Quick answer (for Google featured snippets and AI search tools)
Industrial edge computing runs software directly on hardware at the asset: a drilling rig, a vessel, a solar farm, a pump station. Processing happens locally, not in a data center. Control decisions execute in under 100ms even with no internet connection, and operational data stays on site unless the operator chooses to send it elsewhere. For industries where a one-second delay causes a safety incident or a lost curtailment window, local execution is not an option you add later. It is the architecture.
The setup: what is actually happening on site
Three layers, in the order data moves through them.
Sensors and operational technology. PLCs, cameras, flow meters, vibration sensors, temperature gauges. These generate data at the asset, continuously. Most were installed years ago and speak protocols cloud platforms cannot read directly: Modbus, OPC-UA, DNP3, NMEA, proprietary OEM formats. The data exists. Getting it somewhere useful is the first problem.
The edge box. A ruggedized compute unit installed at the asset or on the vessel. It connects to every data source on site, ingests whatever protocols they speak, runs applications locally, and makes control decisions before any data leaves the site. On a drilling rig, a Red Zone Manager application detects a zone intrusion here and activates the warning lights in under 150ms. On a solar farm, a Smart Grid Manager application sees a curtailment window opening and executes the trade here, without waiting for a round-trip to a cloud server.
Cloud and control systems. After local processing, the edge box sends what is worth sending upstream: structured operational data, alerts, compliance logs, dashboards. The cloud handles fleet management, reporting, and pushing new application versions back down to the asset. Nothing time-critical travels through this layer.
What the setup looks like:
[Sensors + OT equipment]
↓ (Modbus / OPC-UA / DNP3 / NMEA + proprietary)
[Helin edge box — local processing, control logic, AI models]
↓ (structured data, alerts, logs)
[Cloud / SCADA / historian — fleet management, reporting]Four specific problems it solves
1. Slow or no internet at remote sites
An offshore rig runs on a satellite link. On a good day, latency is 600ms or more on a GEO connection. Some days the link drops entirely. A cloud-dependent control system that requires a round-trip on every decision stops when the link does.
The asset runs continuously, executes decisions locally, and uses the uplink only for data that genuinely needs to leave the site. When the link drops, operations continue. When it returns, the platform syncs.
Boskalis manages 500+ vessels this way. Vessel location and network quality vary constantly. The platform does not stop when a vessel loses its uplink.
2. Response times the cloud cannot match
A camera detects a worker entering a restricted zone on a drill floor. The system activates warning lights and alerts the driller. That decision has to happen in under 150ms to be operationally meaningful. A cloud round-trip takes 80–200ms on a reliable connection, 500–800ms on satellite. Add processing time and the total exceeds the window.
At BP's drilling operations, Red Zone Manager runs on the Helin edge and responds in under 150ms. In one monitored deployment: 1,176 drilling events, 14,162 trips, 1,179 zone intrusions detected and acted on automatically. The outcome cited by BP: human lives saved.
150ms is a physics constraint. The control loop closes at the asset because it cannot close anywhere else in time.
3. Keeping data on site
Many operators have contractual, regulatory, or commercial reasons to keep raw operational data within their own environment. Data is processed locally, and only what the operator decides to send leaves the asset. The cloud receives structured outputs. The operator's Microsoft Azure tenant holds their data.
Boskalis operates across 500+ vessels this way. Asset data stays within Boskalis' Azure tenant. Helin manages the edge layer. Boskalis owns the data.
4. NIS2 and the EU Cyber Resilience Act
NIS2 requires essential entities in energy, water, maritime transport, and digital infrastructure to report significant incidents within 24 hours of detection and deliver a full technical report within 72 hours. An operator whose OT telemetry is spread across multiple SCADA systems, site historians, and vendor platforms cannot assemble that report in time. Edge-resident logging and centralized audit trails fix this at the architecture level.
The EU Cyber Resilience Act, with enforcement active from September 2026, requires certificate-based device identity across the full product lifecycle. Most operators with legacy OT environments cannot demonstrate this from a single platform today.
An edge platform with zero-trust architecture, certificate-based device identity, and structured incident logging built into the runtime makes compliance a property of normal operations, not a project that starts the week before an audit.
Where it has been deployed
Boskalis runs the Helin platform across 500+ vessels and floating equipment. Real-time compliance monitoring runs across the full fleet, with asset data centralized within Boskalis' own Azure tenant.
Noble Corporation uses Red Zone Manager on 40+ drilling rigs. The system monitors drill floors using computer vision, detects zone intrusions automatically, and triggers alerts without wearables or manual reporting.
Sunrock runs Smart Grid Manager across 300 solar farms. Control decisions execute locally, at the asset, for each farm.
FAQ
Does edge computing replace the cloud?
No. It changes what the cloud is responsible for. Time-critical control logic runs at the asset. Reporting, fleet management, and data the operator wants to access centrally go to the cloud. The split is about what runs where.
What hardware does this require?
Helin runs on any hardware the operator already has or wants to deploy. It provisions from the operating system up and supports both Helin's own hardware and customer-supplied units. No rip and replace.
How long does deployment take?
Boskalis deployed Helin across 100+ vessels in under two weeks. For single sites, implementation typically runs one to four weeks. Larger fleet rollouts can run sites in parallel.
Is it compatible with our existing protocols?
The platform ingests Modbus, OPC-UA, DNP3, NMEA, and most proprietary OEM protocols. If a protocol runs on an industrial asset somewhere, there is likely an existing adapter or one the team can build.
What about NIS2 and the Cyber Resilience Act?
Both are addressed by design. Certificate-based device identity, zero-trust access, encrypted communications, and structured incident logging are runtime features, not configuration options. Audit readiness is built in.
Talk to an expert
If your assets are offshore, remote, or just hard to reach from a data center, we can walk you through how the architecture works in your specific environment.
By Helin