Dr Dr Oliver Faust

CSP4CMSIS: Bridging Formal Methods and AI at the Edge

Welcome to the official portal for the CSP4CMSIS ecosystem. Based in Cambridge, this project focuses on the implementation of Communicating Sequential Processes (CSP) for ARM-based embedded systems, providing a mathematically sound foundation for high-integrity AI deployment.

The Vision: Formal-AI-Edge (F.A.E.)

In the era of “AI at the Edge,” the industry faces a Safety Wall. Traditional “Super-loop” architectures cannot guarantee the deterministic behavior required for safety-critical systems when integrated with non-deterministic AI inference.

CSP4CMSIS solves this by providing a zero-heap, static-memory mapping of CSP primitives directly onto ARM CMSIS-RTOS and FreeRTOS.

Core Formalisms

We leverage the power of process algebra to model system behavior:


Documentation

Access the core technical resources for designing safe process networks:


Technical Demos & Repositories

Explore our implementation across the ARM ecosystem:

Target Hardware Complexity Repository Link
Nucleo-F401RE Entry Level View Repo
Nucleo-G474RE Intermediate View Repo
Cortex M55 / Ethos NPU Flagship / AI View Repo
DISCO-L475VG-IOT01A (B-L475E-IOT01A) Intermediate View Repo

2026 Funding Call: TechLocal AI Professional Accelerator

We are currently seeking partners for the DSIT TechLocal AI Professional Degree and Traineeship Accelerator.

The Objective: To develop a Level 7 (Masters equivalent) curriculum focused on “Certifiable AI.” We aim to provide graduates with the skills to deploy AI on ARM Ethos NPUs using formally verified process networks.

Calling ARM & Cambridge-based Industry Partners: If you are interested in supporting this traineeship model or integrating these formal methods into your safety-critical workflow, please get in touch.

Contact Oliver Faust via GitHub