SEARCH · Search NASA

Results for “Agent”

Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.

SourceResearch connectionFrom yearThrough year

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 37 records · Page 2

All sources NASA NTRS DOE OSTI

NASA NTRS2000

Adaptivity in Agent-Based Routing for Data Networks

Adaptivity, both of the individual agents and of the interaction structure among the agents, seems indispensable for scaling up multi-agent systems (MAS s) in noisy environments. One important consideration in designing adaptive agents is choosing their action spaces to be as amenable as possible to machine learning techniques, especially to reinforcement learning (RL) techniques. One important way to have the interaction structure connecting agents itself be adaptive is to have the intentions and/or actions of the agents be in the input spaces of the other agents, much as in Stackelberg games. We consider both kinds of adaptivity in the design of a MAS to control network packet routing. We demonstrate on the OPNET event-driven network simulator the perhaps surprising fact that simply changing the action space of the agents to be better suited to RL can result in very large improvements in their potential performance: at their best settings, our learning-amenable router agents achieve throughputs up to three and one half times better than that of the standard Bellman-Ford routing algorithm, even when the Bellman-Ford protocol traffic is maintained. We then demonstrate that much of that potential improvement can be realized by having the agents learn their settings when the agent interaction structure is itself adaptive.

Wolpert, David H.↗

NASA NTRS1999

Cooperation and Coordination Between Fuzzy Reinforcement Learning Agents in Continuous State Partially Observable Markov Decision Processes

Successful operations of future multi-agent intelligent systems require efficient cooperation schemes between agents sharing learning experiences. We consider a pseudo-realistic world in which one or more opportunities appear and disappear in random locations. Agents use fuzzy reinforcement learning to learn which opportunities are most worthy of pursuing based on their promise rewards, expected lifetimes, path lengths and expected path costs. We show that this world is partially observable because the history of an agent influences the distribution of its future states. We consider a cooperation mechanism in which agents share experience by using and-updating one joint behavior policy. We also implement a coordination mechanism for allocating opportunities to different agents in the same world. Our results demonstrate that K cooperative agents each learning in a separate world over N time steps outperform K independent agents each learning in a separate world over K*N time steps, with this result becoming more pronounced as the degree of partial observability in the environment increases. We also show that cooperation between agents learning in the same world decreases performance with respect to independent agents. Since cooperation reduces diversity between agents, we conclude that diversity is a key parameter in the trade off between maximizing utility from cooperation when diversity is low and maximizing utility from competitive coordination when diversity is high.

Berenji, Hamid R.↗

NASA NTRS2001

Learning Sequences of Actions in Collectives of Autonomous Agents

In this paper we focus on the problem of designing a collective of autonomous agents that individually learn sequences of actions such that the resultant sequence of joint actions achieves a predetermined global objective. We are particularly interested in instances of this problem where centralized control is either impossible or impractical. For single agent systems in similar domains, machine learning methods (e.g., reinforcement learners) have been successfully used. However, applying such solutions directly to multi-agent systems often proves problematic, as agents may work at cross-purposes, or have difficulty in evaluating their contribution to achievement of the global objective, or both. Accordingly, the crucial design step in multiagent systems centers on determining the private objectives of each agent so that as the agents strive for those objectives, the system reaches a good global solution. In this work we consider a version of this problem involving multiple autonomous agents in a grid world. We use concepts from collective intelligence to design goals for the agents that are 'aligned' with the global goal, and are 'learnable' in that agents can readily see how their behavior affects their utility. We show that reinforcement learning agents using those goals outperform both 'natural' extensions of single agent algorithms and global reinforcement, learning solutions based on 'team games'.

Turner, Kagan↗

NASA NTRS2002

CATS-based Air Traffic Controller Agents

This report describes intelligent agents that function as air traffic controllers. Each agent controls traffic in a single sector in real time; agents controlling traffic in adjoining sectors can coordinate to manage an arrival flow across a given meter fix. The purpose of this research is threefold. First, it seeks to study the design of agents for controlling complex systems. In particular, it investigates agent planning and reactive control functionality in a dynamic environment in which a variety perceptual and decision making skills play a central role. It examines how heuristic rules can be applied to model planning and decision making skills, rather than attempting to apply optimization methods. Thus, the research attempts to develop intelligent agents that provide an approximation of human air traffic controller behavior that, while not based on an explicit cognitive model, does produce task performance consistent with the way human air traffic controllers operate. Second, this research sought to extend previous research on using the Crew Activity Tracking System (CATS) as the basis for intelligent agents. The agents use a high-level model of air traffic controller activities to structure the control task. To execute an activity in the CATS model, according to the current task context, the agents reference a 'skill library' and 'control rules' that in turn execute the pattern recognition, planning, and decision-making required to perform the activity. Applying the skills enables the agents to modify their representation of the current control situation (i.e., the 'flick' or 'picture'). The updated representation supports the next activity in a cycle of action that, taken as a whole, simulates air traffic controller behavior. A third, practical motivation for this research is to use intelligent agents to support evaluation of new air traffic control (ATC) methods to support new Air Traffic Management (ATM) concepts. Current approaches that use large, human-in-the-loop simulations are unquestionably valuable for this purpose, but pose considerable logistical, fiscal, and experimental control problems. First, data analysis is extremely complicated, owing simply to the large number of participants and data sources in such simulations. In addition, experienced human air traffic controllers working adjacent sectors tend to flexibly adapt to the evolving control problem - potentially shifting to other strategies than those under investigation. In addition, their performance is tightly coupled to the control interface, which in the development phase may support some concepts and supporting strategies better than others. A simple shift in strategy by one controller can change the character of a particular traffic scenario dramatically, which makes experimental comparison of ATC performance under different traffic scenarios difficult. Training a given team of controllers on operations under a new ATM concept for a sufficient period of time could avert such difficulties, but instituting an adequate training program is expensive and logistically difficult.

Callantine, Todd J.↗

DOE OSTI2026

MADA: Multi-Agent Design Assistant

MADA (Multi-Agent Design Assistant) is a Large Language Model (LLM) powered multi-agent framework that coordinates specialized agents for complex design workflows. The system was designed for HPC workflows with the following agents in mind: 1) A Job Management Agent (JMA) launches and manages ensemble simulations on HPC systems, 2) a Geometry Agent (GA) generates meshes, and 3) an Inverse Design Agent (IDA) proposes new designs informed by simulation outcomes. Our framework reduces cumbersome manual workflow setup, and enables automated design exploration at scale. However, the software also enables users to rapidly create new multi-agent systems. Simply define new agents in a configuration file, giving each their own set of tools (via MCP), and then chat and prompt your new multi-agent system. Is

Gunnarson, BrianS [Lawrence Livermore National Lab↗

NASA NTRS1994

DYNACLIPS (DYNAmic CLIPS): A dynamic knowledge exchange tool for intelligent agents

In a dynamic environment, intelligent agents must be responsive to unanticipated conditions. When such conditions occur, an intelligent agent may have to stop a previously planned and scheduled course of actions and replan, reschedule, start new activities and initiate a new problem solving process to successfully respond to the new conditions. Problems occur when an intelligent agent does not have enough knowledge to properly respond to the new situation. DYNACLIPS is an implementation of a framework for dynamic knowledge exchange among intelligent agents. Each intelligent agent is a CLIPS shell and runs a separate process under SunOS operating system. Intelligent agents can exchange facts, rules, and CLIPS commands at run time. Knowledge exchange among intelligent agents at run times does not effect execution of either sender and receiver intelligent agent. Intelligent agents can keep the knowledge temporarily or permanently. In other words, knowledge exchange among intelligent agents would allow for a form of learning to be accomplished.

Cengeloglu, Yilmaz↗

NASA NTRS2003

Mobile Agents: A Distributed Voice-Commanded Sensory and Robotic System for Surface EVA Assistance

A model-based, distributed architecture integrates diverse components in a system designed for lunar and planetary surface operations: spacesuit biosensors, cameras, GPS, and a robotic assistant. The system transmits data and assists communication between the extra-vehicular activity (EVA) astronauts, the crew in a local habitat, and a remote mission support team. Software processes ("agents"), implemented in a system called Brahms, run on multiple, mobile platforms, including the spacesuit backpacks, all-terrain vehicles, and robot. These "mobile agents" interpret and transform available data to help people and robotic systems coordinate their actions to make operations more safe and efficient. Different types of agents relate platforms to each other ("proxy agents"), devices to software ("comm agents"), and people to the system ("personal agents"). A state-of-the-art spoken dialogue interface enables people to communicate with their personal agents, supporting a speech-driven navigation and scheduling tool, field observation record, and rover command system. An important aspect of the engineering methodology involves first simulating the entire hardware and software system in Brahms, and then configuring the agents into a runtime system. Design of mobile agent functionality has been based on ethnographic observation of scientists working in Mars analog settings in the High Canadian Arctic on Devon Island and the southeast Utah desert. The Mobile Agents system is developed iteratively in the context of use, with people doing authentic work. This paper provides a brief introduction to the architecture and emphasizes the method of empirical requirements analysis, through which observation, modeling, design, and testing are integrated in simulated EVA operations.

Clancey, William J.↗

NASA NTRS2005

Quicker Q-Learning in Multi-Agent Systems

Multi-agent learning in Markov Decisions Problems is challenging because of the presence ot two credit assignment problems: 1) How to credit an action taken at time step t for rewards received at t' greater than t; and 2) How to credit an action taken by agent i considering the system reward is a function of the actions of all the agents. The first credit assignment problem is typically addressed with temporal difference methods such as Q-learning OK TD(lambda) The second credit assi,onment problem is typically addressed either by hand-crafting reward functions that assign proper credit to an agent, or by making certain independence assumptions about an agent's state-space and reward function. To address both credit assignment problems simultaneously, we propose the Q Updates with Immediate Counterfactual Rewards-learning (QUICR-learning) designed to improve both the convergence properties and performance of Q-learning in large multi-agent problems. Instead of assuming that an agent s value function can be made independent of other agents, this method suppresses the impact of other agents using counterfactual rewards. Results on multi-agent grid-world problems over multiple topologies show that QUICR-learning can achieve up to thirty fold improvements in performance over both conventional and local Q-learning in the largest tested systems.

Agogino, Adrian K.↗

NASA NTRS2003

Validating the Autonomous Science Agent

This paper describes the validation process for the Autonomous Science Agent, a software agent that will fly onboard the EO-1 spacecraft from 2003-2004. This agent will recognize science events, retarget the spacecraft to respond to the science events, and reduce data downlink to only the highest value science data. The autonomous science agent has been designed using a layered architectural approach with specific redundant safeguards to reduce the risk of an agent malfunction to the EO-1 spacecraft. This 'safe' design is also in the process of being thoroughly validated by informal validation methods and extensive testing. This paper describes the analysis used to define agent safety, elements of the design that increase the safety of the agent, and the process being used to validate agent safety prior to the agent software controlling the spacecraft.

agent safety↗

NASA NTRS1990

A distributed agent architecture for real-time knowledge-based systems: Real-time expert systems project, phase 1

We propose a distributed agent architecture (DAA) that can support a variety of paradigms based on both traditional real-time computing and artificial intelligence. DAA consists of distributed agents that are classified into two categories: reactive and cognitive. Reactive agents can be implemented directly in Ada to meet hard real-time requirements and be deployed on on-board embedded processors. A traditional real-time computing methodology under consideration is the rate monotonic theory that can guarantee schedulability based on analytical methods. AI techniques under consideration for reactive agents are approximate or anytime reasoning that can be implemented using Bayesian belief networks as in Guardian. Cognitive agents are traditional expert systems that can be implemented in ART-Ada to meet soft real-time requirements. During the initial design of cognitive agents, it is critical to consider the migration path that would allow initial deployment on ground-based workstations with eventual deployment on on-board processors. ART-Ada technology enables this migration while Lisp-based technologies make it difficult if not impossible. In addition to reactive and cognitive agents, a meta-level agent would be needed to coordinate multiple agents and to provide meta-level control.

Lee, S. Daniel↗

NASA NTRS1999

Autonomous Agents: The Origins and Co-Evolution of Reproducing Molecular Systems

The central aim of this award concerned an investigation into, and adequate formulation of, the concept of an "autonomous agent." If we consider a bacterium swimming upstream in a glucose gradient, we are willing to say of the bacterium that it is going to get food. That is, we are willing, and do, describe the bacterium as acting on its own behalf in an environment. All free living cells are, in this sense, autonomous agents. But the bacterium is "just" a set of molecules. We define an autonomous agent as a physical system able to act on its own behalf in an environment, then ask, "What must a physical system be to be an autonomous agent?" The tentative definition for a molecular autonomous agent is that it must be self-reproducing and carry out at least one thermodynamic work cycle. The work carried out in this grant involved, among other features, the development of a detailed model of a molecular autonomous agent, and study of the kinetics of this system. In particular, a molecular autonomous agent must, by the above tentative definition, not only reproduce, but must carry out at least one work cycle. I took, as a simple example of a self-reproducing molecular system, the single-stranded DNA hexamer 3'CCGCGG5' which can line up and ligate its two complementary trimers, 5'CCG3' and 5'CGG3'. But the two ligated trimers constitute the same molecular sequence in the 3' to 5' direction as the initial hexamer, hence this system is autocatalytic. On the other hand the above system is not yet an autonomous agent. At the minimum, autonomous agents, as I have defined them, are a new class of chemical reaction network. At a maximum, they may constitute a proper definition of life itself.

Kauffman, Stuart↗

NASA NTRS2002

Time-Extended Payoffs for Collectives of Autonomous Agents

A collective is a set of self-interested agents which try to maximize their own utilities, along with a a well-defined, time-extended world utility function which rates the performance of the entire system. In this paper, we use theory of collectives to design time-extended payoff utilities for agents that are both aligned with the world utility, and are "learnable", i.e., the agents can readily see how their behavior affects their utility. We show that in systems where each agent aims to optimize such payoff functions, coordination arises as a byproduct of the agents selfishly pursuing their own goals. A game theoretic analysis shows that such payoff functions have the net effect of aligning the Nash equilibrium, Pareto optimal solution and world utility optimum, thus eliminating undesirable behavior such as agents working at cross-purposes. We then apply collective-based payoff functions to the token collection in a gridworld problem where agents need to optimize the aggregate value of tokens collected across an episode of finite duration (i.e., an abstracted version of rovers on Mars collecting scientifically interesting rock samples, subject to power limitations). We show that, regardless of the initial token distribution, reinforcement learning agents using collective-based payoff functions significantly outperform both natural extensions of single agent algorithms and global reinforcement learning solutions based on "team games".

Tumer, Kagan↗

NASA NTRS2002

CATS-based Agents That Err

This report describes preliminary research on intelligent agents that make errors. Such agents are crucial to the development of novel agent-based techniques for assessing system safety. The agents extend an agent architecture derived from the Crew Activity Tracking System that has been used as the basis for air traffic controller agents. The report first reviews several error taxonomies. Next, it presents an overview of the air traffic controller agents, then details several mechanisms for causing the agents to err in realistic ways. The report presents a performance assessment of the error-generating agents, and identifies directions for further research. The research was supported by the System-Wide Accident Prevention element of the FAA/NASA Aviation Safety Program.

Callantine, Todd J.↗

NASA NTRS2005

A Scalable and Robust Multi-Agent Approach to Distributed Optimization

Modularizing a large optimization problem so that the solutions to the subproblems provide a good overall solution is a challenging problem. In this paper we present a multi-agent approach to this problem based on aligning the agent objectives with the system objectives, obviating the need to impose external mechanisms to achieve collaboration among the agents. This approach naturally addresses scaling and robustness issues by ensuring that the agents do not rely on the reliable operation of other agents We test this approach in the difficult distributed optimization problem of imperfect device subset selection [Challet and Johnson, 2002]. In this problem, there are n devices, each of which has a "distortion", and the task is to find the subset of those n devices that minimizes the average distortion. Our results show that in large systems (1000 agents) the proposed approach provides improvements of over an order of magnitude over both traditional optimization methods and traditional multi-agent methods. Furthermore, the results show that even in extreme cases of agent failures (i.e., half the agents fail midway through the simulation) the system remains coordinated and still outperforms a failure-free and centralized optimization algorithm.

Tumer, Kagan↗

NASA NTRS2006

Efficient Agent-Based Cluster Ensembles

Numerous domains ranging from distributed data acquisition to knowledge reuse need to solve the cluster ensemble problem of combining multiple clusterings into a single unified clustering. Unfortunately current non-agent-based cluster combining methods do not work in a distributed environment, are not robust to corrupted clusterings and require centralized access to all original clusterings. Overcoming these issues will allow cluster ensembles to be used in fundamentally distributed and failure-prone domains such as data acquisition from satellite constellations, in addition to domains demanding confidentiality such as combining clusterings of user profiles. This paper proposes an efficient, distributed, agent-based clustering ensemble method that addresses these issues. In this approach each agent is assigned a small subset of the data and votes on which final cluster its data points should belong to. The final clustering is then evaluated by a global utility, computed in a distributed way. This clustering is also evaluated using an agent-specific utility that is shown to be easier for the agents to maximize. Results show that agents using the agent-specific utility can achieve better performance than traditional non-agent based methods and are effective even when up to 50% of the agents fail.

Agogino, Adrian↗

NASA NTRS1993

Method and apparatus to characterize ultrasonically reflective contrast agents

A method and apparatus for characterizing the time and frequency response of an ultrasonically reflective contrast agent is disclosed. An ultrasonically reflective contrast agent is injected, under constant pressure, into a fluid flowing through a pump flow circuit. The fluid and the ultrasonically reflective contrast agent are uniformly mixed in a mixing chamber, and the uniform mixture is passed through a contrast agent chamber. The contrast agent chamber is acoustically and axially interposed between an ultrasonic transducer chamber and an acoustic isolation chamber. A pulse of ultrasonic energy is transmitted into the contrast agent chamber from the ultrasonic transducer chamber. An echo waveform is received from the ultrasonically reflective contrast agent, and it is analyzed to determine the time and frequency response of the ultrasonically reflective contrast agent.

Pretlow, Robert A., III↗

NASA NTRS1993

Intelligent Agent Architectures: Reactive Planning Testbed

An Integrated Agent Architecture (IAA) is a framework or paradigm for constructing intelligent agents. Intelligent agents are collections of sensors, computers, and effectors that interact with their environments in real time in goal-directed ways. Because of the complexity involved in designing intelligent agents, it has been found useful to approach the construction of agents with some organizing principle, theory, or paradigm that gives shape to the agent's components and structures their relationships. Given the wide variety of approaches being taken in the field, the question naturally arises: Is there a way to compare and evaluate these approaches? The purpose of the present work is to develop common benchmark tasks and evaluation metrics to which intelligent agents, including complex robotic agents, constructed using various architectural approaches can be subjected.

Rosenschein, Stanley J.↗

NASA NTRS2004

Interaction and Communication of Agents in Networks and Language Complexity Estimates

Knowledge acquisition and sharing are arguably the most critical activities of communicating agents. We report about our on-going project featuring knowledge acquisition and sharing among communicating agents embedded in a network. The applications we target range from hardware robots to virtual entities such as internet agents. Agent experiments can be simulated using a convenient simulation language. We analyzed the complexity of communicating agent simulations using Java and Easel. Scenarios we have studied are listed below. The communication among agents can range from declarative queries to sub-natural language queries. 1) A set of agents monitoring an object are asked to build activity profiles based on exchanging elementary observations; 2) A set of car drivers form a line, where every car is following its predecessor. An unsafe distance cm create a strong wave in the line. Individual agents are asked to incorporate and apply directions how to avoid the wave. 3) A set of micro-vehicles form a grid and are asked to propagate information and concepts to a central server.

Smid, Jan↗