AXON Predict for VxWorks: Extending Edge Analytics to the Final Frontier

September 19, 2017 • Chad Boulanger

VxWorks is the most widely deployed RTOS in the world. It is used in Boeing’s 787 Dreamliner, Magneti Marelli Motorsport’s Formula 1 technologies, Siemens’ industrial PCs, Varian’s medical imaging solutions, Huawei’s international telecom networks, and a plethora of other innovative and far-flung embedded systems deployments. Perhaps most famously, VxWorks powers the NASA JPL Curiosity Rover on the planet Mars.

Today, WindRiver announced the “general availability” (GA) release of AXON Predict™ Analytics for VxWorks, its modular vendor-neutral real-time operating system.

So what value does AXON Predict add to such an esteemed and — literally — out-of-this-world RTOS? As the only analytics solution integrated with VxWorks, AXON Predict’s dynamic analytics engine enables developers to securely extend intelligence to the edge of networks and into connected devices and systems utilizing the RTOS. Predictive analytics, machine learning, and artificial intelligence capabilities are all unleashed so applications can consume critical data at the source of input and within high-volume streams in real time. This facilitates autonomous responses on devices without latency and without the need for constant cloud connectivity. Crucially, AXON Predict introduces minimal application impact with its small footprint, which is ideal for embedded systems and resource-constrained environments.

AXON Predict is enabled through Wind River® Workbench, the VxWorks Eclipse-based integrated development environment, so it’s easy to add analytic capabilities programmatically. Advanced scripts can be designed through a rich interactive interface that includes predefined pattern, computational, and learning templates, as well as an action framework allowing the analytics engine to execute customizable programmatic actions when specific conditions of interest arise. Pattern learning templates are deployed to the system to identify baseline normal behavior of end devices and, by extension, the patterns that indicate anomalies.

The integration presents an iterative development environment where patterns and analytics can initially be quickly defined and deployed to the edge engine, then refined over time as the historical picture necessary for developing custom predictive models is assembled. The dynamic nature of AXON Predict means that as new analytics scripts are developed, they can easily be deployed to the analytics engine through its programmatic API.

Integrating advanced analytics into VxWorks is a natural and necessary feature extension considering the explosive growth in IoT/IIoT implementations. Massive increases in data flows are the natural byproduct of this surge, resulting in a crushing need for to decentralize computation. Transporting all that data to the cloud for management, analysis, and decision-making is costly, inefficient, and far too slow. Real-time edge intelligence addresses all three problems and enables compelling new automation paradigms via the capture and immediate utilization of raw data at the source.

That’s what AXON Predict brings to the equation.

And that kind of power is indescribably valuable whether you’re trying to enhance operational efficiency in an existing IIoT deployment or create the next awesome robotics system for extraplanetary science and space exploration.