Our
Technologies
Stratos Perception develops patented AI technologies that advance physical intelligence across some of the most consequential engineering domains of this decade. Our technologies are grounded in physics and mathematics, and are designed to operate where conventional methods cannot.
​
We currently offer three distinct and complementary technologies, each addressing a fundamental barrier in physical AI, autonomous systems, and space infrastructure.
ACI™ — Artificial Cerebellar Intelligence
ACI™ is Stratos Perception's patented solution to the fundamental sensory gap in physical AI. By approximating the cerebellum, ACI™ provides physical AI systems the autonomic comprehensive estimation layer they have been missing for interpreting the continuously changing stochastic real world.
​
ACI™ can infer 100+ hidden time-varying physical parameters in real time from minimal sensor data, in systems and environments where conventional estimators diverge or fail. It trains entirely on synthetic data, requiring no failure histories or additional sensors, and operates at control system frequencies.
​
Applications span autonomous robotics, spacecraft navigation, fusion reactor control, predictive maintenance, and industrial systems.
​
HFIT — Heterogeneous Flow Interpreter Transducer
The HFIT is Stratos Perception's patented AI technology for real time two-phase flow metering. Developed under NASA SBIR funding and validated on microgravity flow data, it combines machine vision with theoretical flow physics to determine flow velocity, acceleration, flow regimes, void fraction, thermal performance, and other flow conditions that conventional electronic and magnetic sensors cannot resolve.
​
Acceleration field agnostic, the HFIT operates accurately in space, in mobile systems, and in terrestrial environments alike. Primary applications include thermal management for space systems, terrestrial AI data centers where two-phase cooling is rapidly becoming the standard for managing extreme heat flux demands of high density GPU workloads, and the emerging concept of AI data centers deployed in space, where microgravity two-phase flow metering is a core unsolved challenge that HFIT is uniquely positioned to address.
​
Future research will combine HFIT with ACI™ to enable intelligent real time control of two-phase flow systems in space and terrestrial environments.
​
CAVIC — Contextualized Automated Vehicle Intelligent Controller
The CAVIC is a published AI control architecture introduced by Stratos Perception in 2020 for achieving safe, nuanced autonomous vehicle operation in populated, unpredictable environments.
​
Conventional autonomous vehicle control flows perception directly into decision and control systems, creating brittleness when novel or unanticipated perception events are encountered. CAVIC addresses this by abstracting perception into layered, continuously updated context states that are designed to correspond to human interpretable safety priorities and situational awareness, and that map to driving actions via a recurrent neural network.
​
The context state abstraction approach provides an interpretable and auditable control architecture, properties that are increasingly consequential as autonomous systems seek regulatory approval for unsupervised operation in public environments.
​
Stratos Perception continues to develop CAVIC as an active research track.
​
