Alessandro Salatiello

Title: Alessandro Salatiello and Generative AI-Based Stochastic Control

Introduction

Alessandro Salatiello is an inventor associated with Cambridge, Massachusetts, and Mitsubishi Electric Research Laboratories, Inc. His patent work addresses predictive control of mechanical systems under uncertain disturbances, using generative AI to model and anticipate possible conditions.

Latest Patents

Salatiello is named on 1 patent: “Stochastic Control Subject to Generative AI-Based Disturbance.” The invention describes a predictive controller that uses a deep generative decoder model to estimate a conditional probability distribution for latent representations of a disturbance, based on partial observations. It samples from that distribution and decodes the sample to generate predicted disturbance values over a time horizon. The controller then uses the probabilities of at least some predicted values when determining commands that change the mechanical system’s operating state.

Career Highlights

Salatiello’s listed patent reflects work at the intersection of stochastic control, predictive modeling, and generative AI. The approach is designed to help a controller account for uncertainty in disturbances affecting a mechanical system.

Collaborations

Salatiello’s listed coworker is Ye Wang. Both are associated with Mitsubishi Electric Research Laboratories, Inc., in connection with the patent information provided.

Conclusion

Alessandro Salatiello’s patent applies generative modeling to predictive control, offering a method for incorporating uncertain disturbances into mechanical-system decisions.