Less than a year since large language models (LLMs) like ChatGPT were released for wide public use, a lot of experimentation has taken place to learn how the technology can be used in industrial manufacturing environments. Though most of this experimentation has been done to enhance interactions with existing software applications, we’re beginning to see new products introduced that feature LLM capabilities.
At Automate 2023, Beckhoff gave a preview of its forthcoming TwinCAT Chat Client, which is designed to automate tasks such as the creation or addition of function block code. It can also be used for code optimization, documentation and restructuring.
This client, which will be implemented in TwinCAT XAE (extended automation engineering), works by connecting to the host cloud of the LLM. For example, if you’re using Microsoft Azure, it will connect to OpenAI’s ChatGPT. Beckhoff’s TwinCAT Chat Client provides a user interface for communication to the PLC development environment via a corresponding chat window in Visual Studio. According to Beckhoff, the LLM functionality has been optimized with TwinCAT-specific content.
In a demonstration at Automate, Daymon Thompson, director of product management at Beckhoff, showed how to use the technology to deliver step-by-step instructions on how to program a conveyor system. He explained how the software walks the user through each of the required variable inputs needed for an LLM to develop code. With those variables entered, selecting “autocomplete” in the client will send the variables’ data to the ChatGPT engine and return the completed code, typically within a few seconds.
As precise as the LLMs can be when developing code, Thompson stressed the need for users to thoroughly review the code before implementation. “If you're happy with what ChatGPT came back with, you can grab the program icon and drag and drop it into your TwinCAT program to create an entire program for running the conveyor,” he said.
The TwinCAT Chat Client makes it “a lot faster and more efficient to program code,” said Thompson, adding that Beckhoff is “excited to see where this is going to take us in the future as we build this out into our engineering environment even further.”
The video below captures Daymon Thompson's preview of the forthcoming TwinCAT Chat Client from Beckhoff.
