FANUC Integrates AI, Force Control and Digital Twins for IMTS

A manufacturing robot can only act on an AI-generated instruction after that instruction passes through several engineering layers: interpretation, program generation, sensing, motion control and physical validation. FANUC America’s planned IMTS 2026 demonstrations connect those layers into a common robotics and computer-controlled machining workflow, rather than presenting artificial intelligence as a stand-alone factory tool.

The Rochester Hills, Michigan-based automation supplier will demonstrate the systems in Booth 338900 at IMTS, scheduled for September 14–19 in Chicago. Its announced exhibit program combines multimodal AI, robot programming, camera and force feedback, digital twins, machine controls and mobile-robot power integration. The central engineering proposition is that AI interpretation can sit above established control and sensing systems without replacing them.

From human instruction to controlled motion

In a demonstration developed with Google Cloud, AI agents will interpret handwritten instructions and direct robots to identify, locate and kit parts. FANUC’s separate CRX Vibe Coding exhibit will use natural-language commands and generative AI to produce Python code and robot programs.

These demonstrations address the programming interface, but generated code is only one element of a production cell. The robot still needs a defined tool, coordinate system, reachable path, collision constraints and a controller capable of executing motion predictably. Camera and sensor data must also establish that the correct part is present and where it is located. IMTS should therefore provide an opportunity to examine how FANUC constrains AI-generated instructions before they become machine motion, although no comparative programming-time data, error rates or deployment results have been disclosed.

The next layer is feedback at the point of contact. A dual-arm CRX-5iA system will combine vision and force data for connector insertion and assembly. That pairing matters because vision can guide the arms toward a nominal location, while force information can indicate physical contact during alignment and insertion. Another assembly demonstration will use a CRX-30iA’s integrated force control to assemble a planetary gear system without external force sensors.

For work that does not remain stationary, FANUC will combine Inbolt tracking, NVIDIA-powered processing, a CRX-20iA/L robot and an R-30iB Mini Plus Controller to tighten bolts on moving parts. This shifts the control problem from reaching a fixed coordinate to continuously reconciling perceived part motion with the robot trajectory. FANUC has not supplied tracking accuracy, allowable part speed, processing latency or recovery behavior when tracking confidence falls, so the practical operating envelope remains unquantified.

Digital twins connect programming with commissioning

NVIDIA Isaac Sim will support a live digital twin of a robotic spot-welding application integrated with FANUC ROBOGUIDE. A separate CRX-5iA demonstration will synchronize changes between ROBOGUIDE and a physical robot handling computer hard drives. This bidirectional link is more consequential than a visualization alone: it is intended to let engineers transfer changes between virtual and physical representations of a cell.

Simulation can expose reach, timing and interference issues before equipment installation, but its value depends on model fidelity. Robot dynamics, tooling, fixtures, sensor behavior and process loads must match the deployed system closely enough for virtual results to remain useful. FANUC expects digital twins and virtual commissioning to shorten deployment, but it has not released benchmark comparisons or defined which commissioning checks still require physical validation.

Mobile power and machining complete the architecture

FANUC will also introduce the portable CRX-3iA collaborative robot, which has a stated payload of 3 kilograms and weighs 11 kilograms. More significant for mobile integration is the new R-50iA Compact DC Controller. Designed for CRX robots mounted on autonomous mobile robots, it accepts direct 24- to 48-volt battery power. FANUC says this removes the need for a separate inverter, reducing the number of power-conversion components between a mobile platform’s battery and the robot controller.

That design can simplify packaging on a mobile system, where mass, enclosure volume, cabling and conversion losses all affect deployment. However, the disclosed material does not quantify controller power demand, operating duration, thermal limits or the effect of robot duty cycle on vehicle range. Collaborative operation also remains application-specific; the presence of a collaborative robot does not by itself establish that a complete mobile cell is safe for unrestricted human proximity.

On the machining side, FANUC will display its 500i-A computer numerical control platform and debut three ROBODRILL DC Series machining centers: the D54CS, D74CS and D116CS. The largest, the D116CS, expands travel to 1,100 millimeters on the X-axis and 600 millimeters on the Y-axis. An M-810/190-20B robot will also perform wet milling, drilling and tool changes using FANUC machine control and robot G-code technologies.

Together, the exhibits outline a connected engineering chain: human intent becomes software, sensors qualify the physical task, controllers execute motion, and synchronized models support commissioning across robots and machine tools. What IMTS has not yet established is how reliably that chain performs outside a prepared demonstration. The decisive measurements will be interpretation errors, force and tracking tolerances, simulation-to-cell correlation, safe exception handling and commissioning time against a conventional workflow.

By Jonathan Barrett — Editor for AMI’s future mobility and autonomous systems section, with two decades covering robotics, e-mobility, drone-vehicle convergence, and transport mechanical systems.