View The Document

Accession Number:

AD1209329

Title:

Model Aware Reinforcement Learning

Author(s):

• Kamalapurkar, Rushikesh
• Bell, Zachary I.

Author Organization(s):

• Oklahoma State University

Report Date:

2023-05-31

Abstract:

The overall goal in this proposal was to develop a novel model-aware RL (MARL) framework for nonlinear systems incontinuous time and space specifically focused on mitigating large modeling errors and maintaining closed-loop stabilityduring the learning phase. The original scope of work was to focus on the development of model-aware RL methods that utilizeparametric models of the environment, are robust to modeling errors, and can adapt online to changing models and objectives.Data-driven adaptive estimation techniques were proposed to achieve online model estimation. Novel real-time modelvalidation methods were proposed to gauge the quality of the estimated models. Development of fall-back policies wasproposed to achieve robust learning in the presence of inaccurate models. In addition, the development model-aware RLmethods that utilize non-parametric models such as Gaussian Processes (GPs) was proposed along with the use of theconfidence bounds obtained for the GPs to guide model-aware virtual exploration. The development of model-aware RLtechniques that utilize local parametric and non-parametric models was also proposed to synthesize locally optimal policies

Document Type:

Conference:

Journal:

Pages:

19

File Size:

1.96MB

Contracts:

Grants:

Descriptors:

Identifiers:

• Model-based reinforcement learning

Communities of Interest:

• Materials and Manufacturing Processes

Distribution Statement:

Approved For Public Release

View The Document