Driver enables safety critical graphics in autonomous systems -

Driver enables safety critical graphics in autonomous systems

CoreAVI said VkCore SC supports multiple GPU architectures allowing developers to migrate safety critical software stack seamlessly across different silicon implementations.

CoreAVI has developed a safety critical graphics and compute driver called VkCore SC, aligned with the new Vulkan SC API from the Khronos Group, to enable deployment of safety certifiable software stacks on autonomous systems that interact directly with people.

Whether for collaborative robots, driverless transportation systems, or autonomous systems in factories and warehouses, the key challenge for engineers is to move these from prototype to large scale deployment while ensuring a safety critical implementation of graphics and compute drivers.

The company said VkCore SC offers the option for ISO 26262 ASIL D, RTCA DO-178C/EUROCAE ED-12C certification up to DAL A, and IEC 61508 SIL3 certification. The driver supports multiple GPU architectures allowing the developer to migrate their safety critical software stack seamlessly across different silicon implementations, increasing flexibility, scalability and reducing overall total cost of ownership for safety systems. The driver forms the basis of CoreAVI’s platforms for safety critical applications, addressing the needs of safe graphics and compute applications across various market verticals.

CoreAVI’s platform for safety critical applications. (Source: CoreAVI)

While VkCore SC is the driver, the company’s VkCoreVX SC is its safety critical implementation of OpenVX1.3, the Khronos Group’s OpenVX industry standard API that provides a feature set for implementing and deploying neural networks in safety critical environments. This is the backbone of CoreAVI’s artificial intelligence and computer vision platform for safe autonomy, built on top of the safety critical Vulkan SC implementation, and providing both graphics and compute capabilities within the same safety critical framework.

The computer vision subset provides algorithms for performing pre-processing and post-processing tasks on sensory data streams. The collection of algorithms and neural network inferencing engines provide what the company said is a true safety certifiable software stack that facilitates powerful computer vision executing on modern GPUs.

Related Content:

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.