Back to the portfolio

Reinforcement learning framework with embedded optimizer

Computing, Software & AIRobotics & Autonomous SystemsConceptPatent pending

This technology combines two established AI techniques—reinforcement learning and optimization solvers—into a single decision-making system. Rather than having a neural network learn everything from scratch, the system uses a prediction module to forecast future conditions and an optimization module to figure out the best action given those forecasts, while a reinforcement learning agent tunes the parameters that govern that optimization. The result is an AI agent that can make sequential decisions in complex environments more reliably and interpretably than pure neural network approaches. Think of it as giving an AI a built-in calculator it learns to configure, rather than asking it to do all the math implicitly.

What you could build

A middleware framework or software library that lets engineers drop an RL-plus-optimizer agent into industrial control, logistics, or energy-management systems—sold to automation software vendors or industrial AI platform companies that need auditable, constraint-aware decision engines.

Who in Virginia should care

Defense contractors and government agencies in Northern Virginia (e.g., DARPA-adjacent AI programs, logistics optimization for DoD supply chains) would find interpretable RL relevant to autonomous systems certification.

Readiness: Concept

Concept — described but not yet demonstrated. Lab validated — supported by experimental results in the patent. Prototype likely — the text describes a built, working embodiment.

Readiness is inferred from the patent text, not from a lab visit.

The record

Inventors
Ming Jin
Filed
Patent pending — filed March 9, 2023
Status
Application
Publication number
US20230289612A1

Ready to talk?

Virginia Tech Intellectual Properties handles licensing for this technology.

VTIP contact coming shortly

Prosim summaries are generated from public patent text and are not legal advice.