Search NASA⌕ Search

SEARCH · Search NASA

Results for “Learning”

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.

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

At least 55 records · Page 3
All sourcesNASA NTRSDOE OSTI
NASA NTRSDate not supplied

Exploring Informal Learning at the Airlines

Airline pilot training is extensive, highly structured, and defined by aircraft and airspace system operating requirements, yet pilots describe a tradition of between-pilot knowledge transfer and self-directed learning. This learning supplements their approved training programs. While industry and regulators focus on “formal learning” systems, pilots report relying on “informal learning” to build operational expertise. The persistence of informal learning suggests gaps in how successfully formal learning prepares pilots to handle operational complexities. The community that researches learning has extensively studied informal learning, and its characteristics seem to align with how pilots report increasing their skills and knowledge informally. However, no research into informal learning practices among airline pilots seems to exist. In this paper we provide examples of informal learning in commercial aviation, how they fit into two existing frameworks for workplace learning, and propose that researching informal learning might help identify opportunities to improve formal aviation learning systems.

pilot learning↗
NASA NTRS2004

Perceptual learning through optimization of attentional weighting: human versus optimal Bayesian learner

Human performance in visual detection, discrimination, identification, and search tasks typically improves with practice. Psychophysical studies suggest that perceptual learning is mediated by an enhancement in the coding of the signal, and physiological studies suggest that it might be related to the plasticity in the weighting or selection of sensory units coding task relevant information (learning through attention optimization). We propose an experimental paradigm (optimal perceptual learning paradigm) to systematically study the dynamics of perceptual learning in humans by allowing comparisons to that of an optimal Bayesian algorithm and a number of suboptimal learning models. We measured improvement in human localization (eight-alternative forced-choice with feedback) performance of a target randomly sampled from four elongated Gaussian targets with different orientations and polarities and kept as a target for a block of four trials. The results suggest that the human perceptual learning can occur within a lapse of four trials (<1 min) but that human learning is slower and incomplete with respect to the optimal algorithm (23.3% reduction in human efficiency from the 1st-to-4th learning trials). The greatest improvement in human performance, occurring from the 1st-to-2nd learning trial, was also present in the optimal observer, and, thus reflects a property inherent to the visual task and not a property particular to the human perceptual learning mechanism. One notable source of human inefficiency is that, unlike the ideal observer, human learning relies more heavily on previous decisions than on the provided feedback, resulting in no human learning on trials following a previous incorrect localization decision. Finally, the proposed theory and paradigm provide a flexible framework for future studies to evaluate the optimality of human learning of other visual cues and/or sensory modalities.

Non-NASA Center↗
NASA NTRS2020

Amino Acid Encoding for Deep Learning Applications

Background: The number of applications of deep learning algorithms in bioinformatics is increasing as they usually achieve superior performance over classical approaches, especially, when bigger training datasets are available. In deep learning applications, discrete data, e.g. words or n-grams in language, or amino acids or nucleotides in bioinformatics, are generally represented as a continuous vector through an embedding matrix. Recently, learning this embedding matrix directly from the data as part of the continuous iteration of the model to optimize the target prediction – a process called ‘end-to-end learning’ – has led to state-of-the-art results in many fields. Although usage of embeddings is well described in the bioinformatics literature, the potential of end-to-end learning for single amino acids, as compared to more classical manually-curated encoding strategies, has not been systematically addressed. To this end, we compared classical encoding matrices, namely one-hot, VHSE8 and BLOSUM62, to end-to-end learning of amino acid embeddings for two different prediction tasks using three widely used architectures, namely recurrent neural networks (RNN), convolutional neural networks (CNN), and the hybrid CNN-RNN. Results: By using different deep learning architectures, we show that end-to-end learning is on par with classical encodings for embeddings of the same dimension even when limited training data is available, and might allow for a reduction in the embedding dimension without performance loss, which is critical when deploying the models to devices with limited computational capacities. We found that the embedding dimension is a major factor in controlling the model performance. Surprisingly, we observed that deep learning models are capable of learning from random vectors of appropriate dimension. Conclusion: Our study shows that end-to-end learning is a flexible and powerful method for amino acid encoding. Further, due to the flexibility of deep learning systems, amino acid encoding schemes should be benchmarked against random vectors of the same dimension to disentangle the information content provided by the encoding scheme from the distinguishability effect provided by the scheme.

Deep-learning↗
NASA NTRS1994

Learning fuzzy logic control system

The performance of the Learning Fuzzy Logic Control System (LFLCS), developed in this thesis, has been evaluated. The Learning Fuzzy Logic Controller (LFLC) learns to control the motor by learning the set of teaching values that are generated by a classical PI controller. It is assumed that the classical PI controller is tuned to minimize the error of a position control system of the D.C. motor. The Learning Fuzzy Logic Controller developed in this thesis is a multi-input single-output network. Training of the Learning Fuzzy Logic Controller is implemented off-line. Upon completion of the training process (using Supervised Learning, and Unsupervised Learning), the LFLC replaces the classical PI controller. In this thesis, a closed loop position control system of a D.C. motor using the LFLC is implemented. The primary focus is on the learning capabilities of the Learning Fuzzy Logic Controller. The learning includes symbolic representation of the Input Linguistic Nodes set and Output Linguistic Notes set. In addition, we investigate the knowledge-based representation for the network. As part of the design process, we implement a digital computer simulation of the LFLCS. The computer simulation program is written in 'C' computer language, and it is implemented in DOS platform. The LFLCS, designed in this thesis, has been developed on a IBM compatible 486-DX2 66 computer. First, the performance of the Learning Fuzzy Logic Controller is evaluated by comparing the angular shaft position of the D.C. motor controlled by a conventional PI controller and that controlled by the LFLC. Second, the symbolic representation of the LFLC and the knowledge-based representation for the network are investigated by observing the parameters of the Fuzzy Logic membership functions and the links at each layer of the LFLC. While there are some limitations of application with this approach, the result of the simulation shows that the LFLC is able to control the angular shaft position of the D.C. motor. Furthermore, the LFLC has better performance in rise time, settling time and steady state error than to the conventional PI controller. This abstract accurately represents the content of the candidate's thesis. I recommend its publication.

Lung, Leung Kam↗
NASA NTRS2020

Amino Acid Encoding for Deep Learning Applications

Background: The number of applications of deep learning algorithms in bioinformatics is increasing as they usually achieve superior performance over classical approaches, especially, when bigger training datasets are available. In deep learning applications, discrete data, e.g. words or n-grams in language, or amino acids or nucleotides in bioinformatics, are generally represented as a continuous vector through an embedding matrix. Recently, learning this embedding matrix directly from the data as part of the continuous iteration of the model to optimize the target prediction – a process called ‘end-to-end learning’ – has led to state-ofthe-art results in many fields. Although usage of embeddings is well described in the bioinformatics literature, the potential of end-to-end learning for single amino acids, as compared to more classical manually-curated encoding strategies, has not been systematically addressed. To this end, we compared classical encoding matrices, namely one-hot, VHSE8 and BLOSUM62, to end-to-end learning of amino acid embeddings for two different prediction tasks using three widely used architectures, namely recurrent neural networks (RNN), convolutional neural networks (CNN), and the hybrid CNN-RNN. Results: By using different deep learning architectures, we show that end-to-end learning is on par with classical encodings for embeddings of the same dimension even when limited training data is available, and might allow for a reduction in the embedding dimension without performance loss, which is critical when deploying the models to devices with limited computational capacities. We found that the embedding dimension is a major factor in controlling the model performance. Surprisingly, we observed that deep learning models are capable of learning from random vectors of appropriate dimension. Conclusion: Our study shows that end-to-end learning is a flexible and powerful method for amino acid encoding. Further, due to the flexibility of deep learning systems, amino acid encoding schemes should be benchmarked against random vectors of the same dimension to disentangle the information content provided by the encoding scheme from the distinguishability effect provided by the scheme.

Hesham ElAbd↗
NASA NTRS1994

Learning procedures from interactive natural language instructions

Despite its ubiquity in human learning, very little work has been done in artificial intelligence on agents that learn from interactive natural language instructions. In this paper, the problem of learning procedures from interactive, situated instruction is examined in which the student is attempting to perform tasks within the instructional domain, and asks for instruction when it is needed. Presented is Instructo-Soar, a system that behaves and learns in response to interactive natural language instructions. Instructo-Soar learns completely new procedures from sequences of instruction, and also learns how to extend its knowledge of previously known procedures to new situations. These learning tasks require both inductive and analytic learning. Instructo-Soar exhibits a multiple execution learning process in which initial learning has a rote, episodic flavor, and later executions allow the initially learned knowledge to be generalized properly.

Huffman, Scott B.↗
NASA NTRS2004

Planning Mars Memory: Learning from the Mer Mission

Knowledge management for space exploration is part of a multi-generational effort at recognizing, preserving and transmitting learning. Each mission should be built on the learning, of both successes and failures, derived from previous missions. Knowledge management begins with learning, and the recognition that this learning has produced knowledge. The Mars Exploration Rover mission provides us with an opportunity to track how learning occurs, how it is recorded, and whether the representations of this learning will be optimally useful for subsequent missions. This paper focuses on the MER science and engineering teams during Rover operations. A NASA team conducted an observational study of the ongoing work and learning of the these teams. Learning occurred in a wide variety of areas: how to run two teams on Mars time for three months; how to use the instruments within the constraints of the martian environment, the deep space network and the mission requirements; how to plan science strategy; how best to use the available software tools. This learning is preserved in many ways. Primarily it resides in peoples memories, to be carried on to the next mission. It is also encoded in stones, in programming sequences, in published reports, and in lessons learned activities, Studying learning and knowledge development as it happens allows us to suggest proactive ways of capturing and using it across multiple missions and generations.

Linde, Charlotte↗
NASA NTRSDate not supplied

Towards Autonomous Lunar Resource Excavation via Reinforcement Learning

To continue on a sustainable and flexible path, NASA needs to address the challenge of collecting and moving large amounts of regolith at the destination. NASA’s Regolith Advanced Surface Systems Operations Robot (RASSOR) is principally designed to mine and deliver regolith for In-Situ Resource Utilization (ISRU) processing. RASSOR’s design enables it to efficiently collect and deposit regolith, return collected material for processing, and myriad related ISRU activities. To reliably perform these operations on the lunar surface, RASSOR software and sensory systems need to be robust and maximize the information extracted from a reduced sensor payload. Herein, we present preliminary findings from the Intelligent Capabilities Enhanced RASSOR project. We created reduced-order simulation environments to develop autonomous trenching controllers via reinforcement learning and prototype state estimation architectures. The goal of reinforcement learning is for an agent to learn a policy (task strategy) through interactions with an environment. When the agent performs an action, a change occurs in environment state and a numerical reward is received which informs the agent whether the action performed was good or not. Since reinforcement learning algorithms learn through trial-and-error, a simulation is a desirable first environment for development and learning. We developed two simulations, the first is a 2D excavation simulation developed to facilitate parameter selection, and a 3D simulation developed using a game physics engine, to simulate simplified soil interactions and increase the fidelity of the dynamic models of the robotic agents. The development of this 3D simulation has enabled the training of additional sensing capabilities and research both at the granular mechanics and operations levels. We experimented with various virtual sensor payloads to identify a combination that enabled efficient excavation operation and learning. Our reward function is based on how much material is excavated per step. A penalty is also received for leaving the dig site and to smooth the acceleration of the drum arms. We implemented pseudo time-of-flight sensors to report distance from each drum to ground and the height above ground which was found to be more efficient than existing solutions. Our findings suggest that reinforcement learning for autonomous operations has learned viable trenching strategies within 3000 training episodes in our simplified 2D environment and helped identify desirable sensing capabilities, arrangements, and considerations such as the positioning of time-of-flight sensors. Future work includes expanding our simulation to more complex environments and scenarios, and transfer learning from simulation to RASSOR 2.0 hardware for deployment in the Regolith Test Bin at NASA's Kennedy Space Center.

rassor↗
NASA NTRS2008

Best Practices for Researching and Documenting Lessons Learned

Identification, resolution, and avoidance of technical and programmatic issues are important for ensuring safe and successful space missions. Although the importance of applying lessons learned to reduce risk is frequently stressed, there is little material available to help technical and management personnel research and document lessons learned. Collecting, researching, identifying, and documenting lessons learned that will be useful to current and future management and engineering personnel is not always a straightforward task. This white paper presents lessons learned and best practices concerning the research and documentation of technical and organizational lessons learned. It is intended to enable organizations to initiate or improve lessons learned research and documentation efforts. The content of this white paper is based on four technical lessons learned projects conducted by the United Space Alliance (USA) Flight Design and Dynamics Department, in support of the NASA/Johnson Space Center (JSC) Flight Design and Dynamics Division. Each project published a report, titled as follows: (1) GPS Lessons Learned From the ISS, Space Shuttle and X-38; (2) Lessons Learned From Seven Space Shuttle Missions; (3) Space Shuttle Rendezvous and Proximity Operations Experience Report; and (4) Navigation Technical History with Lessons Learned. The four projects were different in availability of subject matter experts and primary source material, subject scope, and the level of effort required to produce the final report. However, generic lessons can be drawn from all of them. The best practices will be discussed by the phases of report research and development: (1) Defining Report Requirements, Project Organization, and Schedule; (2) Collection and Analysis of Source Material; (3) Writing and Integrating the Report; and (4) Review and Revision of the Report.

Goodman, John L.↗
NASA NTRS2004

Lessons Learned Study Final Report for the Exploration Systems Mission Directorate

This report is the final product of a 90-day study performed for the Exploration Systems Mission Directorate. The study was to assemble lessons NASA has learned from previous programs that could help the Exploration Systems Mission Directorate pursue the Exploration vision. It focuses on those lessons that should have the greatest significance to the Directorate during the formulation of program and mission plans. The study team reviewed a large number of lessons learned reports and data bases, including the Columbia Accident Investigation Board and Rogers Commission reports on the Shuttle accidents, accident reports from robotic space flight systems, and a number of management reviews by the Defense Sciences Board, Government Accountability Office, and others. The consistency of the lessons, findings, and recommendations validate the adequacy of the data set. In addition to reviewing existing databases, a series of workshops was held at each of the NASA centers and headquarters that included senior managers from the current workforce as well as retirees. The full text of the workshop reports is included in Appendix A. A lessons learned website was opened up to permit current and retired NASA personnel and on-site contractors to input additional lessons as they arise. These new lessons, when of appropriate quality and relevance, will be brought to the attention of managers. The report consists of four parts: Part 1 provides a small set of lessons, called the Executive Lessons Learned, that represent critical lessons that the Exploration Systems Mission Directorate should act on immediately. This set of Executive Lessons and their supporting rationale have been reviewed at length and fully endorsed by a team of distinguished NASA alumni; Part 2 contains a larger set of lessons, called the Selected Lessons Learned, which have been chosen from the lessons database and center workshop reports on the basis of their specific significance and relevance to the near-term work of the Exploration Directorate. These lessons frequently support the Executive lessons but are more general in nature; Part 3 consists of the reports of the center workshops that were conducted as part of this activity. These reports are included in their entirety (approximately 200 pages) in Appendix G and have significance for specific managers; Part 4 consists of the remainder of the lessons that have been selected by this effort and assembled into a database for the use of the Explorations Directorate. The database is archived and hosted in the Lessons Learned Knowledge Network, which provides a flexible search capability using a wide variety of search terms. Finally, a spreadsheet lists databases searched and a bibliography identifies reports that have been reviewed as sources of lessons for this task. NASA has been presented with many learning opportunities. We have conducted numerous programs, some extremely successful and others total failures. Most have been documented with a formal lessons learned activity, but we have not always incorporated these learning opportunities into our normal modes of business. For example, the Robbins Report of 2001 clearly indicates that many project failures of the past two decades were the result of violating well documented best practices, often in direct violation of management instructions and directives. An overarching lesson emerges: that disciplined execution in accordance with proven best practices is the greatest single contributor to a successful program. The Lessons Learned task team offers a sincere hope that the lessons presented herein will be helpful to the Exploration Systems Directorate in charting and executing their course. The success of the Directorate and of NASA in general depends on our collective ability to move forward without having to relearn the lessons of those who have gone before.

Van Laak, Jim↗
NASA NTRS2010

The Value of Identifying and Recovering Lost GN&C Lessons Learned: Aeronautical, Spacecraft, and Launch Vehicle Examples

Within the broad aerospace community the importance of identifying, documenting and widely sharing lessons learned during system development, flight test, operational or research programs/projects is broadly acknowledged. Documenting and sharing lessons learned helps managers and engineers to minimize project risk and improve performance of their systems. Often significant lessons learned on a project fail to get captured even though they are well known 'tribal knowledge' amongst the project team members. The physical act of actually writing down and documenting these lessons learned for the next generation of NASA GN&C engineers fails to happen on some projects for various reasons. In this paper we will first review the importance of capturing lessons learned and then will discuss reasons why some lessons are not documented. A simple proven approach called 'Pause and Learn' will be highlighted as a proven low-impact method of organizational learning that could foster the timely capture of critical lessons learned. Lastly some examples of 'lost' GN&C lessons learned from the aeronautics, spacecraft and launch vehicle domains are briefly highlighted. In the context of this paper 'lost' refers to lessons that have not achieved broad visibility within the NASA-wide GN&C CoP because they are either undocumented, masked or poorly documented in the NASA Lessons Learned Information System (LLIS).

Dennehy, Cornelius J.↗
NASA NTRS1992

Indirect decentralized learning control

The new field of learning control develops controllers that learn to improve their performance at executing a given task, based on experience performing this specific task. In a previous work, the authors presented a theory of indirect learning control based on use of indirect adaptive control concepts employing simultaneous identification and control. This paper develops improved indirect learning control algorithms, and studies the use of such controllers in decentralized systems. The original motivation of the learning control field was learning in robots doing repetitive tasks such as on an assembly line. This paper starts with decentralized discrete time systems, and progresses to the robot application, modeling the robot as a time varying linear system in the neighborhood of the nominal trajectory, and using the usual robot controllers that are decentralized, treating each link as if it is independent of any coupling with other links. The basic result of the paper is to show that stability of the indirect learning controllers for all subsystems when the coupling between subsystems is turned off, assures convergence to zero tracking error of the decentralized indirect learning control of the coupled system, provided that the sample time in the digital learning controller is sufficiently short.

Longman, Richard W.↗
NASA NTRS1998

Preliminary Work for Examining the Scalability of Reinforcement Learning

Researchers began studying automated agents that learn to perform multiple-step tasks early in the history of artificial intelligence (Samuel, 1963; Samuel, 1967; Waterman, 1970; Fikes, Hart & Nilsonn, 1972). Multiple-step tasks are tasks that can only be solved via a sequence of decisions, such as control problems, robotics problems, classic problem-solving, and game-playing. The objective of agents attempting to learn such tasks is to use the resources they have available in order to become more proficient at the tasks. In particular, each agent attempts to develop a good policy, a mapping from states to actions, that allows it to select actions that optimize a measure of its performance on the task; for example, reducing the number of steps necessary to complete the task successfully. Our study focuses on reinforcement learning, a set of learning techniques where the learner performs trial-and-error experiments in the task and adapts its policy based on the outcome of those experiments. Much of the work in reinforcement learning has focused on a particular, simple representation, where every problem state is represented explicitly in a table, and associated with each state are the actions that can be chosen in that state. A major advantage of this table lookup representation is that one can prove that certain reinforcement learning techniques will develop an optimal policy for the current task. The drawback is that the representation limits the application of reinforcement learning to multiple-step tasks with relatively small state-spaces. There has been a little theoretical work that proves that convergence to optimal solutions can be obtained when using generalization structures, but the structures are quite simple. The theory says little about complex structures, such as multi-layer, feedforward artificial neural networks (Rumelhart & McClelland, 1986), but empirical results indicate that the use of reinforcement learning with such structures is promising. These empirical results make no theoretical claims, nor compare the policies produced to optimal policies. A goal of our work is to be able to make the comparison between an optimal policy and one stored in an artificial neural network. A difficulty of performing such a study is finding a multiple-step task that is small enough that one can find an optimal policy using table lookup, yet large enough that, for practical purposes, an artificial neural network is really required. We have identified a limited form of the game OTHELLO as satisfying these requirements. The work we report here is in the very preliminary stages of research, but this paper provides background for the problem being studied and a description of our initial approach to examining the problem. In the remainder of this paper, we first describe reinforcement learning in more detail. Next, we present the game OTHELLO. Finally we argue that a restricted form of the game meets the requirements of our study, and describe our preliminary approach to finding an optimal solution to the problem.

Clouse, Jeff↗
NASA NTRS1996

Cascade Error Projection: A Learning Algorithm for Hardware Implementation

In this paper, we workout a detailed mathematical analysis for a new learning algorithm termed Cascade Error Projection (CEP) and a general learning frame work. This frame work can be used to obtain the cascade correlation learning algorithm by choosing a particular set of parameters. Furthermore, CEP learning algorithm is operated only on one layer, whereas the other set of weights can be calculated deterministically. In association with the dynamical stepsize change concept to convert the weight update from infinite space into a finite space, the relation between the current stepsize and the previous energy level is also given and the estimation procedure for optimal stepsize is used for validation of our proposed technique. The weight values of zero are used for starting the learning for every layer, and a single hidden unit is applied instead of using a pool of candidate hidden units similar to cascade correlation scheme. Therefore, simplicity in hardware implementation is also obtained. Furthermore, this analysis allows us to select from other methods (such as the conjugate gradient descent or the Newton's second order) one of which will be a good candidate for the learning technique. The choice of learning technique depends on the constraints of the problem (e.g., speed, performance, and hardware implementation); one technique may be more suitable than others. Moreover, for a discrete weight space, the theoretical analysis presents the capability of learning with limited weight quantization. Finally, 5- to 8-bit parity and chaotic time series prediction problems are investigated; the simulation results demonstrate that 4-bit or more weight quantization is sufficient for learning neural network using CEP. In addition, it is demonstrated that this technique is able to compensate for less bit weight resolution by incorporating additional hidden units. However, generation result may suffer somewhat with lower bit weight quantization.

Duong, Tuan A.↗
NASA NTRS1994

Goal Directed Model Inversion: Learning Within Domain Constraints

Goal Directed Model Inversion (GDMI) is an algorithm designed to generalize supervised learning to the case where target outputs are not available to the learning system. The output of the learning system becomes the input to some external device or transformation, and only the output of this device or transformation can be compared to a desired target. The fundamental driving mechanism of GDMI is to learn from success. Given that a wrong outcome is achieved, one notes that the action that produced that outcome "would have been right if the outcome had been the desired one." The algorithm makes use of these intermediate "successes" to achieve the final goal. A unique and potentially very important feature of this algorithm is the ability to modify the output of the learning module to force upon it a desired syntactic structure. This differs from ordinary supervised learning in the following way: in supervised learning the exact desired output pattern must be provided. In GDMI instead, it is possible to require simply that the output obey certain rules, i.e., that it "make sense" in some way determined by the knowledge domain. The exact pattern that will achieve the desired outcome is then found by the system. The ability to impose rules while allowing the system to search for its own answers in the context of neural networks is potentially a major breakthrough in two ways: 1) it may allow the construction of networks that can incorporate immediately some important knowledge, i.e. would not need to learn everything from scratch as normally required at present, and 2) learning and searching would be limited to the areas where it is necessary, thus facilitating and speeding up the process. These points are illustrated with examples from robotic path planning and parametric design.

Colombano, Silvano P.↗
NASA NTRS1994

Goal Directed Model Inversion: Learning Within Domain Constraints

Goal Directed Model Inversion (GDMI) is an algorithm designed to generalize supervised learning to the case where target outputs are not available to the learning system. The output of the learning system becomes the input to some external device or transformation, and only the output of this device or transformation can be compared to a desired target. The fundamental driving mechanism of GDMI is to learn from success. Given that a wrong outcome is achieved, one notes that the action that produced that outcome "would have been right if the outcome had been the desired one." The algorithm makes use of these intermediate "successes" to achieve the final goal. A unique and potentially very important feature of this algorithm is the ability to modify the output of the learning module to force upon it a desired syntactic structure. This differs from ordinary supervised learning in the following way: in supervised learning the exact desired output pattern must be provided. In GDMI instead, it is possible to require simply that the output obey certain rules, i.e., that it "make sense" in some way determined by the knowledge domain. The exact pattern that will achieve the desired outcome is then found by the system. The ability to impose rules while allowing the system to search for its own answers in the context of neural networks is potentially a major breakthrough in two ways: (1) it may allow the construction of networks that can incorporate immediately some important knowledge, i.e., would not need to learn everything from scratch as normally required at present; and (2) learning and searching would be limited to the areas where it is necessary, thus facilitating and speeding up the process. These points are illustrated with examples from robotic path planning and parametric design.

Colombano, Silvano P.↗
NASA NTRS2001

Lessons Learned and Technical Standards: A Logical Marriage

A comprehensive database of lessons learned that corresponds with relevant technical standards would be a boon to technical personnel and standards developers. The authors discuss the emergence of one such database within NASA, and show how and why the incorporation of lessons learned into technical standards databases can be an indispensable tool for government and industry. Passed down from parent to child, teacher to pupil, and from senior to junior employees, lessons learned have been the basis for our accomplishments throughout the ages. Government and industry, too, have long recognized the need to systematically document And utilize the knowledge gained from past experiences in order to avoid the repetition of failures and mishaps. The use of lessons learned is a principle component of any organizational culture committed to continuous improvement. They have formed the foundation for discoveries, inventions, improvements, textbooks, and technical standards. Technical standards are a very logical way to communicate these lessons. Using the time-honored tradition of passing on lessons learned while utilizing the newest in information technology, the National Aeronautics and Space Administration (NASA) has launched an intensive effort to link lessons learned with specific technical standards through various Internet databases. This article will discuss the importance of lessons learned to engineers, the difficulty in finding relevant lessons learned while engaged in an engineering project, and the new NASA project that can help alleviate this difficulty. The article will conclude with recommendations for more expanded cross-sectoral uses of lessons learned with reference to technical standards.

Gill, Paul↗
NASA NTRS2006

QUICR-learning for Multi-Agent Coordination

Coordinating multiple agents that need to perform a sequence of actions to maximize a system level reward requires solving two distinct credit assignment problems. First, credit must be assigned for an action taken at time step t that results in a reward at time step t > t. Second, credit must be assigned for the contribution of agent i to the overall system performance. The first credit assignment problem is typically addressed with temporal difference methods such as Q-learning. The second credit assignment problem is typically addressed by creating custom reward functions. 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. QUICR-learning is based on previous work on single-time-step counterfactual rewards described by the collectives framework. Results on a traffic congestion problem shows that QUICR-learning is significantly better than a Q-learner using collectives-based (single-time-step counterfactual) rewards. In addition QUICR-learning provides significant gains over conventional and local Q-learning. Additional results on a multi-agent grid-world problem show that the improvements due to QUICR-learning are not domain specific and can provide up to a ten fold increase in performance over existing methods.

Agogino, Adrian K.↗