Higher-level policy
Objectives, modes, strategies
Akreon Decision Engine
Decision systems under uncertainty.
The layer between models and action—connecting state estimates, dynamics, objectives, constraints, uncertainty, feedback, and implementation limits.
Decision architecture
Layered decisions, closed around models and signals.
Useful decision systems do not live at one rate. Fast loops stabilize and reject disturbances. Optimization layers plan under constraints. Higher-level policies choose objectives, modes, and strategies as conditions change.
Models
Uncertainty
Constraints
Constrained optimization
Planning, allocation, MPC
Feedback control
Stabilization, estimation, safety filters
Operating system
Signals, dynamics, timing limits
Feedback control for constrained decisions
Akreon's work connects feedback, planning, and policy through models, constraints, uncertainty, and numerical software.
Fast loops
High-rate feedback control, safety filters, and state estimation for systems that need stable behavior under disturbance and measurement noise.
Mid-rate optimization
Trajectory optimization, model predictive control, planning, and allocation under explicit constraints and imperfect models.
Higher-level policies
Objective selection, mode logic, strategy updates, and learned or neural components where they fit within a broader model-based system.
Technical foundation
Akreon is grounded in systems where models are imperfect, measurements are noisy, constraints matter, and decisions must become reliable action.
Constrained optimization
Numerical formulations that make objectives, limits, feasibility, and tradeoffs explicit.
Control and estimation
Feedback, observers, filters, and decision loops built around uncertain state and imperfect signals.
Physics-informed modeling
Models that preserve the structure needed for reasoning about dynamics, limits, and operating envelopes.
Nonlinear and hybrid systems
Methods for systems with switching behavior, saturation, contact, modes, or nonconvex dynamics.
Simulation and validation
Testable artifacts, scenario analysis, and closed-loop evaluation before operational use.
Public work
The Library organizes thesis notes, tutorial arcs, rendered notebooks, videos, and technical notes around the recurring structure of decision problems.