FogHorn Lets SIs Build Offline Machine Learning Apps For Android

FogHorn Systems is using its expertise in developing machine learning solutions for edge gateways and extending it to the world of Android.

The Sunnyvale, Calif.-based edge intelligence vendor announced the launch of Lightning Mobile, an edge computing solution that brings its machine learning and streaming analytics capabilities to ruggedized Android smartphones and tablets for the industrial market.

[Related: Falkonry's Machine-Learning Edge System Needs No Data Scientists]

Ramya Ravichandar, FogHorn's vice president of product management, told CRN that Lightning Mobile can run completely offline, meaning that all of the computing happens on the device, which is necessary for industrial companies who rely on operators working in low-connectivity environments.

"We're really opening up the expanse in terms of going after use cases that require mobility but don't require connectivity," she said. Targeted verticals include manufacturing, oil and gas, transportation and healthcare, according to the company.

FogHorn is taking more of an app-driven approach with Lightning Mobile than it does with its flagship product, the Lightning edge intelligence platform, which provides the means for ingesting, enriching and analyzing data in edge gateways for use cases like predictive maintenance and condition monitoring. For Lightning Mobile, the company will focus on developing use case-driven apps while allowing system integrators to do the same.

Keith Higgins, FogHorn's vice president of marketing, said Lightning Mobile will be a good channel sales play with plenty of opportunities for margins, especially for systems integrators who build high-value apps. The company will charge a license fee per instance of the software, according to the executive.

"If the systems integrator developed an app that is built with replicable use cases across a lot of companies in the same industry, they can write once and sell many – just like consumer app stores today," he said.

To start, the company has developed two apps. The first is barcode scanning app that uses machine learning to reconstruct images of barcodes that aren't legible, which Ravichandar said can save operators valuable time, preventing them from rescanning a barcode or manually entering the numbers.

The second is a battery prediction app that uses adaptive learning to tell operators whether a device will run out of power while they're out in the field.

Ravichandar said Lightning Mobile can ingest, enrich and analyze any kind of data coming from an Android device, including images, audio, time-series data and sensor data, which can go a long way in improving the way operators at industrial companies work.

"It's opening up new opportunities and ways of leveraging machine learning at the edge," she said. "This liberates them in the sense of being able to make operational decisions in real time."