Industrial Artificial Intelligence And Software Defined Automation Systems

Bosch Rexroth and AMD are cooperating in the development of automation solutions for the direct execution of artificial intelligence in industrial production environments. The technical partnership focuses on the decoupling of hardware and software to deploy machine learning and agentic AI locally at plant facilities.

Technical Architecture and System Integration
The open, Linux-based ctrlX OS operating system from Bosch Rexroth forms the software foundation and supports a wide range of embedded processors from the chip manufacturer AMD. The deployed AMD Ryzen hardware platform combines CPU, GPU, and NPU resources within a unified system. This system architecture provides the necessary computing power for classic automation tasks, graphics processing, and AI inference directly at the machine, without the need to implement additional accelerator cards or an external cloud infrastructure.

Local Data Processing and AI Agents
The execution of AI applications occurs entirely locally on the edge hardware. Multiple AI agents as well as integrated large language models access engineering and current process data via the ctrlX Data Layer and a central Model Context Protocol (MCP) server. The MCP server running on ctrlX OS connects all relevant interfaces and provides the artificial intelligence with the necessary industrial context in real time. Through this local approach, critical machine data remains within the proprietary IT infrastructure, thereby minimizing latency during data transmission and eliminating ongoing inference costs for cloud services.

Software Defined Automation and Hardware Selection
The cooperation follows the fundamental structural principle of software defined automation, in which software stacks and hardware components are consistently separated from each other. Steffen Winkler, Sales Management of the Automation and Electrification Solutions Business Unit at Bosch Rexroth, explains that ctrlX OS 'connects the automation world with modern computing and AI resources'. The standardization of the software foundation enables machine builders and original equipment manufacturers (OEMs) to flexibly adapt physical computing power to the respective automation task without having to develop a separate software environment for each hardware platform.

Edited by Maria Brueva, Induportals editor - adapted by AI.

www.boschrexroth.com