From Ports to Welding Arcs: Automation Systems Bring AI Onto the Factory Floor
Universal Robots, part of Teradyne Robotics, unveiled its seventh-generation robot platform, Gen 7, at the International Manufacturing Technology Show in Chicago. It had been years since the previous generation launched. The most noteworthy change is not payload or speed. It is in a place many people overlook: the tool flange.
In the past, industrial cameras, force sensors, and end effectors required separate wiring. Cable harnesses bent repeatedly with each joint movement, making them the most failure-prone and hardest-to-validate part of integration. Gen 7 integrates Gigabit Ethernet, 24/48V power, and safety I/O directly into the flange. Data, power, and safety signals reach the tool end through the same channel. Built-in force/torque sensing and impedance control give the robot a sense of touch, enabling force-sensitive assembly and delicate manipulation tasks. The significance of this design change is that the robot is no longer just an actuator. It becomes a node with a real-time data link and local computing power, capable of dividing traffic and permissions with industrial PCs, vision systems, and MES platforms.
Around the same time, Lincoln Electric revealed during an investor call that its first Physical AI product will debut at FABTECH in October, with orders starting in the fourth quarter. The system combines vision technology, welding process intelligence, and collaborative robot capabilities. Lincoln Electric previously acquired Inrotech to obtain vision system technology, enabling cobots to recognize and respond to unstructured welding conditions rather than being limited to preset paths. CFO Bruno said the automation business is spreading from automotive into general industrial, heavy industrial, and structural fabrication. Automotive currently accounts for about 40 percent of its automation business, down from a peak of nearly half after the Fori Automation acquisition. Lincoln Electric aims to adjust the mix so general industrial, heavy industry and structural fabrication, and automotive each account for roughly one-third.
Lincoln Electric’s order data reflects the breadth of automation demand. Bruno noted that capital investment confidence typically follows stable production activity. Standard welding equipment demand has reached an inflection point. Automation investment continues. Consumables sales posted mid-single-digit growth in the Americas welding segment in the second quarter, a signal that industrial activity is stabilizing and growing.
Domestic automation market data points in a similar direction. China‘s OEM automation market grew 8 percent year-on-year in the second quarter of 2026. The battery segment grew 35 percent. Industrial robots and semiconductors both exceeded 20 percent. PCB was the most elastic, growing 55.6 percent. The buildout of AI servers, computing hardware, high-frequency PCBs, and wafer lines is creating demand for micron-level positioning and high-tempo flexible production. Order growth for automation equipment across the full supply chain exceeded 30 percent.
This demand is penetrating more specialized scenarios. Yijiahe and Huawei Cloud jointly launched the next-generation live-line distribution network robot Z200Pro, marking a shift from autonomous operation in known scenarios to embodied intelligence capable of autonomous perception and dynamic decision-making in complex, unknown field conditions. Hangcha Group announced plans to issue convertible bonds to raise up to 2.259 billion yuan for a forklift robot smart factory and logistics robot R&D projects. The traditional forklift category is being redefined as a mobile automation platform.
B&R released 13 automation and industrial AI products. Its M series industrial AI solutions connect machine perception, autonomous understanding, and process optimization, offering equipment manufacturers an end-to-end path from data collection to intelligent optimization. Universal Robots’ Gen 7 runs on the PolyScope X operating system, which opens APIs and ROS2 communication, allowing developers to run edge AI applications directly on the robot while the safety controller retains veto power over final decisions.
From the Gigabit interface on the tool flange to visual recognition in welding robots to dynamic decision-making in live-line distribution robots, the evolution of automation systems is converging. Equipment no longer just executes preset programs. It is beginning to perceive its environment, understand processes, and make judgments within safety boundaries. The speed of this shift depends on a simple question: are production lines willing to redesign their interfaces and workflows for machines that can think?