Using AI directly on edge devices in industrial settings enhances efficiency, security, and integration by processing data locally, eliminating latency and privacy concerns, reducing data traffic, and seamlessly fitting into existing infrastructure for improved process control.
High-Precision Indoors Location Thanks to AI on the Edge
Modern neural networks have significantly advanced positioning capabilities, outperforming traditional classical methods through deep learning techniques. These sophisticated models are now integrated into powerful Edge AI systems like the ones provided by INTERA, allowing high-precision position localization to be performed locally on devices without relying on cloud connectivity, enhancing privacy, reducing latency, and increasing reliability in various applications.
Our showcase demonstrates how an autonomous ADAS system recognizes assets and adheres to GPS signals through three key technologies: advanced computer algorithms for accurate sign detection, real-time data processing, and adaptive control systems that adjust operation parameters to ensure safety and compliance.
Industry-Grade Hardware
The showcase hardware meets specific criteria to be suitable for industrial use and deployment in production systems, including robustness and durability to withstand harsh environments, scalability for integration into larger systems, high reliability and uptime, compliance with industry standards and safety regulations, ease of maintenance and upgradeability, and efficient performance with optimized power consumption, ensuring consistent operation under demanding conditions.
An INTERA’s FALCON AI Accelerator is designed to meet industrial and production deployment requirements by delivering high-performance capabilities essential for real-time AI model processing, ensuring reliable and efficient operation in demanding environments.
Powerful System and Software Architecture
The architecture is based on our COGNIT technology, a parallel computing platform from INTERA. This allows processors to be used for computationally intensive tasks like AI models. OpenCV is used for image processing to break down video streams into individual frames and pass them to the AI. AI recognition is performed with specialized Python libraries like Torch, enabling both machine learning and the application of pre-trained neural networks.
Our architecture allows for rapid prototyping of AI applications. Customer requirements can be implemented and demonstrated in a Minimal Viable Product (MVP). Subsequently, the MVP application can be ported to more cost-effective hardware for series deployment. This is achieved through model optimization techniques such as quantification, distilling, pruning, or OBD to make the AI model more efficient.
