Have you ever been in a car with an on-board computer system that identifies stop signs? This is what is called “real-time image inferencing” in the artificial intelligence (AI) world, and it is amazing technology. Modern cars now have cameras that can stream video to computers that “score” each frame with a pre-trained AI model. Each frame is pre-processed, fed into the model, and the model identifies every stop sign it sees. And this whole process takes place in a fraction of a second (much faster than a human’s reaction time), improving passenger safety by an order of magnitude.

This type of AI model deployment is called “edge deployment” – computation that is done close to the source of the data, rather than in the cloud or in on-premises servers. And it is the best way to get low latency (i.e. very fast) processing of images, video, audio, or other large data sources. Unfortunately, the only way large automotive companies can do this now is with large teams of full-stack developers. But what if your company does not have these kinds of resources? How can you capture imagery from a fleet of edge devices? How can you build AI models and push them to these devices without hiring teams of developers to write mountains of complex computer code?

Altair is pleased to introduce its innovative solution to this problem: Altair® AI Edge™. It is a new hardware device designed for AI edge applications. And it integrates with both Altair® IoT Studio™ and Altair® RapidMiner® to make it easy to solve AI image challenges on the edge.

What is Altair AI Edge?

AI Edge is a single-board-computer (SBP) with a high-resolution, auto-focus camera all enclosed in one compact aluminum enclosure.  The computer is powered by a Broadcom BCM2711, Quad core Cortex-A72 (ARM v8) 64-bit SoC with 8GB of RAM and 64GB of on-board MicroSD storage. It only has one plug: a standard ethernet connection that provides both network connectivity and power via PoE. The camera has a 12MP sensor with auto-focus for amazing imagery. And the all-aluminum enclosure has a standard 3/8-16 (tripod) socket to make it easy to install in a variety of applications.

Why Altair AI Edge?

The key to AI Edge is its seamless integration with IoT Studio and the Altair RapidMiner Platform. AI Edge comes pre-provisioned with software that allows it to stream data directly to IoT Studio via MQTT or HTTPS. Powered by k3s, a lightweight implementation of kubernetes, IoT Studio’s Edge Compute Platform (ECP) manages the entire kubernetes cluster residing on AI Edge. Customers can plug hundreds of AI Edge devices into their network and they will instantly make a connection to IoT Studio. And all fleet management can be done from IoT Studio’s cloud-native interface. Once AI Edge devices are connected, users can build drag-and-drop workflows and machine learning models in Altair® AI Studio™ and push them instantly down to the entire fleet. Streamed real-time scoring can be viewed using Altair® Panopticon™, or consumed by external systems via REST API or MQTT.