Industrial Automation

The invention is usually software, and eligibility turns on what it does to the machine.

Much of what is patentable in industrial automation is now software: the environment a control project is engineered in, the pipeline that carries plant data to the cloud, the digital twin that stands in for a process before it runs. Claimed as information gathered, analyzed and displayed, that software draws the same eligibility rejection as any other. Claimed through a specific change in how a controller, a safety system or a running process operates, it is a claim to an improved machine, and that is the ground on which eligibility is argued.

Published patents in this area: 309
Professionals: 10

We draft and prosecute patent applications across industrial control and the systems built around it: industrial controllers and the way they store and exchange data, the design environments in which automation projects are engineered, edited and shared among developers, digital twins and the simulation of industrial processes, and the pipelines that move plant data from edge devices to cloud analytics. The same work reaches functional safety and emergency-stop systems, operator interfaces and dashboards, adaptive control and controller tuning, equipment diagnostics and maintenance, and the connections between the plant floor and manufacturing execution, distributed control and enterprise resource planning systems.

This work shares its eligibility discipline with Software & Computing, and it is an area of its own because its claims are anchored in controllers and the processes they run rather than in general-purpose computing. The boundaries follow what is being controlled: manipulators and robot control are handled in Robotics, vehicle guidance in Autonomous Vehicles, and the mechanisms and machine tools themselves in Mechanical & Materials. Where a controller or design environment uses a trained model, the model architecture belongs to AI & Machine Learning and the control system around it belongs here; the security of control networks and devices is handled in Cybersecurity.

Section 101 and the plant-data claim

Many automation inventions are about data: acquiring it from separate device subsystems on a common time base, giving it context, transforming it for the system that will consume it, and presenting it on a dashboard or in the cloud. A claim that stops at collecting, analyzing and displaying information matches the pattern the Federal Circuit held abstract in Electric Power Group, and data from a plant is not exempt because it came from a machine.

A physical step added at the end does not cure that. Diehr and Flook, both process-control cases, mark the line: Diehr upheld a rubber-curing process in which a computer repeatedly recalculated the cure time and opened the press, and Flook rejected a method whose only new element was the formula for updating an alarm limit, even though the limit governed a running chemical process. What holds is a claim in which the control step is where the invention does its work, or one that improves how the controller or its network itself operates, such as keeping data consistent in controller memory. The specification has to present the invention that way from the start, because an examiner will not supply it.

Obviousness and the known software technique

Collaborative editing, version control, rollback to a milestone, synchronized text and graphical views of the same program: an examiner can usually find each of these in general software engineering and will argue that applying it to an automation project was obvious. After KSR, a predictable use of a known technique in a neighboring field is likely to be held obvious, and that is the rejection a design-environment application should expect.

The answer lies in what makes a control project different from ordinary source code. The program is bound to physical I/O and to a particular controller, it may be edited while that controller is running a process, and returning it to an earlier state may have to account for the hardware as well as the text. A specification that sets out those constraints, with claims that recite the step meeting them, gives the examiner a reason the general technique does not simply transfer; an application that describes the tool only as software gives that argument nothing to stand on.

Evidence

Patents we drafted in this area

Patents the firm prosecuted, drawn from its own published listing and from the USPTO's attorney-of-record index. A record, not a claim.

All patent listings

12,724,404Industrial data destination transformation
12,717,296Generative artificial intelligence for creation of instruction code from an input
12,693,648Industrial artificial intelligence model aggregation
12,687,824Manufacturing cloud system
12,681,469Generative AI for industrial automation design environment
12,681,464Front end user interface for manufacturing cloud system
12,675,101Industrial automation relational data extraction, connection, and mapping
12,675,099Industrial control program component exploration and playback