Open-Source Physical AI, Robotic Welding and Resilient Manufacturing — September 24, 2026
Automation History does not claim authorship or ownership of the third-party articles featured in our Daily Articles section. All articles, research, and reporting are credited to their original authors and publishers, with links provided to the original sources. Automation History shares these links solely for educational and informational purposes, helping readers discover and learn from relevant content. All rights remain with the respective authors and publisher
1. Intrinsic open-sources a foundation for physical AI
Intrinsic announced on September 22 that it has released Intrinsic Core under the Apache 2.0 license, providing ROS-compatible real-time control, asset models and a native digital twin that can run on local hardware. Its companion Open Machine Tending Solution is a customizable reference design for AI-enabled CNC tending rather than a finished turnkey cell.
Why it matters: Open components can reduce custom integration work, but they also transfer more lifecycle responsibility to the operating team. Repair leaders should freeze validated versions, document drivers and calibration assets, and test rollback before an open software stack becomes production-critical.
Read Intrinsic’s September 22 announcement
2. KUKA brings production-scale friction stir welding into research
KUKA said September 24 that Magdeburg-Stendal University of Applied Sciences has installed an industrial research cell built around the KR FORTEC ultra MT for high-force friction stir welding and milling. The platform adds a digital twin and continuous process-data capture so researchers can develop, optimize and validate complete process chains under production-relevant conditions.
Why it matters: High process forces make rigidity, tool condition, fixturing and robot calibration inseparable from weld quality. Maintenance teams should treat spindle health, force data and fixture inspection as one condition-monitoring system rather than separate checks.
Read KUKA’s September 24 announcement
3. Kuehne+Nagel and CATL link battery logistics with fleet electrification
Kuehne+Nagel and CATL signed a memorandum announced September 24 covering battery logistics, European transport electrification, recycling, aftersales support and storage and charging systems at logistics sites. Their first projects include electrifying CATL transport flows in Europe and launching a Qiji battery-swapping pilot for heavy trucks in China this year.
Why it matters: A battery swap is simultaneously a vehicle, charging, material-handling and inventory operation. Supervisors need common ownership for connector wear, lift alignment, thermal monitoring, quarantine of damaged packs and recovery when the energy system or transport schedule falls out of sync.
Read Kuehne+Nagel’s September 24 announcement
4. Unilever uses digital twins to tune real factory utilities and recipes
In a September 23 manufacturing analysis, Unilever said energy-efficiency projects have saved it more than €1.5 billion since 2008 and that it plans more than 40 AI-enabled digital twins over the next 18 months. At its Haldia detergent plant, a twin is being used to optimize fan speeds, temperature and moisture settings while balancing emissions, waste, quality and demand.
Why it matters: A useful twin must stay synchronized with the physical plant after sensors drift, equipment is replaced and recipes change. Repair teams should own data-quality checks and configuration updates so simulated savings do not rest on an obsolete machine state.
Read the September 23 Manufacturing Dive analysis
5. Extreme heat is already disrupting electronics supply chains
A Global Electronics Association survey reported September 23 found that two-fifths of manufacturers worldwide experienced heat-related supply-chain disruption in 2026, with roughly half in North America, Europe and Asia-Pacific reporting supplier-fulfillment effects. More than four-fifths of affected companies saw productivity or throughput suffer, while cooling limits and power reliability ranked among the leading future concerns.
Why it matters: Heat resilience belongs in reliability planning, not only environmental reporting. Repair supervisors should test chiller capacity, electrical derating, enclosure temperature alarms, shift plans and critical-supplier contingencies against multiweek heat scenarios.
Read the September 23 Manufacturing Dive report
6. Banks warn that AI purchasing agents are outrunning payment safeguards
Banks including NatWest, Bank of America, ING and Commonwealth Bank of Australia warned September 22 that AI shopping agents are advancing faster than standards and consumer protections. Their report highlights direct handling of card details, steering toward weaker payment protections, unclear accountability and exposure to scams, fraud and privacy breaches.
Why it matters: The same control problem appears when an industrial agent orders parts, books freight or authorizes service: convenience can bypass approval and segregation of duties. Operations leaders should enforce spending limits, verified vendors, complete action logs and human approval for exceptions before agents receive purchasing authority.
Read the September 22 Reuters report
7. Accelevation’s IPO puts power and cooling at the center of the AI boom
Accelevation and selling shareholders aim to raise as much as $720 million in a U.S. IPO valuing the company at up to $5.37 billion, Reuters reported September 22. Founded in 2017, the Ohio company designs, manufactures and installs power distribution, cooling and modular infrastructure for data centers.
Why it matters: The valuation story is also an operations story: AI capacity is constrained by switchgear, cooling loops, fabrication and field installation. Leaders should evaluate maintainability, spares and single-vendor dependencies in those systems with the same rigor applied to servers and accelerators.
Read the September 22 Reuters report
8. Court blocks New Mexico’s PFAS label, but factory obligations keep moving
Manufacturing Dive reported September 23 that a federal judge temporarily blocked New Mexico’s consumer label for products containing intentionally added PFAS, which had been scheduled for January 1, 2027. The ruling targets the label requirement, while the state’s broader reporting and phaseout framework continues and the case may still be appealed.
Why it matters: A paused label does not justify stopping material traceability or change preparation. Manufacturing teams should maintain PFAS bills of material, supplier declarations and packaging configurations so legal changes do not trigger a rushed redesign across multiple lines.
Read the September 23 Manufacturing Dive report
9. AI tracking closes the loop for miniature robots inside living systems
A Science Robotics research article published September 23 presents AI-driven, actuation-compatible tracking for closed-loop navigation of miniature robots in vivo. The approach addresses a central constraint of medical microrobots: their bodies are often too small for onboard sensing and power, so external tracking must work continuously with the actuation system.
Why it matters: For invention tracking, this is a shift from merely moving a tiny device to observing and correcting it during a live mission. The industrial parallel is direct: autonomy becomes dependable when sensing, actuation and feedback timing are engineered as one validated loop.
Find the September 23 research article in Science Robotics
10. Maya meliponary preserves a preindustrial production system
INAH announced new details September 23 about a Maya stingless-bee facility excavated near Tulum: a 19.5-meter-long masonry shelter designed for rows of hollow-log hives, with 40 limestone and coral discs that may have served as lids. The probable Late Postclassic site also contained honey-storage pottery and formed part of a homestead combining residence, agriculture, storage and production.
Why it matters: For Automation History, the site records an organized production system built around standardized containers, environmental protection, storage and resource management without powered machinery. It is a reminder that good operations begin with flow, repeatable interfaces and protection of the process—not with automation itself.