EtherX AI

Neuromorphic Vision Intelligence

See FASTER.
Think at the EDGE.
Act in REAL TIME.

EtherX builds complete event-based vision systems, from AI algorithms to edge hardware, for real-time detection, tracking, and motion analysis of satellites, debris, drones, and aerial entities.

Neuromorphic Vision Intelligence

EtherX.AI

Neuromorphic Vision To Real-Time Drone Detection And Tracking

The next generation of intelligent machines will view the world differently! They will respond only to the dynamics of the world.

At EtherX.AI, we are building next generation intelligent perception systems using neuromorphic/event cameras that capture visual change asynchronously at extremely fine temporal resolution. We are putting this technology to work on one of the most demanding perception challenges today, namely, real-time detection and tracking of small, fast-moving drones.

Conventional cameras capture complete images at fixed intervals, even when most of the scene remains unchanged. A neuromorphic event camera works very differently. Each pixel responds independently when it detects a change in the corresponding intensity, producing a sparse and asynchronous stream of visual events.

For small aerial targets, this is a powerful advantage. A drone may occupy only a few pixels on an image sensor, accelerate rapidly, change direction abruptly, or move significantly between consecutive video frames. A neuromorphic event camera does not need to wait for the next frame. It captures change dynamically as it happens.

Event based tracking

At EtherX.AI, we have developed a complete neuromorphic processing pipeline that transforms asynchronous neuromorphic data into real-time drone detection and continuously updated target trajectories. The complete perception pipeline operates at millisecond-scale latency, enabling rapid response to fast target motion. Our system maintains robust, uninterrupted tracking under challenging conditions, including sun-glare, rain, multiple drones, and noisy environments

Our neuromorphic detection and tracking approach outperforms the state-of-the-art techniques.

Event Sensor

Neuromorphic
Processing Pipeline

Localization and Tracking

The pipeline is being optimized, quantized and deployed directly on edge hardware, reducing latency, bandwidth and dependence on cloud computing. Our goal is to build intelligent systems that are compact, power-efficient and ready for real-world deployment. This enables a new generation of intelligent systems across high-speed and edge-sensing applications.

Counter UAS

Drone detection and tracking

Sense and Avoid

Airborne collision awareness

Critical Infrastructure

Perimeter and asset monitoring

High Speed Robotics

Fast motion perception

Persistent Surveillance

Efficient edge monitoring

Dynamic Target Tracking

Responsive trajectory estimation

As seen in these demo videos, the conventional RGB camera struggles to maintain reliable drone detection and tracking under challenging illumination due to its limited dynamic range and frame-based acquisition. In contrast, the neuromorphic camera provides high dynamic range and microsecond-scale temporal resolution, enabling more robust and continuous tracking under strong contrast, shadows, and rapid motion.

RGB Camera
Neuromorphic Camera

Hardware Platform

Meet HOMI

HOMI is our end-to-end EdgeAI platform, combining a high-speed event sensor with FPGA compute and our in-house AI accelerator in a compact board designed for deployment in space, aerial, and defense systems.

10x5cm

Compact form factor

Sub-ms

Ultra-low latency architecture

On-device

Full edge inference - no cloud

Multi-task

Detection, tracking & classification - simultaneously

Front view

HOMI front view hardware

Top view

HOMI top view hardware

Event Sensor

Asynchronous, sparse event-driven output that captures only change in the scene with microsecond temporal resolution.

FPGA

FPGA-based compute backbone with hardware-optimized preprocessing pipelines and headroom for multi-task deployments.

In-House AI Accelerator

Quantization-aware neural network accelerator optimized for sparse event data across detection, classification, and tracking.

Compact & Deployable

10 cm x 5 cm form factor designed for edge deployment in satellites, UAVs, ground stations, and embedded systems.

Algorithms built for event-based intelligence

01

Motion Estimation

Compute velocity vectors in real time from sparse event streams for trajectory prediction at speeds frame cameras cannot resolve.

02

Object Detection & Tracking

Continuously detect and track satellites, debris, drones, and aerial objects using event-based clustering and centroid trajectories.

03

Star vs. Satellite Classification

Distinguish satellites from stars and clutter in dense fields using AI models trained on event-based motion signatures.

04

Collision Avoidance & Prediction

Run real-time trajectory analysis and motion prediction to estimate collision risk and enable early warning.

Why event cameras outperform frame-based

Microsecond temporal resolution.

No motion blur and no missed fast events.

Sparse, low-power data.

Only changing pixels generate data.

High dynamic range.

Handles dark space and sunlit objects without saturation.

Edge-native.

Full inference near 1 ms latency with ultra-low power.

FeatureFrame-basedEvent-based
Latency~33ms~1ms
Temporal res.~30 fps>1M events/s
Data volumeFull framesSparse events
Motion blurSignificantNone
Power drawHighUltra-low
Dynamic range~60 dB~120 dB
Edge inferenceCloud-dependentOn-device

Built for any domain where speed matters

HOMI and our algorithm stack are domain-agnostic. If it moves fast and you need to understand it in real time, we can track it.

Space Domain

Track satellites and orbital debris in LEO, detect tumbling behavior, support collision avoidance research, and analyze object attitude.

Aerial & Drone

Detect and track UAVs, drones, and low-altitude flying objects in real time even in cluttered visual environments.

Defense & Security

Enable autonomous airspace surveillance with ultra-low-latency threat detection and on-device inference.

Research & Development

Use a full-stack neuromorphic testbed from event preprocessing to on-device AI with flexible latency/accuracy studies.

Build smarter vision systems with us

Whether you are working in space operations, defense, aerial robotics, or research, we bring neuromorphic hardware and AI algorithms to turn raw events into real-time intelligence.

Contact us