DDC-I Announces Klepsydra Artificial Intelligence and Edge Processing for Deos DO-178C Safety-Critical RTOS
The integration of Klepsydra and Deos provides a compact, modular, safety-critical platform with high-performance AI and edge data processing, making it ideal for a broad range of mission-critical applications requiring sensor fusion, situational awareness, and other AI applications for both manned and autonomous vehicles (air, ground, and water).
“Deos and Klepsydra make certified AI at the edge not just a possibility, but a reality,” said Bob Morris, president/CEO at DDC-I. “Now, developers requiring advanced AI, edge processing, and cybersecurity capabilities have a high-performance, modular, lightweight, reusable platform with secure partitioning that can be readily certified, ported, and upgraded with the latest technology for the most demanding mission-critical applications.”
“We are thrilled to collaborate with DDC-I to bring Klepsydra’s AI and edge computing technology to the Deos real-time operating system,” said Pablo Ghiglino, CEO, Klepsydra Technologies. “This integration marks a significant milestone in making high-performance, safety-critical AI available for the most demanding mission-critical applications. By combining Deos’ robust safety credentials with Klepsydra’s efficient and deterministic data processing, we’re enabling a new generation of intelligent systems in avionics, autonomous mobility, and Beyond.”
Klepsydra AI is a cutting-edge software framework for edge AI applications. Delivering high-efficiency AI inference, Klepsydra AI employs a lightweight, independent design without third-party dependencies, making it robust and adaptable for next-generation edge computing.
Klepsydra provides a safety-certified, cybersecurity-protected AI software stack that is compliant with ECSS software standards and compatible with standard AI frameworks such as PyTorch and SensorFlow. Utilizing RTEMS (Real-Time Executive for Multiprocessor Systems) real-time services, Klepsydra features a native API that provides access to Deos SafeMC multicore capabilities, and a streaming distribution optimizer that enables developers to configure the framework for maximum data throughput.
Klepsydra’s acceleration technology combines two-dimensional parallelization with lock-free programming to deliver a flexible and efficient inference engine, achieving up to 4x lower latency, 10x higher throughput, and 50% reduced power consumption than competitive solutions. Its autotuning tool optimizes AI inference for various constraints, while a unified API eliminates processor lock-in, enabling greater flexibility and improved performance across hardware platforms.
Deos is a safety-critical embedded RTOS that employs patented cache partitioning, memory pools, and safe scheduling to deliver higher CPU utilization than any other certifiable safety-critical COTS RTOS on multicore processors. First certified to DO-178 DAL A in 1998,
Deos provides FACE® Conformant OSS Safety Base and Safety Extended Profiles that feature hard real-time response, time, and space partitioning, with support for RateMonotonic, ARINC-653, and POSIX interfaces.
SafeMC™ technology extends Deos’ advanced capabilities to multiple cores, enabling developers of safety-critical systems to achieve best-in-class multicore performance without compromising safety-critical task response and guaranteed execution time. SafeMC employs a bounded multiprocessing (BMP) extension of the symmetric multiprocessing architecture (SMP), safe scheduling, and cache partitioning to minimize cross-core contention and interference patterns that affect the performance, safety criticality, and certifiability of multicore systems. These features enable avionics developers to address issues specified by AC/AMC 20-193 multicore objectives that could impact the safety, performance, and integrity of a software airborne system.