Results for “Reinforcement Learning” — Page 14

NASA NTRSDate not supplied

Discovery and Analysis of Rare High-Impact Failure Modes using Adversarial RL-Informed Sampling

Adaptive learning agents have tremendous potential to handle critical tasks currently performed by humans. Unfortunately, due to their complexity, it can be difficult to verify that these learning agents do not have critical failure modes. Standard verification and validation methods often do not apply directly to learning agents and Monte Carlo methods have difficulty covering even a small fraction of the state space, especially in multiagent systems or over long time horizons. To overcome this difficulty, we demonstrate an adaptive stress-testing method based on reinforcement learning of correlations that raise the probability of failure. This approach has three key properties: (1) it is able to find rare failure modes with far greater sample efficiency than Monte Carlo methods, (2) it can estimate the true probability of a failure mode despite the inherent bias in the learning method, and (3) it is capable of learning and resampling compact representations of multimodal failure spaces. These properties are important in practice as we need to find disparate failure modes while accounting for their actual relevance. This is a significant advantage over traditional adaptive stress testing methods that give abstract likelihoods of particular failure instances, but cannot estimate the probability of a broader failure mode. We test our algorithm on a simple problem from the aviation domain where an autonomous aircraft lands in gusty wind conditions. The results suggest that we can find failure modes with far fewer samples than the Monte Carlo approach and simultaneously estimate the probability of failure.

reinforcement learning↗

NASA NTRSDate not supplied

Discovery and Analysis of Rare High-Impact Failure Modes using Adversarial RL-Informed Sampling

Adaptive learning agents have tremendous potential to handle critical tasks currently performed by humans. Unfortunately, due to their complexity, it can be difficult to verify that these learning agents do not have critical failure modes. Standard verification and validation methods often do not apply directly to learning agents and Monte Carlo methods have difficulty covering even a small fraction of the state space, especially in multiagent systems or over long time horizons. To overcome this difficulty, we demonstrate an adaptive stress-testing method based on reinforcement learning of correlations that raise the probability of failure. This approach has three key properties: (1) it is able to find rare failure modes with far greater sample efficiency than Monte Carlo methods, (2) it can estimate the true probability of a failure mode despite the inherent bias in the learning method, and (3) it is capable of learning and resampling compact representations of multimodal failure spaces. These properties are important in practice as we need to find disparate failure modes while accounting for their actual relevance. This is a significant advantage over traditional adaptive stress testing methods that give abstract likelihoods of particular failure instances, but cannot estimate the probability of a broader failure mode. We test our algorithm on a simple problem from the aviation domain where an autonomous aircraft lands in gusty wind conditions. The results suggest that we can find failure modes with far fewer samples than the Monte Carlo approach and simultaneously estimate the probability of failure.

Validation↗

NASA NTRSDate not supplied

Discovery and Analysis of Rare High-Impact Failure Modes using Adversarial RL-Informed Sampling

Adaptive learning agents have tremendous potential to handle critical tasks currently performed by humans. Unfortunately, due to their complexity, it can be difficult to verify that these learning agents do not have critical failure modes. Standard verification and validation methods often do not apply directly to learning agents and Monte Carlo methods have difficulty covering even a small fraction of the state space, especially in multiagent systems or over long time horizons. To overcome this difficulty, we demonstrate an adaptive stress-testing method based on reinforcement learning of correlations that raise the probability of failure. This approach has three key properties: (1) it is able to find rare failure modes with far greater sample efficiency than Monte Carlo methods, (2) it can estimate the true probability of a failure mode despite the inherent bias in the learning method, and (3) it is capable of learning and resampling compact representations of multimodal failure spaces. These properties are important in practice as we need to find disparate failure modes while accounting for their actual relevance. This is a significant advantage over traditional adaptive stress testing methods that give abstract likelihoods of particular failure instances, but cannot estimate the probability of a broader failure mode. We test our algorithm on a simple problem from the aviation domain where an autonomous aircraft lands in gusty wind conditions. The results suggest that we can find failure modes with far fewer samples than the Monte Carlo approach and simultaneously estimate the probability of failure.

Validation↗

DOE OSTI2024

Coherency-Aware Learning Control of Inverter-Dominated Grids: A Distributed Risk-Constrained Approach

Here, this letter investigates the importance of integrating the coherency knowledge for designing controllers to dampen sustained oscillations in wide-area power networks with significant penetration of inverter-interfaced resources. Coherency is a fundamental property of power systems, where time-scale separation in frequency dynamics leads to clustered behavior among generators of different groups. Large-scale penetration of inverter-driven low inertia resources replacing conventional synchronous generators (SGs) can lead to perturbation in the coherent partitioning; hence, integrating such information is of utmost importance for oscillation control designs. We present the coherency-aware design of a distributed output feedback-based reinforcement learning method that additionally incorporates risk constraints to capture the uncertainties related to net-load fluctuations. The use of domain-aware coherency information has produced improved training and oscillation performance than the coherency-agnostic control design, hence proving to be effective in controller design. Finally, we validated the proposed method with numerical experiments on the benchmark IEEE 68-bus test system.

24 POWER TRANSMISSION AND DISTRIBUTION↗

NASA NTRS2014

Exact and Approximate Probabilistic Symbolic Execution

Probabilistic software analysis seeks to quantify the likelihood of reaching a target event under uncertain environments. Recent approaches compute probabilities of execution paths using symbolic execution, but do not support nondeterminism. Nondeterminism arises naturally when no suitable probabilistic model can capture a program behavior, e.g., for multithreading or distributed systems. In this work, we propose a technique, based on symbolic execution, to synthesize schedulers that resolve nondeterminism to maximize the probability of reaching a target event. To scale to large systems, we also introduce approximate algorithms to search for good schedulers, speeding up established random sampling and reinforcement learning results through the quantification of path probabilities based on symbolic execution. We implemented the techniques in Symbolic PathFinder and evaluated them on nondeterministic Java programs. We show that our algorithms significantly improve upon a state-of- the-art statistical model checking algorithm, originally developed for Markov Decision Processes.

Symbolic Execution↗

DOE OSTI2024

Artificial-intelligence-driven shot reduction in quantum measurement

Variational Quantum Eigensolver (VQE) provides a powerful solution for approximating molecular ground state energies by combining quantum circuits and classical computers. However, estimating probabilistic outcomes on quantum hardware requires repeated measurements (shots), incurring significant costs as accuracy increases. Optimizing shot allocation is thus critical for improving the efficiency of VQE. Current strategies rely heavily on hand-crafted heuristics requiring extensive expert knowledge. This paper proposes a reinforcement learning (RL)-based approach that automatically learns shot assignment policies to minimize total measurement shots while achieving convergence to the minimum of the energy expectation in VQE. The RL agent assigns measurement shots across VQE optimization iterations based on the progress of the optimization. This approach reduces VQE's dependence on static heuristics and human expertise. When the RL-enabled VQE is applied to a small molecule, a shot reduction policy is learned. The policy demonstrates transferability across systems and compatibility with other wavefunction Ansätze. In addition to these specific findings, this work highlights the potential of RL for automatically discovering efficient and scalable quantum optimization strategies.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

DOE OSTI2024

Model-Based Detection of Coordinated Attacks (DCA) in Distribution Systems

The fast-paced growth in digitization of smart grid components enhances system observability and remote-control capabilities through efficient communication. However, enhanced connectivity results in heightened system vulnerability towards cybersecurity risks in the cyber-physical power system. Coordinated cyber-attacks (CCA), when undetected, lead to system-wide impact in terms of large disturbances or widespread outages. Detecting CCA in the cyber layer is critical to thwart cyber-attacks in real-time before the attack impacts the physical system. The challenge of locating CCA stems from the complex grid dynamics, making it difficult to distinguish between normal operational variations and cyber-attack impact. CCA often employs multiple attack vectors targeting geographically distributed components, further complicating CCA identification. Existing research in intrusion detection is primarily focused on the transmission network and limited to detecting individual attacks. In this paper, a novel proactive DCA strategy is proposed for early detection of CCA by establishing correlations among distinct attack events through model-based reinforcement learning that utilizes abductive reasoning to conclude the attacker goal. The solution includes understanding the system model, learning the system dynamics, and correlating individual cyber-attacks to extract the attacker’s objective. The developed learning algorithm identifies the most probable attack path to reach the attacker’s objective by predicting the next attack steps. A DNP3-based cyber-physical co-simulation testbed is developed to test the proposed algorithm using the IEEE 13-node test feeder.

24 POWER TRANSMISSION AND DISTRIBUTION↗

NASA NTRS2019

Interplanetary Low-Thrust Design Using Proximal Policy Optimization

This paper aims to demonstrate a reinforcement learning technique for developing complex, decision-making policies capable of planning interplanetary transfers.Using Proximal Policy Optimization (PPO), a neural network agent is trained to produce a closed-loop controller capable of transfers between Earth and Mars.The agent is trained in an environment that utilizes a medium fidelity solar electric propulsion model and a real ephemeris model of the Earth and Mars. The results are compared against those generated by the Evolutionary Mission Trajectory Generator (EMTG) tool.

proximal policy optimization↗

DOE OSTI2024

Enhancing Distribution System Resilience: A First-Order Meta-RL Algorithm for Critical Load Restoration

The increasing frequency of extreme events and the integration of distributed energy resources (DERs) into modern grids have elevated the need for resilient and efficient critical load restoration strategies in distribution systems. However, the stochastic nature of renewable DERs, limited energy resource availability and the intricate nonlinearities inherent in complex grid control problem make the problem challenging. Although reinforcement learning (RL) and warm-start RL methods have shown promising results, their performance often falls short in rapidly adapting to new, unseen situations and typically requires exhaustive problem-specific tuning. To address these gaps, we propose a First-Order Meta-based RL (FOM-RL) algorithm within an online framework for adaptive and robust critical load restoration. By harnessing local DERs as the enabling technology, FOM-RL allows the RL agent to swiftly adapt to new unseen scenarios by leveraging previously acquired knowledge of different tasks. Experimental results provide evidence that proposed algorithm learns more efficiently and showcases generalization capabilities across diverse set of operational scenarios. Moreover, a rigorous theoretical analysis yields a tight sublinear regret bound, sensitive to temporal variability, with a task-averaged optimality gap bounded by O(VM+D*/(Tsquare root(M))). These results suggest that optimality improves with task similarity and an increased number of tasks M, reaffirming the efficacy and scalability of the proposed approach in addressing the complexities of critical load restoration in distribution systems.

complexity theory↗

DOE OSTI2024

From Sim to Real: A Pipeline for Training and Deploying Traffic Smoothing Cruise Controllers

Designing and validating controllers for connected and automated vehicles to enhance traffic flow presents significant challenges, from the complexity of replicating real-world stop-and-go traffic dynamics in simulation, to the intricacies involved in transitioning from simulation to actual deployment. In this work, we present a full pipeline from data collection to controller deployment. Specifically, we collect 772 km of driving data from the I-24 in Tennessee, and use it to build a one-lane simulator, placing simulated vehicles behind real-world trajectories. Using policy-gradient methods with an asymmetric critic, we improve fuel efficiency by over 10% when simulating congested scenarios. Our comprehensive approach includes reinforcement learning for controller training, software verification, hardware validation and setup, and navigating various sim-to-real challenges. Furthermore, we analyze the controller's behavior and wave-smoothing properties, and deploy it on four Toyota Rav4’s in a real-world validation experiment on the I-24. Lastly, we release the driving dataset, the simulator and the trained controller, to enable future benchmarking and controller design.

42 ENGINEERING↗

DOE OSTI2024

Design and optimization of a modular hydrogen-based integrated energy system to maximize revenue via nuclear-renewable sources

Here, this paper demonstrates a novel modular distributed framework that uses optimal energy-dispatching strategies to enable greater flexibility and profitability in nuclear-renewable integrated energy systems (NR-IES). Hydrogen is used as a commodity in this framework since its production can improve grid stability and system operational flexibility, decarbonize heavy industry, and create an additional revenue stream for electricity generators, particularly nuclear power plants with high operational expenses. The proposed solution addresses the challenges associated with merging multiple software and services from various domains by using functional mock-up units (FMU) to co-simulate diverse subsystems designed in various platforms. The tightly coupled integrated energy system (IES) is optimized to maximize revenue by utilizing the deep reinforcement learning (DRL) technique to make smart dispatching decisions based on variable electricity prices and the availability of renewable energy. Proximal policy optimization (PPO) algorithm is used in training and testing the DRL agent. Over a period of 120 days, the proposed hydrogen-based IES framework showed about 10% revenue boost compared to a non-hydrogen generating baseline IES while also providing an easily-adoptable framework which can help to improve the flexibility of future generation nuclear power plants.

29 ENERGY PLANNING, POLICY, AND ECONOMY↗

DOE OSTI2026

Optimal CO 2 storage management considering safety constraints in multi-stakeholder multi-site GCS projects: A Markov game perspective

Geological carbon storage (GCS) projects could involve a diverse array of stakeholders or players from public, private, and regulatory sectors, each with different objectives and responsibilities. Given the complexity, scale, and long-term nature of GCS operations, determining whether individual stakeholders can independently optimize their interests — or whether collaborative coalition agreements are needed — remains a central question for effective GCS project planning and management. To access large, high-quality storage resources, future GCS deployment may increasingly occur in geologically connected sites, where shared geological features such as pressure space and reservoir pore capacity can lead to competitive behavior among stakeholders. In this work, we propose a paradigm based on Markov games to quantitatively investigate how different coalition structures affect the goals of stakeholders. We frame this multi-stakeholder multi-site problem as a multi-agent reinforcement learning problem with safety constraints. Our approach enables agents to learn optimal strategies while complying with safety regulations. We present an example where multiple operators are injecting CO 2 into their respective project areas in a geologically connected basin. To address the high computational cost of repeated simulations of high fidelity models, a previously developed surrogate model based on the Embed-to-Control (E2C) framework is employed. Our results demonstrate the effectiveness of the proposed framework in addressing optimal management of CO 2 storage when multiple stakeholders with different objectives and goals are involved.

58 GEOSCIENCES↗

DOE OSTI2025

Sim2Real Autonomous Robotic Exploration [Poster]

Autonomous robots offer promising solutions for exploration in environments that are inaccessible or hazardous to humans. Despite this, physical training of such robots is often constrained by safety risks, high cost or limited accessibility. This project presents an end-to-end simulation to reality pipeline leveraging Nvidia Isaac Sim and Boston Dynamics' Spot to enable autonomous navigation in indoor environments. A reinforcement learning policy is first trained using Nvidia Isaac Lab to establish Spot's locomotion pattern. Virtual LiDAR sensors are then integrated to perform SLAM-based navigation using simulated odometry. Finally, the simulated navigation scheme is transferred to a physical Spot robot to inspect and record images of a real-world room by repeating the learnt trajectory. The proposed framework highlights the potential of scalable training in simulation and reliable deployment in physical environments. Future directions include dynamic trajectory generation in unseen and challenging environments and integration of environmental sensing like temperature, radiation or humidity via sensor and material simulation.

97 - MATHEMATICS AND COMPUTING↗

NASA NTRS2017

Implementation of a Space Communications Cognitive Engine

Although communications-based cognitive engines have been proposed, very few have been implemented in a full system, especially in a space communications system. In this paper, we detail the implementation of a multi-objective reinforcement-learning algorithm and deep artificial neural networks for the use as a radio-resource-allocation controller. The modular software architecture presented encourages re-use and easy modification for trying different algorithms. Various trade studies involved with the system implementation and integration are discussed. These include the choice of software libraries that provide platform flexibility and promote reusability, choices regarding the deployment of this cognitive engine within a system architecture using the DVB-S2 standard and commercial hardware, and constraints placed on the cognitive engine caused by real-world radio constraints. The implemented radio-resource allocation-management controller was then integrated with the larger spaceground system developed by NASA Glenn Research Center (GRC).

reinforcement learning↗

NASA NTRS2017

Implementation of a Space Communications Cognitive Engine

Although communications-based cognitive engines have been proposed, very few have been implemented in a full system, especially in a space communications system. In this paper, we detail the implementation of a multi-objective reinforcement-learning algorithm and deep artificial neural networks for the use as a radio-resource-allocation controller. The modular software architecture presented encourages re-use and easy modification for trying different algorithms. Various trade studies involved with the system implementation and integration are discussed. These include the choice of software libraries that provide platform flexibility and promote reusability, choices regarding the deployment of this cognitive engine within a system architecture using the DVB-S2 standard and commercial hardware, and constraints placed on the cognitive engine caused by real-world radio constraints. The implemented radio-resource allocation-management controller was then integrated with the larger spaceground system developed by NASA Glenn Research Center (GRC).

reinforcement learning↗

DOE OSTI2025

FIRE: A Failure-Adaptive RL Framework for Edge Computing Migrations

In edge computing, users' service profiles are migrated between edge servers due to user mobility. Reinforcement Learning (RL) frameworks have been proposed to do so, often trained on simulated data. However, existing RL frameworks overlook occasional server failures, which although rare, impact latency-sensitive applications like AR/VR and real- time obstacle detection. These rare failures, being not adequately represented in historical training data, pose a challenge for data-driven RL algorithms. We introduce FIRE, a framework that adapts to rare events by training a RL policy in an edge computing digital twin environment. We propose FIRE-ImRE, an importance sampling-based Q-learning algorithm, which samples rare events proportionally to their impact on the value function. FIRE considers delay, migration, failure, and backup placement costs across individual and shared service profiles. We prove FIRE-ImRE's boundedness and convergence to optimality. Next, we introduce novel deep Q-learning (FIRE-ImDQL) and actor critic (FIRE-ImACRE) versions of our algorithm to enhance scalability. Here, we extend our framework to accommodate users with varying risk tolerances of rare failure events. Through trace-driven experiments, we show that FIRE reduces edge computing costs compared to vanilla RL and the greedy baseline in the event of failures.

Edge computing↗

NASA NTRS2015

Adaptive Stress Testing of Airborne Collision Avoidance Systems

This paper presents a scalable method to efficiently search for the most likely state trajectory leading to an event given only a simulator of a system. Our approach uses a reinforcement learning formulation and solves it using Monte Carlo Tree Search (MCTS). The approach places very few requirements on the underlying system, requiring only that the simulator provide some basic controls, the ability to evaluate certain conditions, and a mechanism to control the stochasticity in the system. Access to the system state is not required, allowing the method to support systems with hidden state. The method is applied to stress test a prototype aircraft collision avoidance system to identify trajectories that are likely to lead to near mid-air collisions. We present results for both single and multi-threat encounters and discuss their relevance. Compared with direct Monte Carlo search, this MCTS method performs significantly better both in finding events and in maximizing their likelihood.

Verification and Validation↗

DOE OSTI2026

Performance-Aligned LLMs for Generating Fast HPC Code

Optimizing scientific software is a difficult task because codebases are often large and complex, and performance can depend upon several factors including the algorithm, its implementation, and hardware among others. Causes of poor performance can originate from disparate sources and be difficult to diagnose. Recent years have seen a multitude of work that use large language models (LLMs) to assist in software development tasks. However, these tools are trained to model the distribution of code as text, and are not specifically designed to understand performance aspects of code. In this work, we introduce a reinforcement learning based methodology to align the outputs of code LLMs with performance. This allows us to build upon the current code modeling capabilities of LLMs and extend them to generate better performing code. Here, we demonstrate that our fine-tuned model improves the expected speedup of generated code over base models for a set of benchmark tasks from 0.9 to 1.6 for serial code and 1.9 to 4.5 for OpenMP parallel code.

Computer science↗