top of page

AI on the edge

Machine learning solutions for real-time sensor data processing on mobile, wearable, IoT, and edge devices.

edgeml_logo.jpg
Edge ML

With our Edge ML software suite and framework, we provide computationally efficient, real-time classification of time-series data on low-power devices to generate, log, and deliver actionable events at the edge, to local networks, or to the cloud.  

tablet_inertial copy.png

Real-time ML Ops

​

The low processing requirements of our Edge ML solution means that real-time Machine Learning Model Operationalization Management (MLOPs) can be done at the edge.  This allows users to train, generate, and test models on-the-fly without ever needing network connectivity.

Edge ML Applications

Solving real-world problems requires real-world data.  Edge ML has been applied to many applications, including human performance assessment, GPS-denied navigation, underwater diver navigation, and real-time search-and-rescue training.

Hardware on the Edge

Edge compute is a term that includes a broad range of devices, from Internet-of-Things (IoT) devices to laptop computers.  Our unique ML algorithms are not only optimized for wearable and mobile devices, but also run on Raspberry Pis, laptops, and various embedded compute modules.

What makes Edge ML different?

Our solution provides classification of real-time data on energy-constrained and computationally-constrained devices with end-to-end MLOPs performed remotely or on-device, with optional network connectivity.

Case Study: Pedometry

We compare our Edge ML solution to commercially available devices and tinyML EVKs for basic pedometry.  Running our solution on an Android smartphone, we attempt to trigger false positives while walking a marked course.

Extensible Framework

The Edge ML framework can support modern Artificial Intelligence (AI) and Machine Learning techniques. We can adapt Edge ML for your specifications, requirements, and hardware.

EdgeML-Detail
About Us

We have a mission to improve contextual data.

All sorts of innovative approaches to sensor data processing have been made in the past decade due to the ubiquity of low-cost sensors found in smartphones and Internet-of-Things (IoT) devices.  Today, contextual information about users has been used by health and fitness apps despite low accuracy of sensor fusion algorithms, often requiring internet connectivity and electrical infrastructure.

With Edge ML we aim to deliver user contextual data at the edge, enabling low-power, low-compute ML pipelines in commercial, industrial, and end-user settings.

About
bottom of page