Automation & Robotics

Automation is about orchestration¶

A robot that sorts packages is an actuator. A conveyor belt is transport. A PLC is a controller. The value isn’t any of them individually — the value is the system that knows where each package belongs, coordinates dozens of robots and conveyors simultaneously, reacts to exceptions in real-time, and reports performance to ERP.

We build the orchestration layer between the physical world (OT) and business systems (IT). PLCs control machines. Our software controls the process.

Sorting lines¶

Control architecture¶

A sorting line is a real-time system. A package on the belt waits 200ms — if software doesn’t decide where to go, the package goes to fallback output.

Decision engine: 1. Camera/scanner reads identifier (barcode, QR, OCR) 2. Lookup in local cache (WMS data, routing rules) 3. Decision: output 1-N based on destination, priority, dimensions 4. Signal to PLC: “activate diverter at position X in Y milliseconds”

Entire flow under 50ms. Local processing, no dependency on cloud or WAN.

PLC integration¶

OPC-UA (Unified Architecture): Modern standard for industrial communication. Client-server and pub/sub model. Security built-in (encryption, authentication). Structured data model — you read “motor 3 temperature” instead of “register 40001”.

Integration pattern:

WMS → [Routing Rules Engine] → [OPC-UA Client] → PLC → Diverter
       ↑                                          ↓
  [Dashboard] ← [Telemetry Collector] ← [OPC-UA Subscription]

PLC remains master for real-time control (safety, timing). Our layer is master for business logic (where things belong, prioritization, exceptions).

Monitoring and optimization¶

• Throughput: Packages per hour per output. Trend, shift comparison.
• Jam detection: Photocells + accelerometers. Automatic jam detection, operator alert, optional automatic recovery.
• Misroute rate: How many packages ended up at wrong output. Root cause: bad scan, wrong routing rule, timing issue.
• Utilization: Line capacity utilization. Peak vs. off-peak. Bottleneck identification.

AMR (Autonomous Mobile Robots)¶

Fleet coordination¶

One AMR robot is simple. Ten AMR robots in the same space is an orchestration challenge:

Traffic management: - Path planning: A or RRT algorithms for optimal route - Collision avoidance: Dynamic replanning when two robots meet - Intersection management: Virtual traffic lights at intersections — one robot passes, others wait - Deadlock prevention:* Detection and resolution of circular blocks

Task assignment: - WMS generates tasks: pick item X from location A, deliver to location B - Fleet manager assigns task to optimal robot (closest, most battery, right type) - Multi-task optimization: robot picks up multiple items on route - Priority queue: urgent orders bypass standard ones

WMS integration¶

AMR fleet is execution layer for WMS:

• Inbound: Robot brings pallet from receiving to storage location
• Replenishment: Robot replenishes pick location from storage
• Picking: Robot brings shelves to operator (goods-to-person)
• Outbound: Robot takes completed order to shipping

Real-time visibility: WMS sees position of each robot, status of each task, ETA completion.

Fleet management¶

• Charging orchestration: Battery level prediction, automatic return to charger before depletion. Optimization — not all robots charge simultaneously.
• Maintenance scheduling: Preventive maintenance based on operational hours, error count, sensor health.
• Performance metrics: Tasks per hour, average transport time, utilization rate per robot. Underperformer identification.

IT/OT convergence¶

Why it’s difficult¶

IT and OT worlds have fundamentally different priorities:

	IT	OT
Priority	Confidentiality	Availability
Update cycle	Weeks	Years
Protocols	HTTP, SQL, AMQP	Modbus, PROFINET, OPC-UA
Downtime tolerance	Minutes	Seconds
Lifecycle	3-5 years	15-20 years

Connecting these worlds requires respect for both. You can’t force IT approaches (frequent updates, cloud-first) on OT systems, nor IT systems OT rigidity.

Integration architecture¶

DMZ between IT and OT networks:

[OT Network]                    [DMZ]                      [IT Network]
PLC, SCADA, HMI  →  [OPC-UA Gateway]  →  [Integration Layer]  →  ERP, WMS, BI
                     [Firewall]           [Kafka/MQ]             [Cloud]
                     Unidirectional       Protocol translation

Key principles: - Network segmentation: OT network isolated from IT and internet. Communication only through defined gateways. - Unidirectional where possible: Data from OT to IT (telemetry, status). Commands from IT to OT through explicit, audited channels. - Protocol translation: OPC-UA or custom adapters translate industrial protocols to IT standards (REST, Kafka, MQTT). - Audit trail: Every command from IT to OT logged. Who, when, what, why.

Legacy integration¶

Not every PLC speaks OPC-UA. Older systems require:

• Modbus TCP/RTU: Most widespread legacy protocol. Simple register-based read/write. Custom Modbus gateway for translation to modern API.
• PROFINET: Siemens ecosystem. Real-time Ethernet protocol.
• EtherNet/IP: Rockwell/Allen-Bradley ecosystem. CIP protocol over Ethernet.
• Proprietary: Serial port, custom binary protocols. Reverse engineering where documentation doesn’t exist.

Digital Twin¶

Virtual model of physical process¶

Digital twin isn’t 3D visualization (though it can include it). It’s a simulation model that reflects current state of physical system and enables what-if analysis:

• Current state: Real sensor data → live model. Operator sees entire line status in real-time.
• What-if simulation: What happens if we increase belt speed by 20%? Where’s the bottleneck? How many robots do we need for peak season?
• Predictive: Model predicts future behavior based on historical data and current trend.
• Training: New operator learns on virtual model. No risk of damaging actual hardware.

Implementation¶

• Data layer: Real-time telemetry from OT systems (OPC-UA, MQTT)
• Simulation engine: AnyLogic, SimPy, or custom discrete event simulation
• Visualization: 3D model (Unity, Unreal Engine) or 2D schematic view (Grafana, custom web app)
• Integration: Simulation results → capacity planning, maintenance scheduling, layout optimization

Technology stack¶

Communication: OPC-UA, Modbus TCP/RTU, PROFINET, EtherNet/IP, MQTT, Kafka.

AMR platforms: MiR, OTTO Motors, Locus Robotics, custom AGV integration.

Simulation: AnyLogic, SimPy, FlexSim, custom discrete event simulation.

Fleet management: ROS 2, custom fleet orchestration, VDA 5050 interface.

Monitoring: Grafana, InfluxDB, Prometheus, custom OEE dashboards.

Souvisí s