Today’s autonomous or augmented applications require a safe and robust compute framework. CoreAVI offers a suite of algorithms and libraries, developed to the most stringent safety standards, used to facilitate advanced compute applications. CoreAVI’s Vulkan-based graphics and compute driver provides the foundation for safety critical GPU Compute. GPU Compute is the application of a modern GPU’s data-level parallelism through many-core SIMD engines to solve high volume computations. It is the basis for modern data processing, machine learning and computer vision systems.
Why Vulkan?
Vulkan SC is a next-gen safety critical graphics and compute API introduced by the Khronos group. Built from the ground up for parallel processing, Vulkan is a revolutionary, high efficiency, open standards API that enables wide, flexible, low level access to modern GPUs. It provides enhanced functionality with increased performance and flexibility. Vulkan not only reduces CPU workload, but also provides multi-threading capabilities, and moves management of some functions to the application layer allowing more control with less overhead resulting in significant performance gains. Vulkan is suited for a wide variety of GPU compute applications currently supported by OpenCL or CUDA.
Use Cases
Safe Autonomous Vehicles
As we march towards the inevitable mass deployment of autonomous vehicles, for truly pervasive adoption into our everyday lives to occur we have to trust in the safety of that vehicle whether as a passenger, a driver of a vehicle in close proximity or as a pedestrian. Huge investments are going into the development of the control systems and the AI models that will sense and perceive the environment around the vehicle before it decides what actions should be taken. These neural network-based designs will execute on complex heterogeneous systems that must have safety ‘in mind’ right from the outset of the design for a truly safe system to be deployable and meet the rigorous standards and expectations. For the developer to keep costs under control, these safe neural networks should be based on open standards so they can be safely deployed across a wide range of underlying hardware architectures with minimized rework.
In-Air Refuelling
In-air refuelling involves the transfer of fuel from one airborne aircraft to another. This capability requires the height of precision, with alignment kept between the tip of the boom and the receiver receptacle as both aircraft move at high speeds. One false calculation has the potential for disaster. Compute algorithms used to calculate the distance between refuelling components, as well as computations required for connection, disconnection, and fuel transfer all need to be safety critical to ensure a successful refuelling. CoreAVI’s Vulkan compute capabilities are currently used by Airbus Defence & Space in their automatic refuelling program.
Industrial Robots
Traditionally, factory and warehouse walls have served as a dam holding back the progress of autonomous technologies. Industrial robots have been safely caged in yellow steel enclosures to protect human workers from heavy, high-speed robotic arms with rudimentary safety systems shutting down the robot before a human can venture inside the cage. However, newer collaborative robots, where the barriers between humans and robots have been removed, require the complexity of compute safety technologies to grow rapidly such as safe path planning and vision systems to safety detect human proximity. The algorithms working inside industrial robots must adhere to safety critical standards in order to keep their human counterparts safe in the workplace.
Solution
CoreAVI’s Platform for Safety Critical Applications (PSCA) is built on VkCore SC, our Vulkan graphics and compute driver, and includes all the elements needed to build safety critical compute applications. With PSCA, customers can obtain all the building blocks – including hardware IP, semiconductors, graphics and compute drivers, libraries, firmware, and even development hardware – right from CoreAVI. Having a single source for all these pre-integrated components reduces risk, speeds time to certification and product launch, and prevents integration challenges.
Click the boxes below to view our products:
