Romanian Journal of Information Technology and Automatic Control / Vol. 35, No. 3, 2025

Accelerating intelligent vehicle vision: A hybrid Python-Rust architecture with partial-blocking inter-process communication

Dávid SZILÁGYI, Răzvan FILEA, Kuderna-Iulian BENȚA

Abstract:

Modern autonomous systems require high-throughput, reliable vision pipelines with real-time guarantees. This paper presents a hybrid Python-Rust architecture that leverages synchronized shared memory and introduces a novel partial-blocking inter-process communication (IPC) model to decouple perception modules and enable parallelism without GIL-induced bottlenecks. In order to ensure a consistent data handling across asynchronous pipelines, the proposed system embeds lightweight runtime verification through watchdogs and health monitoring, and avoids redundant memory copying through efficient shared buffers. Across embedded and desktop platforms, this system delivers a fourfold speedup over naive multiprocessing while reducing the memory and processing overhead. The system is evaluated under various runtime configurations and demonstrates real-world applicability on a 1:10 scale autonomous vehicle. Its architecture provides a scalable foundation for safety-critical, real-time perception pipelines for robotics applications.

Keywords:
Intelligent Vehicles, Real-Time Vision, Shared Memory IPC, Python-Rust Integration, Partial- Blocking Communication.

View full article:

CITE THIS PAPER AS:
Dávid SZILÁGYI, Răzvan FILEA, Kuderna-Iulian BENȚA, "Accelerating intelligent vehicle vision: A hybrid Python-Rust architecture with partial-blocking inter-process communication", Romanian Journal of Information Technology and Automatic Control, ISSN 1220-1758, vol. 35(3), pp. 101-116, 2025. https://doi.org/10.33436/v35i3y202508