System Architecture for Safety-Critical Image Processing
Mixed-criticality system design: image processing, object detection, and frontend distributed across FPGA, MPU, and PC — separated by safety level, optimized for performance and resources.
- 01
Starting Point
Safety-critical and non-critical functions on a shared platform — from real-time image processing to visualization. On top of that, tight performance and resource budgets across all system boundaries. Without a clear architecture, that means certification effort for the entire system and resource conflicts between the functions.
- 02
Solution
A system architecture with consistent separation by safety level. Each function runs where it's most efficient: deterministic image processing in the FPGA, object detection and sequence control on the MPU with embedded Linux (Yocto), visualization on the PC. The domains are decoupled via DMA interfaces — high throughput without CPU load. Safety-relevant decisions remain exclusively within the critical domain.
- 03
Result
Each component only needs to be certified at its required safety level — significantly less certification effort. Real-time and throughput requirements met, resources optimally used. A scalable platform for further product variants.