LG CNS, Figure move toward AI guidance

Robotics is moving past the old factory playbook. The old model was simple — script every motion, fence off the environment, and pray nothing weird happens. But warehouses, hospitals, and back-of-house service work are full of weirdness. That is why the interesting news is not just “better robots.” It is the rise of software layers that tell mixed fleets what to do, when to escalate, and how to recover when reality stops matching the script. LG CNS and Figure are both moving in that direction, just from different ends of the stack. (msn.com) ### What did LG CNS actually show? LG CNS demonstrated a system it calls Physical Works, where different robot types — including humanoid, quadruped, and wheeled machines — work together on warehouse tasks instead of operating as isolated appliances. That matters because most industrial automation still breaks work into narrow islands: one robot for (msn.com)the hardware. (msn.com) ### Why is the software layer the real story? A robot arm or mobile base is only useful if somebody decides what it should do next. In older systems, that “somebody” was a hard-coded workflow. In newer systems, it is increasingly an AI policy layer that can interpret goals, assign tasks, and handle exceptions. LG CNS said its Mobile Shuttle now inclu(msn.com)g. That is a big shift — from remote control to supervised autonomy. (lgcns.com) ### What makes warehouses the first big proving ground? Warehouses are messy enough to need adaptation, but structured enough to measure whether the software works. LG CNS says its upgraded shuttle operates in -15°F cold-chain sites, carries up to 3,307 pounds, and improves storage efficiency by more than 30% versus conventional two-way systems. Those are concrete operational gains, not just(lgcns.com)ask assignment in real time, adding more autonomous decision-making becomes a natural next step. (lgcns.com) ### Where does Figure fit into this? Figure is pushing the same idea through humanoids. Its Helix system started as a vision-language-action model for manipulation, then moved into logistics tasks like package triage, and by January 2026 Helix 02 extended control to the full body — walking, balancing, and manipulating as one continuous system. The important part is not the humanoid form by i(lgcns.com)s in the real world. (figure.ai) ### Why does “full-body autonomy” matter so much? Because real work is not a sequence of frozen poses. A robot has to notice a shifted box, change its footing, reach around clutter, and keep going. That is the hard part. Figure describes Helix 02 as long-horizon autonomy with no resets during multi-minute tasks. In plain English, the robot is doing more of the “figure it out as you go” work that humans do automatically. (figure.ai) ### What about healthcare? The healthcare angle is less about humanoids roaming hospitals today and more about the same orchestration logic moving into regulated, data-heavy workflows. LG CNS has been expanding its AI transformation business in pharma and digital health, including an AI clinical-trial platform and an automated quality-review system that cut document preparation time by 90%. That is not a warehouse robo(figure.ai)gentic software coordinating complex steps, humans, and exceptions. (chosun.com) ### So what changes next? The near-term winner is not the most human-looking robot. It is the company that can run a mixed fleet reliably when sensors fail, priorities change, or a human needs to step in. PrismaX is explicitly positioning itself as a service layer for exactly that world — robots, data, and human work tied together in one operational system. Basically, the control plane is becoming the product. (prismax.ai) ### Bottom line? The industry is shifting from robots as single-purpose machines to robots as endpoints in a larger AI operating system. LG CNS is showing that in logistics and enterprise workflows. Figure is showing it in humanoid autonomy. The hardware still matters — but the real moat is starting to look like orchestration.