Overview
A global leader in semiconductor display manufacturing is driving digital transformation across its production lines by integrating AI and industrial IoT technologies. As robotic equipment became more widely used to boost efficiency, the company faced growing challenges in monitoring and managing these devices in real time. To address this, it set out to build a centralized diagnostics and alerting platform that could unify data from hundreds of robots, reduce downtime, and enable smarter, data-driven maintenance.
Challenges
Building a real-time diagnostics platform for factory-floor robotics came with significant technical hurdles:
- Low-latency performance: The system needed to handle high volumes of time-sensitive data in milliseconds, even under hardware limitations and constrained bandwidth typical of industrial environments.
- Device diversity: Robots from different vendors produced data in inconsistent formats. The platform had to standardize and integrate data across multiple protocols and interfaces.
- Massive data flow: With 24/7 operations, data volume reached hundreds of millions of messages per day. Ensuring data integrity, continuity, and security was critical.
- System resource limits: Due to legacy infrastructure, server CPU usage had strict limits, making efficient resource scheduling a key concern.
- Real-time data processing: Operators needed the ability to visualize robot performance by day, week, or month, and quickly diagnose anomalies based on historical trends and real-time metrics.
Solution
To meet these demands, the company built an end-to-end robotic management platform using EMQX Platform as its messaging backbone. Combined with Intel EII’s containerized edge computing framework, the solution collects, preprocesses, and routes robot data from controllers and industrial PCs to the backend systems with flexibility and scalability.
Seamless Data Collection
Using the self-developed data collection tool, ia-publisher, the company gathered detailed robotic metrics—including torque, motion pulses, Z-axis positions, vacuum time, speed, temperature, and positional offsets—via socket connections from central robot controllers.
Efficient Data Transmission
With over 500 robots generating millions of data points daily, EMQX served as the high-throughput MQTT broker at the core of the system. It ensured millisecond-level latency and millions of messages per second throughput, enabling uninterrupted, real-time data flow from robot edge devices to central servers.
Reliable Data Storage and Integration
Out-of-the-box integration with InfluxDB enabled reliable and scalable storage of time-series data, ensuring long-term retention, compliance, and fast querying.
Intuitive Data Visualization
Grafana dashboards provided visual insights into robot performance across multiple timeframes (daily, weekly, monthly), helping operators monitor operations at a glance.
Smart Alerting and Predictive Maintenance
Using Kapacitor, the system defined alert thresholds to trigger email notifications for abnormal data points. This helped reduce response time, minimize losses, and improve operational efficiency. Leveraging the real-time data streams, it powered AI-based predictive models to forecast future failures—enabling a shift from reactive to preventive maintenance.
Results
Since launch, the robotic diagnostics platform has delivered measurable success:
- 500+ robots connected and managed in a single plant.
- 10+ million messages processed daily, with low latency and high reliability.
- Over ¥5 million saved in reduced downtime and repair costs.
- Operational efficiency improved through visualized dashboards and automated alerts.
- Anomaly root cause analysis and predictive maintenance capabilities enhanced, reducing the risk of unexpected failures.
Most importantly, the platform helped the company standardize robot health monitoring and empowered operators to take proactive action before faults escalated.
Looking Ahead
EMQX Platform enabled the company to overcome key Industrial IoT challenges, shifting equipment management from reactive to proactive. The solution with end-to-end capabilities from data collection to analysis has become a model for smart manufacturing in the semiconductor display industry.
With EMQX as the foundation of its data pipeline, the company plans to expand this solution to other areas such as warehouse automation and environmental monitoring. As AI models evolve, the platform’s predictive capabilities will continue to grow—unlocking new value from industrial data and accelerating the journey toward fully intelligent manufacturing.