides multiple entry points to fit your workflow:
InterfaceBest ForDescription
Web UIVisual managementFull-featured dashboard for configuration, monitoring, and analysis
CLIAutomation & scriptingCommand-line tool for CI/CD pipelines and programmatic access
Copilot (beta)Natural languageAI assistant to query events and configure triggers conversationally
All three interfaces access the same platform—choose based on your preference or combine them as needed.

​
Key Benefits

BenefitHow Heex Delivers
Risk-free adoptionStart on historical data, deploy to production when ready
Reduced storage costs70-90% less data through intelligent filtering
Faster incident responseEvents are tagged and searchable instantly
Remote fleet managementOTA updates across your entire fleet in minutes
Privacy by DesignMetadata on Heex, recordings stay on your infrastructure
Minimal edge impactLess than 6% CPU, less than 1% RAM on edge devices
Scalable deploymentSame triggers work across 1 or 1000 robots
Flexible accessWeb UI, CLI, or Copilot—work the way you prefer

​
Next Steps