Industrial Intelligence Is Moving Closer to the Machine
Artificial intelligence is changing how industrial automation systems process information, make decisions, and respond to changing operating conditions. Rather than relying entirely on centralized computing resources or cloud-based analytics, manufacturers are increasingly exploring Edge AI, which brings data processing and intelligent decision-making closer to sensors, controllers, embedded platforms, and production equipment.
The IIoT Forum at electronica 2026 will examine these developments on November 10–11 in Hall C6 at the Future Tech Stage in Munich, Germany. Organized in cooperation with Embedded Computing Design, the forum will cover Edge AI, industrial connectivity, cybersecurity, motor control, robotics, and predictive maintenance.
For industrial engineers, the discussion extends beyond computing performance. The real challenge is integrating AI into existing automation architectures while maintaining predictable control behavior, protecting operational data, and ensuring that intelligent functions operate within the technical constraints of industrial environments.
In my view, the most important question is not how much intelligence can be placed at the edge, but how effectively that intelligence can be integrated into a production system without introducing new operational risks.
Day 1: Industrial Computing Platforms and Long-Term Cybersecurity
The first day begins at 2:00 p.m. with “Qualcomm Dragonwing: Modernizing Industrial Automation for the AI Era.” Anand Venkatesan, Senior Director of Product Management at Qualcomm Technologies, will discuss the role of modern computing platforms in supporting AI-enabled industrial applications.
As embedded processors become capable of handling more demanding workloads, automation designers have additional options for machine vision, local data analysis, and condition monitoring. However, selecting a processor is only one part of system design. Engineers must also consider software compatibility, communication interfaces, thermal management, power consumption, and integration with existing PLC and industrial control architectures.
At 2:45 p.m., Balraj Kathiresan, Principal Engineer at Infineon, will present “Mitigating Security Risks in the Era of Quantum Attacks.” The session focuses on the security implications of emerging quantum computing capabilities and the long service lives of connected industrial equipment.
Industrial systems often remain operational for many years after their initial deployment. Security mechanisms that are adequate today may require updates as cryptographic standards and threat capabilities evolve. This creates a practical challenge for equipment manufacturers and plant operators: cybersecurity must be considered throughout the product lifecycle, not treated as a one-time commissioning task.
For automation engineers, this means evaluating secure communications, authentication, firmware update mechanisms, access control, and the ability to maintain cryptographic protections across legacy and modern equipment.
The day's program concludes with the “AI at the Edge for IIoT” panel from 4:30 to 5:15 p.m. Participants include Vinay Agarwal of Texas Instruments, Karthi Gopalan of Renesas, Robin Saltnes of Nordic Semiconductor, and Yankin Tanurhan of MIPS/GlobalFoundries. Rich Nass of Embedded Computing Design will moderate the discussion.
The panel will explore the opportunities and engineering considerations associated with processing AI workloads closer to industrial devices. Local processing can reduce dependence on cloud connectivity and support faster responses, although actual performance depends on the application, network architecture, computational workload, and control requirements.
From an engineering perspective, Edge AI should complement established automation functions rather than automatically replace them. Deterministic control loops, interlocks, and safety-related functions must retain their defined behavior, even when AI-based applications are introduced for optimization or decision support.
Day 2: Physical AI and Intelligent Robotic Systems
The second day shifts the focus toward physical AI, robotics, embedded storage, and autonomous machine behavior.
At 2:00 p.m., Jean-Jacques Adragna, Staff Manager at Qualcomm France, will deliver “Advancing Physical AI with Qualcomm Dragonwing Robotics.” The presentation will examine computing platforms for intelligent robotic systems, where perception, decision-making, and physical interaction must operate within practical processing and power limits.
Unlike applications that generate recommendations or analyze historical data, robotic systems must translate information into physical actions. Cameras, sensors, actuators, motor drives, embedded processors, and control software must work together within the timing and operating constraints of the machine.
This introduces a distinction that is particularly important for industrial applications: recognizing an object or predicting a condition does not automatically mean that a robot can act on that information safely. The complete system must account for sensor accuracy, communication latency, actuator response, motion constraints, and the consequences of incorrect decisions.
At 2:45 p.m., Donnie Garcia, PSoC Marketing Manager at Infineon, will present “Beyond Bigger Memory: Smart Storage and AI Performance on PSoC Edge.”
The session addresses the relationship between memory, storage architecture, and AI execution. Embedded AI performance depends on more than processor throughput. Memory capacity, data-transfer rates, storage access, and resource allocation can all influence application performance.
For equipment designers, these factors become particularly relevant when deploying machine vision, sensor fusion, or multiple AI functions on a resource-constrained platform. Hardware selection should therefore be based on the complete application workload rather than processor specifications alone.
At 3:30 p.m., Ivo Marocco, Vice President and Head of UX at Renesas, will discuss “Managing Complexity in the Humanoid Robotics Ecosystem: A Systems Approach to Scalable Autonomy.”
Humanoid robots combine sensing, actuation, computation, communications, and control software within increasingly complex architectures. Coordinating these subsystems requires careful management of interfaces, resource allocation, system behavior, and fault conditions.
The same principle applies to industrial robotics more broadly. As machines become more autonomous, system integration becomes increasingly important. A capable processor or sophisticated AI model cannot compensate for poorly coordinated subsystems, inadequate feedback mechanisms, or unclear fault-handling strategies.
Physical AI on the Factory Floor: From Demonstration to Deployment
The second day's final panel, “Physical AI on the Factory Floor,” begins at 4:30 p.m. The discussion will feature Lee Gonzales-Fuentes, Global Product Marketing Manager supporting PSoC Control MCUs at Infineon, and Larry Olivas, Vice President of Secure Connected Edge at NXP Semiconductors. Rich Nass will again serve as moderator.
The discussion returns to the practical requirements of industrial production, where intelligent machines must interact with real equipment, processes, and working environments. Deploying AI on a factory floor involves more than executing a model locally. Engineers must determine how AI outputs interact with machine controls, production sequences, safety mechanisms, and supervisory systems.
For example, an AI-based condition-monitoring application may identify an unusual vibration pattern and recommend an inspection. Integrating that information into a maintenance workflow is different from allowing an AI system to directly modify machine motion or process parameters. Each application requires a clearly defined level of authority, appropriate validation, and suitable responses to uncertain or incorrect results.
In my assessment, successful Physical AI deployment will depend on establishing clear boundaries between probabilistic AI functions and deterministic control functions. AI can provide valuable perception, classification, and optimization capabilities, while conventional control logic continues to manage time-critical sequences and validated interlocks. Safety-related functions must follow the applicable safety architecture and cannot rely on an AI model alone unless the complete implementation has been appropriately assessed and certified for that purpose.
This approach allows manufacturers to explore AI capabilities while retaining the operational discipline required for industrial production.
Predictive Maintenance and Industrial Connectivity Remain Core Requirements
Predictive maintenance and industrial connectivity are also part of the IIoT Forum's broader program. These technologies provide the infrastructure needed to collect equipment data, identify developing faults, and coordinate maintenance activities.
Edge processing can help analyze vibration, temperature, current, pressure, and other operational measurements close to their source. This may reduce the volume of data sent to centralized systems and support faster detection of abnormal conditions. However, the usefulness of predictive maintenance still depends on sensor quality, data consistency, appropriate analytical models, and validation against actual equipment behavior.
Connectivity presents a related challenge. Industrial environments frequently contain equipment from different generations and manufacturers, each with its own interfaces, protocols, and maintenance requirements. Integrating edge devices into existing PLC, SCADA, DCS, and manufacturing execution environments requires careful attention to interoperability, network segmentation, data ownership, and system availability.
A practical deployment strategy should therefore begin with a clearly defined operational problem. Engineers should establish baseline performance, determine which data is required, evaluate the consequences of communication or processing failures, and measure whether the resulting application improves maintenance or production outcomes.
What Industrial Engineers Should Take Away from electronica 2026
The IIoT Forum highlights several connected engineering priorities: edge computing, long-term cybersecurity, embedded resource management, robotic autonomy, and the integration of AI with physical control systems.
These topics should not be treated as independent technology trends. They are parts of the same industrial architecture. More capable edge processors create opportunities for local intelligence, but they also increase software and security requirements. More autonomous robots introduce new integration demands, while predictive maintenance depends on reliable data collection and meaningful analysis.
For manufacturers, the priority should be to identify applications where edge intelligence delivers a measurable benefit, such as reduced diagnostic latency, improved fault detection, or more efficient use of machine resources. Those applications should then be evaluated against operational constraints, cybersecurity requirements, lifecycle costs, and the need to maintain established control and safety functions.
The transition from AI experimentation to industrial deployment will not be determined by computing performance alone. It will depend on how well engineers integrate intelligent software with proven automation principles, maintainable hardware, secure communications, and clearly defined operational responsibilities.
The IIoT Forum at electronica 2026 provides a venue for examining these challenges through technical presentations and expert discussions. Attendees will also have opportunities to ask questions at the conclusion of each session, allowing them to explore how emerging technologies may apply to their own industrial systems.
Ultimately, the strongest industrial AI solutions will be those that solve specific engineering problems, fit into existing operational architectures, and deliver measurable improvements without compromising the control, security, and safety requirements of the equipment they support.
