Збирач потоків

Mineral exploration company Volta Metals Ltd of Toronto, Canada (which owns, has optioned and is currently exploring a critical minerals portfolio of rare-earths, gallium, lithium, cesium and tantalum projects in Ontario) has begun laboratory-scale bioleaching recovery test work primarily targeting gallium and secondarily rare-earth elements (REEs) at Dr Vasu Appanna’s laboratory of Biomine Research and Development at Laurentian University in Sudbury, Ontario. Laurentian University is recognized for its applied research expertise in mining, mineral processing, and earth sciences...

High-performance semiconductor, Internet of Things (IoT) systems and cloud connectivity service provider Semtech Corp of Camarillo, CA, USA has acquired HieFo Corp of Alhambra, CA –which manufactures indium phosphide (InP) optoelectronic devices for optical transceivers used across data-center interconnects (DCI) and intra-data-center interconnects – for about $34m in cash...

In booth #2027 at the IEEE Applied Power Electronics Conference (APEC 2026) in San Antonio, Texas (22–26 March), Navitas Semiconductor Corp of Torrance, CA, USA is exhibiting a 250kW solid-state transformer (SST) platform developed by the Power Electronics Laboratory of Switzerland’s École Polytechnique Fédérale de Lausanne (EPFL) that enables the grid architecture required by next-generation data centers, eliminating bulky low-frequency transformers while improving end-to-end efficiency...

Київська політехніка отримала додаткову грантову підтримку від Amazon Web Services kpi чт, 03/05/2026 - 09:54

Arrow Electronics and Infineon Technologies AG have announced REF_ARIF240GaN, a 240W USB Power Delivery (PD) 3.2 reference design for battery-powered motor control applications that require high performance and power efficiency in a compact form factor. This design complements the existing portfolio of joint reference design solutions from Arrow and Infineon, supporting the ongoing migration of customer designs to USB-C technology.

REF_ARIF240GaN is specifically designed to support the launch of EZ-PD PMG1-B2, Infineon’s newest USB PD 3.2 controller, featuring up to 240W USB sink capability and integrated buck-boost functionality in a compact single package. It provides developers with a ready-to-use platform for implementing high-power USB-C charging alongside efficient motor drive control features. It brings fast charging capabilities for 2- to 12-cell Li-ion battery packs, simplifying the overall design and reducing components count.

Motor control functionality is delivered using Infineon’s PSOC C3, a 180MHz Arm Cortex-M33 microcontroller, and highly efficient 100V CoolGaN G5 transistors. By combining a fully interoperable USB-C PD stack with high-performance sensor and sensorless GaN motor control on a single platform, the reference design enables compact, high-efficiency battery-powered systems while shortening development time, reducing bill of materials cost and space required.

Target applications include light electric vehicles (e-bikes, e-scooters and personal mobility devices), along with power tools, vacuum cleaners, kitchen appliances, garden equipment and robotics.

The reference design can be obtained upon request. Advanced technical support and customisation services are available from Arrow’s engineering solutions centre (ESC).

Visitors to embedded world 2026 can see the joint Arrow and Infineon solutions for motor control and battery-powered applications at Arrow’s stand 4A-342.

About Arrow Electronics
Arrow Electronics (NYSE:ARW) sources and engineers technology solutions for thousands of leading manufacturers and service providers. With 2025 sales of $31 billion, Arrow’s portfolio enables technology across major industries and markets. Learn more at arrow.com.

The post Arrow Electronics and Infineon introduce 240W USB-C PD 3.2 reference design for battery-powered motor control applications appeared first on ELE Times.

Factories today operate as dense mechanical ecosystems, whether in automotive assembly lines or semiconductor fabrication units. Traditionally, each robotic and mechanical element performed predefined, deterministic functions within isolated automation cells. However, as shop floors become increasingly machine-intensive and interconnected, operational complexity rises proportionally. Managing these environments now requires more than mechanical precision—it demands architectural coordination across layers of control and intelligence.

In this context, the convergence of Information Technology (IT) and Operational Technology (OT) is fundamentally reshaping robotics engineering. Data processing layers—analytics engines, business logic systems, and enterprise platforms—are no longer separated from operational control systems. At the same time, the physical layer, comprising sensors, actuators, servo drives, and Programmable Logic Controllers (PLCs), is becoming increasingly tightly integrated with edge compute and network infrastructure. Robotics systems are no longer designed as standalone motion units; they are engineered as nodes within a larger, connected control ecosystem.

“Traditional automation tools were built for a high-volume, low-variability environment. But today’s market demands agility,” says Ujjwal Kumar, Former Group President of Teradyne Robotics.

This architectural integration is shifting robotics engineering from a purely mechanical discipline toward system-level design—where communication protocols, deterministic networking, cybersecurity, and software orchestration are as critical as torque curves, kinematics, and payload specifications.

Adaptive Systems

At the core of this transformation lies the emergence of adaptive robotic systems. In practical terms, adaptability on the shop floor means the ability to reconfigure, scale, and modify operational behavior through software-defined control and network orchestration, rather than through mechanical redesign. Modern robots are no longer confined to fixed, pre-programmed routines. Equipped with AI models, IIoT connectivity, and high-resolution sensor feedback, they can interpret environmental inputs, process real-time data streams, and dynamically adjust execution parameters.

“The big difference is that traditional automation was a custom-made, perfect solution for one application. The new age of AI-integrated robotics has standard products serving multiple applications. You go into multiple applications through software and some end-of-arm tooling differences,” says Ujjwal Kumar, Former Group President of Teradyne Robotics.

As manufacturers pursue higher efficiency alongside greater product diversity, such adaptability becomes essential. Integrated control and data layers allow robots to transition between production tasks or product variants with minimal downtime, supporting high-mix manufacturing environments. Simultaneously, context-aware operations enable robotic systems to respond to signals from enterprise platforms such as ERP and MES, aligning execution with demand fluctuations, material availability, and downstream constraints.

The Build Architecture: Sensors, Control, and Communication Layers

To understand the engineering behind IT–OT convergence, it is useful to examine the architectural layers that define modern shop-floor robotics. Traditionally, industrial systems followed hierarchical models such as ISA-95, where field devices, control systems, and enterprise platforms operated in structured tiers with limited cross-layer interaction. Today’s robotic systems, however, are increasingly designed around a more unified Industrial Internet of Things (IIoT) architecture—where sensing, control, computation, and enterprise integration operate within a tightly interconnected framework.

“The groundbreaking automation innovations of the future won’t come from one single company but from close cross-technology ecosystem collaborations,” says Ujjwal Kumar, Former Group President of Teradyne Robotics.

At the foundation lies the physical and sensing layer. Modern robots are embedded with dense networks of encoders, force–torque sensors, high-resolution vision systems, vibration monitors, and environmental sensors—particularly critical in semiconductor manufacturing. Unlike earlier generations, where sensors primarily supported local closed-loop motion control, today’s sensing infrastructure generates continuous, time-synchronised data streams. These data flows serve a dual purpose: ensuring precision motion control while simultaneously feeding analytics and optimisation engines upstream.

Above this sits the control and communication layer, where deterministic execution remains paramount. PLCs, motion controllers, industrial PCs, and real-time operating systems govern microsecond-level synchronisation of servo drives and actuators. However, this layer has evolved from rigid, ladder-logic-driven hierarchies to hybrid architectures that combine deterministic control with networked intelligence. Industrial Ethernet, fieldbus systems, and increasingly Time-Sensitive Networking (TSN) ensure that motion commands and data packets coexist without compromising latency or jitter requirements. Control systems are no longer isolated—they are communicative nodes within a broader industrial network.

The next shift occurs at the edge. Edge computing nodes now preprocess high-frequency sensor data, execute AI inference models, and filter operational information before it propagates upward. Event-driven architectures and publish–subscribe communication patterns allow machines to update a shared operational state across the plant continuously. Rather than relying solely on hierarchical polling mechanisms, modern factories operate through near real-time data dissemination, enabling contextual awareness across production assets.

James Davidson, Chief Artificial Intelligence Officer, Teradyne Robotics, says, ” AI is transforming robots from tools into intelligent collaborators that can perceive, learn, and adapt.” 

At the enterprise integration level, robotics systems increasingly interact with MES and ERP platforms, digital twin environments, and predictive maintenance engines. Data flow is no longer unidirectional. Demand signals, material constraints, and quality metrics can influence robotic execution parameters in near real time. This bidirectional exchange is the practical manifestation of IT–OT convergence—where business logic and machine logic intersect.

Underpinning all these layers is a security and infrastructure framework that ensures resilience. As robots become connected assets, cybersecurity, network segmentation, device authentication, and secure firmware management become integral engineering considerations rather than afterthoughts. Connectivity without security would undermine determinism and operational continuity.

Redefining the Core of Robotics Engineering 

For decades, robotics engineering on shop floors was largely centred on mechanical excellence. Engineers focused on motion accuracy, payload capacity, repeatability, structural rigidity, and cycle-time optimisation. The primary goal was to design a robot that could execute a defined task with precision and reliability within a controlled cell.

That foundation still matters—but it is no longer enough. As IT–OT convergence reshapes shop floors, robotics engineering now extends far beyond mechanical design. Engineers must integrate advanced sensors, real-time communication networks, edge computing systems, AI-driven analytics, and enterprise software interfaces into the robot’s architecture. A robot is no longer just a mechanical arm with a controller; it is a connected, data-producing, and data-consuming system embedded within a larger digital ecosystem.

This means engineering decisions are no longer confined to gears, motors, and control loops. Network latency can influence motion stability. Data accuracy affects predictive maintenance outcomes. Software updates can modify operational behaviour. Cybersecurity vulnerabilities can interrupt production. Mechanical performance is now intertwined with software reliability and network integrity.

Physical AI equips robots with the capacity to perceive and respond to the real world, providing the versatility and problem-solving capabilities that are often required by complex use cases that have been out of scope until now,” says James Davidson, Chief AI Officer, Teradyne Robotics. 

In practical terms, robotics engineers are moving from designing machines to designing intelligent systems. They must think about interoperability, data structures, communication protocols, and secure integration—alongside torque curves and kinematics. The robot is no longer an isolated automation asset; it is part of a coordinated production architecture that responds to real-time information from across the enterprise.

The shift is clear: robotics engineering is evolving from a purely mechanical discipline into a multidisciplinary field where mechanics, electronics, networking, and software operate as a unified whole.

Conclusion 

As factories continue to evolve into connected, data-driven environments, robotics can no longer be engineered as standalone mechanical systems. The convergence of IT and OT is embedding intelligence, connectivity, and responsiveness directly into the core of robotic architecture. What was once a discipline defined by mechanical precision is now defined by system integration. 

“Taking a modern Industry 5.0 approach requires prioritisation of adaptability, empowering line workers with robots that can be reprogrammed and redeployed as demand shifts, which is the biggest benefit of having these very flexible systems coming online quickly,”  says Ujjwal Kumar, Former Group President of Teradyne Robotics.

The competitive edge will not belong merely to the fastest or strongest robots, but to those designed as intelligent, interoperable components of a unified production ecosystem. In this new industrial reality, robotics engineering is no longer just about motion—it is about orchestration.

The post Robotics Engineering: The Architectural Evolution Behind IT–OT Convergence appeared first on ELE Times.

As enterprises accelerate toward cloud-native infrastructure, edge computing, and virtualised network functions, data volumes and traffic patterns have become increasingly dynamic and unpredictable. This shift has significantly complicated network management, making traditional monitoring and testing approaches insufficient for modern workloads.

Software-Defined Networking (SDN) emerged as a response to this complexity. By decoupling the control plane from the data plane and centralising network intelligence in software-based controllers, SDN introduced programmability, agility, and fine-grained policy enforcement into network architecture. Networks were no longer static hardware constructs — they became programmable systems capable of real-time configuration and orchestration.

However, this programmability has introduced a new challenge: protocol behaviour is no longer deterministic. Dynamic flow rules, frequent controller updates, real-time policy changes, and multi-controller orchestration have made protocol validation exponentially more complex. Traditional pre-defined test scripts and static regression libraries struggle to keep pace with continuously evolving network states.

“AI applications are driving an entirely new set of requirements in our customers’ network equipment and in their network architectures,” says  Joel Conover, senior director at Keysight Technologies

In programmable environments, protocols must be validated not just for correctness, but for adaptive behaviour across changing topologies and traffic conditions. This is precisely where Artificial Intelligence is beginning to redefine network protocol testing — shifting it from rule-based verification to intelligent, adaptive validation.

Traditional Protocol Testing Failing with SDNs

With legacy traditional networks, the protocol behaviour remains largely uniform and predictable. Routing tables were static, firmware updates were infrequent, and network state changes followed predictable patterns. Testing technologies evolved accordingly – with pre-defined test cases, fixed traffic simulations, and rule-based regression suites. But with Software Defined Network, that isn’t the case. 

SDN disrupts this very uniformity and predictability. As with SDN, the control plane is abstracted into centralised controllers, and the network remains largely flexible- not hardcoded into individual devices. Flow rules are dynamically installed, modified, or withdrawn based on application demands, policy engines, and real-time telemetry. As a result, network state becomes fluid rather than fixed. This also puts forth tremendous testing challanges including: 

• Dynamic Flow Table Updates: In SDN environments, flow entries can change in milliseconds. Traditional test scripts, designed for static configurations, cannot continuously validate transient states or short-lived rule conflicts.
• Controller-Driven Logic Complexity: Unlike legacy networks, where protocols like Open Shortest Path First (OSPF) or Border Gateway Protocol (BGP) operate autonomously within devices, SDN controllers introduce centralized decision-making logic. Testing must now validate not only protocol compliance, but also controller algorithms, northbound applications, and southbound API interactions.
• Multi-Controller and Multi-Domain Orchestration: Large deployments often rely on distributed controller clusters for scalability and redundancy. Synchronisation delays, inconsistent state propagation, or split-brain scenarios introduce validation complexity beyond conventional test frameworks.
• CI/CD-Driven Network Updates: Modern SDN deployments increasingly follow DevOps models, where network policies and configurations are updated frequently. Regression cycles that once ran quarterly may now need to be executed daily or continuously.
• Emergent Behavior in Programmable Networks: When multiple applications interact through a controller — security policies, load balancers, traffic optimizers — unintended rule interactions can produce emergent protocol behavior. Static test matrices cannot anticipate such combinations.

In this evolving environment, traditional test automation tools operate reactively. They verify what has been explicitly defined, but struggle to discover what has not been anticipated. As SDN architectures scale in complexity, protocol testing must evolve from deterministic validation — capable of learning network behaviour rather than merely executing predefined scenarios.

The Limits of Automation in Modern SDN Testing

As SDN environments grew in complexity, testing frameworks also adopted automation. Continuous integration pipelines began validating controller updates, traffic replay tools simulated workloads, and orchestration layers executed regression suites at scale. Usually, the traditional automated testing systems operate on predefined logic. They execute scripted scenarios, compare outputs against expected results, and flag deviations. While this approach accelerates validation cycles, it remains fundamentally reactive. It can only test what engineers anticipate. In programmable networks, however, not all behaviours are foreseeable.

With SDNs, Flow rules interact dynamically, policies overlap, and controllers adapt in real time to the telemetry inputs. Under such conditions, failure modes are often emergent rather than explicit. They arise from complex interactions between components rather than from isolated configuration errors.

This is where the limitations of deterministic automation become evident:

• Static rule engines cannot adapt to evolving topology states.
• Regression libraries cannot scale combinatorially with policy variations.
• Manual definition of edge cases becomes impractical in large-scale SDN fabrics.

As networks increasingly resemble distributed software systems, testing must adopt characteristics of software intelligence — the ability to learn patterns, detect deviations autonomously, and anticipate risk scenarios. It is within this context that Artificial Intelligence begins to move from experimental concept to architectural necessity.

How is AI replacing the Automation Debate in Testing? 

As Software-Defined Networks evolve into highly dynamic, programmable infrastructures, testing frameworks must move beyond deterministic execution models. AI-driven protocol testing becomes the obvious and most promising strategy as it is enhanced with contextual learning, predictive analysis, and adaptive decision-making. An effective AI-enabled SDN testing architecture operates across multiple functional layers.

“AI is being infused into many aspects of communications technology – it shows particular promise in predicting channel conditions, essentially creating new forms of ‘smart radios’ that can achieve higher throughput and/or longer distances by incorporating machine learning in the radio itself,” says  Mr Conover. 

At the foundation lies a telemetry intelligence layer. SDN environments generate vast volumes of real-time data — including flow table updates, controller logs, latency metrics, packet drops, topology transitions, and API interactions across northbound and southbound interfaces. Rather than relying solely on post-event log analysis, AI models ingest and process this telemetry continuously. By establishing behavioural baselines, the system distinguishes between acceptable adaptive changes and genuine protocol anomalies.

Built upon this is the Behavioral Modeling Layer. In programmable networks, protocol validation must account for interactions between controllers, applications, and dynamic policies. Machine learning models analyse how control-plane decisions influence data-plane outcomes under varying traffic loads, topology shifts, and failover scenarios. Through supervised and unsupervised learning techniques, the system identifies normal operational patterns and detects deviations that static scripts might overlook — such as cascading latency effects, unstable rule propagation, or intermittent synchronization gaps.

The next layer introduces Intelligent Test Case Generation and Prioritisation. Traditional regression testing treats all scenarios uniformly, often leading to inefficiencies. AI-enhanced systems instead evaluate historical defect data, configuration change patterns, and policy dependency graphs to calculate risk scores. Testing resources are then dynamically allocated to high-risk areas. Reinforcement learning techniques can further simulate targeted disruptions, enabling adversarial-style validation that exposes weaknesses before deployment.

Finally, Predictive Validation capabilities elevate protocol testing from reactive detection to proactive assurance. By analysing patterns across multiple test cycles, AI systems can forecast potential congestion points, controller overload risks, and policy conflicts at scale. This predictive insight is particularly valuable in CI/CD-driven SDN environments, where frequent updates demand continuous and reliable validation.

Together, these layers transform protocol testing from a script-driven verification exercise into an adaptive, intelligence-led framework. As networks become software-defined, testing infrastructures are becoming learning-defined — capable not only of validating correctness, but of anticipating instability before it manifests in production environments.

Conclusion

Software-Defined Networking transformed networks into programmable, software-driven systems — but in doing so, it also made protocol validation far more complex. Static test scripts and deterministic regression cycles are no longer sufficient for environments defined by dynamic flows, controller logic, and continuous updates.

“The use case for network testing is emulating the unique properties of that environment, and delivering it at a scale we’ve never seen before,”  says  Mr Conover. 

Artificial Intelligence is emerging as the natural evolution of network testing. By learning behavioural patterns, detecting anomalies in real time, and prioritising risk intelligently, AI shifts protocol validation from reactive verification to predictive assurance.

The future of SDN will not depend solely on how programmable networks become, but on how intelligently they are tested. As infrastructure grows more dynamic, validation must become equally adaptive — combining automation, intelligence, and human oversight to ensure resilient, scalable network operations.

The post How AI Is Transforming Network Protocol Testing in Software-Defined Networks? appeared first on ELE Times.