The relentless drive to integrate sophisticated artificial intelligence directly into devices at the network’s edge has created a significant engineering challenge, demanding solutions that merge immense processing power with robust security in an ever-shrinking physical footprint. Addressing this critical need, Tria Technologies has introduced the OSM-LF-IMX95, a compact and powerful module meticulously designed for the next wave of advanced embedded systems and edge AI applications. Based on NXP Semiconductors’ state-of-the-art i.MX 95 applications processor, this module adheres to the Open Standard Module (OSM) Specification 1.2, ensuring a high degree of interoperability and future-proofing. Despite its diminutive 45 × 45 mm size, the module is engineered for direct soldering onto a carrier board. This strategic design choice empowers system designers to develop highly cost-optimized products by eliminating the need for expensive connectors, thereby streamlining the manufacturing process and accelerating the journey from concept to market-ready product.
A Convergence of Performance and Fortified Security
At the heart of the OSM-LF-IMX95 lies a formidable processing architecture designed to handle complex, real-time workloads. The module is equipped with a six-core Arm Cortex-A55 processor capable of running at speeds up to 2 GHz, providing the computational muscle for demanding applications. This is complemented by a pair of real-time processors—an 800 MHz Arm Cortex-M7 and a 333 MHz Arm Cortex-M33—which are ideal for deterministic control tasks. The processing power is further amplified by a suite of specialized hardware, including NXP’s eIQ Neutron neural processing unit (NPU) for efficient AI model execution, an Arm Mali GPU for advanced graphics, a 4K video processing unit, and a newly developed image signal processor. This potent combination enables the module to perform sophisticated AI-accelerated vision tasks directly at the edge, such as intricate object recognition in factory settings, predictive maintenance analysis on industrial machinery, and real-time production line monitoring. Security is a cornerstone of the module’s design, featuring NXP’s EdgeLock Secure Enclave subsystem. This integrated hardware security module simplifies the implementation of critical security functions, offering secure boot, advanced cryptography, comprehensive key management, and a secure pathway for over-the-air software updates, which are essential for low-power edge platforms in industrial automation, robotics, and vision systems where functional safety and data integrity are non-negotiable.
Extensive Connectivity for Advanced System Integration
The module’s comprehensive design extended beyond raw processing power to include a versatile array of connectivity and memory options, enabling its integration into a wide range of complex systems. It offered support for up to 16 GB of high-speed LPDDR5 SDRAM and up to 256 GB of onboard eMMC flash storage, providing ample resources for both operating systems and large AI models. The inclusion of numerous high-speed interfaces ensured seamless data transfer and communication with peripherals. These interfaces included a flexible SerDes lane supporting up to 10G Ethernet, dual standard Gigabit Ethernet ports for robust network connectivity, USB 3.0 for high-speed data transfer, and a PCI Express Gen 3 lane for connecting high-performance expansion cards. For embedded vision applications, the module provided dual MIPI CSI-2 camera interfaces and a MIPI DSI display output. Furthermore, its support for the CAN-FD protocol made it an ideal choice for automotive and industrial control networks. Tria Technologies provided substantial development support with a dedicated starter kit and a Yocto-based Linux board support package, with options for Android 12 and Windows 10 IoT Enterprise available to meet diverse project requirements.
