Industrial Automation Solutions: Enabling Intelligence at the Edge

Industrial automation is undergoing one of the most significant transformations in its history. Driven by the convergence of robotics, artificial intelligence (AI), Internet of Things (IoT), and edge computing, modern industrial systems are evolving from centralized, rigid architectures into distributed, intelligent ecosystems capable of real-time decision-making.

Across manufacturing, logistics, energy, agriculture, and infrastructure, organizations are increasingly adopting edge-first architectures, where data is processed locally rather than being sent to the cloud. This shift is not only about performance; it is about enabling autonomy, reducing latency, improving reliability, and operating efficiently in environments where connectivity may be limited or intermittent.

At the heart of this transformation lies a new generation of embedded computing platforms: low-power, compact, industrial-grade System-on-Modules (SoMs) that integrate CPU performance, AI acceleration, connectivity, and sensor processing into highly optimized form factors.

By providing advanced embedded edge computing building blocks, SolidRun enables OEMs and solution developers to rapidly design and deploy intelligent industrial systems. These platforms empower customers to bring AI-driven capabilities directly to machines, devices, and field systems operating in harsh and resource-constrained environments.

This article presents several real-world industrial automation use cases demonstrating how edge computing enables smarter, faster, and more resilient industrial systems.

Integrated 3D Sensor Systems with Machine Learning

The Challenge: Vision-Driven Automation at the Edge

Modern Industry 4.0 applications increasingly rely on machine vision systems to guide robotics, ensure quality control, and enable autonomous decision-making. However, implementing advanced 3D vision systems introduces several challenges:

Traditional architectures often rely on cloud-based processing, which introduces unacceptable latency for real-time robotics applications. In addition, high-resolution image streams require significant bandwidth, increasing infrastructure cost and limiting scalability. Integrating depth sensing, image processing, and AI inference into a compact system further complicates hardware design, especially in industrial environments where space, power, and thermal constraints are strict.

To enable next-generation automation, industrial systems require compact, integrated platforms capable of processing visual data locally and in real time.

Evaluation Criteria

To meet the demands of modern machine vision systems, the solution must provide:

Real-time processing of 3D and depth sensor data
Low-latency AI inference at the edge
Compact hardware suitable for embedded robotics
Industrial-grade reliability under continuous operation
Support for multi-camera or advanced vision configurations

The Solution

A customized embedded vision system built on the i.MX8M Plus System on Module enables real-time machine learning inference directly on the device. The platform combines CPU performance with a dedicated neural processing unit (NPU), allowing efficient execution of computer vision workloads without reliance on external compute resources.

Integration Approach

SolidRun worked closely with the customer to develop a tailored hardware solution, including a custom carrier board designed specifically for industrial machine vision requirements. The design included:

Dual camera input support for stereoscopic or depth-based vision
Integrated storage and power management system
Optimized board layout for compact robotic systems
Edge AI acceleration using the onboard NPU

This tight hardware-software integration allowed the customer to deploy a fully autonomous vision system capable of operating directly within robotic platforms.

Outcomes and Benefits

The resulting solution delivered measurable improvements:

Significant reduction in processing latency due to local inference
Increased accuracy in object detection and spatial awareness
Reduced reliance on cloud infrastructure and network connectivity
Smaller system footprint, enabling integration into compact robotic devices
Lower total system cost through hardware consolidation

By bringing intelligence directly to the edge, the customer achieved faster decision-making and improved operational efficiency in automated industrial environments.

Next-Generation Autonomous Harvesting Robotics

The Challenge: Precision Agriculture in Complex Environments

Agriculture is undergoing a technological revolution driven by automation and AI. While drones have been widely used for monitoring crop health, soil conditions, and spraying applications, the next frontier is fully autonomous harvesting.

However, agricultural environments present unique challenges. Crops must be handled with precision to avoid damage, and harvesting decisions must be made in real time based on environmental conditions, plant maturity, and spatial positioning. These tasks require a combination of computer vision, AI inference, and motion control operating in highly dynamic outdoor environments.

Evaluation Criteria

An effective solution must support:

Autonomous decision-making using edge AI
Low-power operation suitable for airborne robotics
Real-time image processing under variable lighting conditions
Flexible hardware for custom robotics integration
High reliability in outdoor agricultural environments

The Solution

An autonomous agricultural robotics platform was developed using the i.MX8M Mini System on Module, providing an optimized balance between performance and energy efficiency.
The system enables onboard processing for navigation, object detection, and harvesting decision-making without relying on external connectivity.

Integration Approach

The customer designed a custom carrier board based on SolidRun reference schematics, enabling:

Integration of flight control systems and AI processing on a unified platform
Edge-based computer vision for crop detection and classification
Autonomous control loops for harvesting actions
Efficient power management for extended drone operation

Outcomes and Benefits

The implementation resulted in:

Increased harvesting precision with reduced crop damage
Improved autonomy in complex orchard environments
Reduced operational dependency on human labor
Extended mission duration due to low-power architecture
Faster development cycles through reference design reuse

This use case demonstrates how compact embedded systems enable entirely new classes of autonomous industrial robotics.

AI-Driven Food Dispensing Systems

The Challenge: Consistency, Waste, and Limited Intelligence

Traditional food dispensing systems operate as static machines with minimal adaptability. This leads to inconsistent portion control, inefficient inventory usage, and high levels of food waste. Additionally, maintenance is typically reactive rather than predictive, resulting in unnecessary downtime.

Modern food service and vending applications require intelligent systems capable of adapting in real time to user behavior, environmental conditions, and supply availability.

Evaluation Criteria

The solution must provide:

Real-time AI inference at ultra-low power
Integrated vision and sensor processing
Compact and fanless system design
Predictive maintenance capabilities
High reliability for continuous operation

The Solution

A smart dispensing system built on the Renesas RZ/V2L System on Module enables efficient edge AI processing in cost-sensitive industrial environments.
The platform integrates a DRP-AI accelerator and image processing capabilities, allowing real-time decision-making without external hardware dependencies.

Integration Approach

Key system design elements include:

On-device image recognition for portion verification
Embedded AI models for quality control
Integrated processing eliminating external ISP requirements
Compact architecture suitable for kiosks and vending systems
Industrial temperature support for continuous operation

Outcomes and Benefits

The system delivered:

Improved consistency in portion control
Reduced food waste through intelligent dispensing
Lower operational costs due to integrated architecture
Predictive maintenance capabilities reducing downtime
Enhanced personalization for end users

This represents a shift from mechanical dispensing systems to intelligent, adaptive food automation platforms.

Next-Generation Industrial Gateways

The Challenge: Real-Time Connectivity and Edge Intelligence

Industrial networks are becoming increasingly complex, requiring gateways that not only connect devices but also process data locally. Traditional architectures struggle to meet the demands of modern industrial systems, especially those involving high-bandwidth sensors, machine vision, and real-time control loops.
Latency, security, and scalability are key concerns, particularly in large distributed environments such as energy grids, water systems, and manufacturing plants.

Evaluation Criteria

Industrial gateways must support:

Sub-10ms real-time processing capability
Multi-protocol industrial connectivity
Edge AI inference for local decision-making
High-bandwidth data handling (e.g., video streams)
Secure and resilient network architectures

The Solution

An advanced industrial gateway platform based on the i.MX8M Plus System on Module integrates computing, AI, and connectivity into a unified edge system.

Integration Approach

The platform includes:

Quad-core Cortex-A53 CPU for application processing
Integrated NPU for AI inference (2.3 TOPS)
Cortex-M7 real-time core for deterministic control
Dual Gigabit Ethernet with TSN support
CAN-FD and camera input support for industrial sensing
Hardware security engines for trusted operations

Outcomes and Benefits

The solution enables:

Reduced latency in industrial control systems
Improved throughput for data-intensive applications
Enhanced system reliability in harsh environments
Stronger cybersecurity and system integrity
Simplified system architecture through hardware integration

This gateway architecture represents the convergence point between connectivity, control, and intelligence at the industrial edge.

Predictive Maintenance: Monitoring Machine Wear

The Challenge: Preventing Unplanned Downtime

Industrial equipment failure remains one of the most significant causes of production loss. Traditional maintenance approaches are often reactive or scheduled, rather than predictive. This leads to inefficiencies and unexpected downtime.

Evaluation Criteria

A predictive maintenance system must provide:

Continuous multi-sensor data collection and monitoring
Real-time edge processing and AI-based anomaly detection
Industrial-grade reliability for 24/7 operation
Flexible connectivity with industrial equipment and sensors
Long lifecycle support for industrial deployments
Scalable architecture for different machine types and production environments

The Solution

A predictive maintenance system built on SolidRun’s Bedrock R8000 industrial edge computer enables continuous machine health tracking using multi-sensor data collection, real-time analytics, and AI-driven anomaly detection at the edge.

Integration Approach

The system integrates:

Sensors measuring vibration, temperature, oil quality, and flow
Local processing for anomaly detection
Industrial temperature-rated embedded platform
Scalable architecture for multiple machine types

Outcomes and Benefits

Reduced unplanned downtime
Early detection of equipment wear
Improved maintenance scheduling efficiency
Extended machinery lifecycle
Lower operational maintenance costs

Pipeline Inspection Systems

The Challenge: Infrastructure Monitoring in Hazardous Environments

Pipeline infrastructure in petrochemical plants and municipal systems requires constant inspection to ensure safety and operational integrity. However, these environments are often dangerous, difficult to access, and unsuitable for manual inspection.

Evaluation Criteria

A pipeline inspection platform must provide:

Reliable operation in harsh industrial environments
Real-time video processing and AI-based analysis
Compact embedded computing for robotic integration
Industrial temperature and long-term operational reliability
High-performance edge processing without cloud dependency
Flexible interfaces for cameras, sensors, and control systems

The Solution

A rugged pipeline inspection platform powered by SolidRun’s Bedrock R8000 industrial edge computer enables real-time video capture, AI-based analysis, and reliable data processing for robotic inspection systems operating in challenging industrial environments.

Integration Approach

Embedded video streaming system for live inspection
Integration with robotic pipeline traversal systems
Optimized for long-term industrial deployment
Designed for stability under harsh environmental conditions

Outcomes and Benefits

Improved inspection accuracy and coverage
Enhanced safety by reducing human entry into hazardous zones
Faster detection of infrastructure faults
Reduced maintenance and inspection costs

Conclusion: The Future of Industrial Intelligence at the Edge

Industrial automation is no longer defined solely by mechanical systems or centralized computing. It is increasingly driven by distributed intelligence, where machines are capable of sensing, processing, and acting in real time at the edge.

Across vision systems, robotics, agriculture, food automation, industrial gateways, and predictive maintenance, the common requirement is clear: compact, efficient, and intelligent embedded platforms capable of operating in demanding environments.

By enabling these capabilities through advanced System-on-Modules and edge computing platforms, SolidRun supports OEMs and developers in building the next generation of industrial systems—faster, smarter, and more resilient than ever before.