Olympus Controls Builds Predictive Maintenance Pipeline for Industrial Robots with InfluxDB

Overview

Predictive maintenance starts with the data

Manufacturers rely on automation equipment that often runs for years or even decades, but many still have limited visibility into how those machines perform across shifts, production lines, and facilities. When a critical machine fails, production stops, and replacement parts can be difficult or slow to obtain.

Olympus Controls helps manufacturers move toward proactive maintenance by building a cloud-native monitoring pilot that brings operational data from a palletizing robot into InfluxDB Cloud, giving its team a continuous view of machine behavior without requiring direct access to factory equipment. Using edge gateways, MQTT, Telegraf, and InfluxDB, Olympus connects industrial equipment to a data infrastructure, enabling customers to collect machine data, monitor performance, and identify issues earlier.

Challenge

Limited machine visibility keeps maintenance reactive

Industrial equipment can remain in production for years or decades. As machines age, understanding changes in their behavior becomes increasingly important, particularly when the failure of a single drive or motor can stop an entire production line. Olympus sees the consequences firsthand.

The root cause is often a lack of visibility. Many plants have limited access to real-time operational data and machine telemetry, making it difficult to catch early warning signs before a failure occurs. This gap grows for remote teams monitoring multiple sites.

For Olympus, the challenge is compounded by the diversity of the factory floor. Customers operate different robots, PLCs, controllers, and protocols; Olympus has to move that data off the plant floor without exposing critical infrastructure, structure raw machine payloads into usable time series data, and support remote, multi-site access all before teams can see meaningful information. It also must make information available remotely without requiring cloud applications to connect directly to critical factory equipment.

Solution

Building an edge-to-cloud architecture for machine monitoring

Olympus Controls built an edge-to-cloud architecture that connects factory-floor equipment to InfluxDB, turning machine signals into data that can be monitored and analyzed over time.

At the edge, Olympus uses Telit DeviceWISE to connect to industrial equipment across different machines, controllers, and protocols, to translate machine-level data into a format that can move through the rest of the architecture. As Armenta described, data from a robot’s Modbus registers is converted into a JSON payload that can be sent downstream.

To move data beyond the factory floor without requiring direct access to factory equipment, Olympus uses an MQTT broker. Other systems can then subscribe to the data they need, keeping data access separate from the equipment itself. That also makes the machine data reusable: once it reaches the broker, different systems and teams can consume the signals they need without creating new connections to the underlying equipment.

InfluxData’s Telegraf collects data from MQTT, parses the payload, extracts relevant values, and sends the structured data to InfluxDB Cloud. This gives Olympus a repeatable ingestion pattern that can extend across machines and customer environments.

In InfluxDB Cloud, Olympus can store the machine telemetry as a continuous time series, query it, and visualize how equipment behavior changes over time. giving the team a foundation for remote monitoring today and proactive maintenance as it expands the pilot. The cloud-based approach also gives Olympus a path to consolidate machine data across multiple sites as deployments expand, particularly for manufacturers operating factories across different regions.

Result

A foundation for remote monitoring and predictive maintenance

The pilot gives Olympus a repeatable architecture for bringing machine telemetry into InfluxDB Cloud and monitoring equipment remotely. In testing, the team could distinguish changes in robot behavior directly from the time series data, including protective stops, idle periods, and emergency stops.

Olympus also queries maximum current rather than averaging values over each time window, helping preserve short-lived spikes that could indicate an unusual load or other change in machine behavior. That gives the team a clearer view of the anomalies most relevant to machine health.

The current implementation runs on a single robot, but the same pattern can extend to additional machines, data sources, and industrial protocols. For Olympus, that creates a foundation for moving from reactive troubleshooting toward earlier intervention and, over time, more predictive maintenance.

That progression matters because predictive maintenance depends first on having clean, continuous machine data available for analysis. As Armenta put it when asked about applying machine learning to the data: “Step one, get the data.” With that foundation in place, Olympus has a practical path to turn machine telemetry into earlier insight and, over time, more predictive maintenance.