Bandwidth is finite.
Continuous video can overwhelm the links connecting distributed sensors to shared compute.
PHYSICAL AI Intelligence in motion
Move less data.
Feature intelligence
Move less data.
See more possibility.
The world is full of sensors. DayStar makes their data work harder.
Machine vision compression and sensor transport built to scale physical AI.
THE WORLD IN DATA
Feature intelligenceINFORMATION THAT MATTERS
Built for the physical world
Autonomous mobility
Industrial automation
Smart infrastructure
Defense & aerospace
THE CHALLENGE
Sensors move fast.
Intelligence needs to keep up.
More cameras. More autonomous systems. More data than networks can carry. Physical AI needs a better way to connect what machines sense with what they understand.
Edge resources matter.
Power, memory, and compute constraints shape what an autonomous platform can do.
Every detail has a purpose.
Machine vision needs the information that supports detection, tracking, and decisions.
THE DAYSTAR APPROACH
Intelligence at the edge.
Understanding across systems.
DayStar develops task aware sensor transport: extracting and compressing useful features close to the sensor, then connecting them to the compute that completes the picture.
01 / CLOSE TO THE SENSOR
Begin the AI model where the data begins.
The edge runs the front of the model and produces compact features for the downstream task. This creates a practical way to share the work between a resource constrained device and more capable compute.
FEATURE CODING FOR MACHINES
A smaller stream.
A bigger opportunity.
Our first offering brings MPEG Feature Coding for Machines (FCM) into real time, with implementations demonstrated on visible, thermal, and low-light imagery.
18–25×
Less bandwidth
Compared with H.265 at matched detection accuracy.
2–8×
Detection accuracy
At the same bitrate in DayStar’s reported evaluations.
30+ fps
Real-time processing
Demonstrated on NVIDIA Orin.
Company-reported results. Performance depends on model, imagery, hardware, task, and configuration. Request benchmark methodology.
FROM RESEARCH TO IMPLEMENTATION
Built to fit your system.
A development path from flexible software integration to optimized hardware.
SOFTWARE
SDK & API
Integrate FCM into existing applications across various platforms
MODEL OPTIMIZATION
AI toolchain
User interface and engine for optimized deployment, adapting models to specific use case
HARDWARE ROADMAP
Implementations in Silicon
A path to hardware acceleration, with IP blocks designed for embedded platforms and future processor integration.
ONE FOUNDATION. MANY POSSIBILITIES.
Where perception
meets the real world.
From a single sensor to a fleet of autonomous systems, efficient machine vision creates room to do more.
AUTONOMOUS MOBILITY
Help every vehicle see further.
Connect cameras, onboard intelligence, and shared compute across autonomous vehicles and fleets. Compact feature streams can support perception while easing the demands of moving and retaining video.
- Autonomous vehicles and robotaxis
- Commercial fleets and logistics
- Collaborative perception
BEYOND VIDEO
A path to
collaborative perception.
FCM is the starting point. Our roadmap extends task aware transport to more sensors and to a shared view across platforms.
- 01
Feature coding
Real-time machine vision compression and an implementation toolchain.
- 02
Task aware prioritization
Deliver the most useful information first as available bandwidth changes.
- 03
More sensor types
Extend the approach toward radar, RF, acoustic, and other sensor inputs.
- 04
Shared perception
Connect information across platforms so systems can build a richer picture together.
THE PEOPLE BEHIND DAYSTAR
Deep research.
Real-world experience.
DayStar brings together expertise in video compression, AI, standards, intellectual property, and commercial deployment. Our team connects foundational research with the systems that put it to work.
Rob Moore
Founder, CEO & President
Experienced founder, engineer, patent lawyer. Founder of OP Solutions. Sold legacy OPS codec portfolio in 2026 for published amount of $37M.
Michael Strauss
Technology & Strategy
More than 30 years of experience commercializing technology across commercial and defense markets.
Dr. Velibor Adzic
Director of Research
Video compression researcher and named inventor on more than 100 U.S. patents.
Dr. Hari Kalva
Senior Researcher
Florida Atlantic University department chair, National Academy of Inventors Fellow, and named inventor on more than 130 U.S. patents.
Donald “Chee” Gansberger
Defense Liaison & Subject Matter Expert
Experience across Air Force special operations, DIUX, AFWERX, and defense technology programs.
Henk Vliestra
Chief Operating Officer
More than 30 years of commercial and defense experience, including serving as COO of Magic Leap.
A CLOSER LOOK
Questions,
answered.
What is Feature Coding for Machines?
FCM represents information used by a machine vision model as compressed features. It supports splitting inference between an edge device and downstream compute, with efficiency evaluated against the intended machine task.
How does DayStar work with existing platforms?
DayStar’s implementation approach includes a software SDK and API, an AI toolchain, and a hardware acceleration roadmap. An engagement starts with your sensors, models, hardware, network constraints, and performance requirements.
Which applications can benefit?
Applications include autonomous mobility, industrial automation, surveillance, infrastructure monitoring, and defense systems—particularly where cameras and compute are separated by limited network capacity.
Can we evaluate the technology?
Yes. Contact the team to discuss a live demonstration, benchmark methodology, or an evaluation using your own application requirements.
Contact DayStarBUILD WHAT COMES NEXT
Let’s put your
sensor data to work.
Explore a demonstration, a technical evaluation,
or a collaboration with DayStar.