Concurrent Learning Research Articles

Concurrent-learning-based visual servo tracking and scene identification of mobile robots

Purpose The purpose of this paper is to present a visual servo tracking strategy for the wheeled mobile robot, where the unknown feature depth information can be identified simultaneously in the visual servoing process. Design/methodology/approach By using reference, desired and current images, system errors are constructed by measurable signals that are obtained by decomposing Euclidean homographies. Subsequently, by taking the advantage of the concurrent learning framework, both historical and current system data are used to construct an adaptive updating mechanism for recovering the unknown feature depth. Then, the kinematic controller is designed for the mobile robot to achieve the visual servo trajectory tracking task. Lyapunov techniques and LaSalle’s invariance principle are used to prove that system errors and the depth estimation error converge to zero synchronously. Findings The concurrent learning-based visual servo tracking and identification technology is found to be reliable, accurate and efficient with both simulation and comparative experimental results. Both trajectory tracking and depth estimation errors converge to zero successfully. Originality/value On the basis of the concurrent learning framework, an adaptive control strategy is developed for the mobile robot to successfully identify the unknown scene depth while accomplishing the visual servo trajectory tracking task.

Dual Quaternions as a Tool for Modeling, Control, and Estimation for Spacecraft Robotic Servicing Missions

In recent years there has been an increasing interest in spacecraft robotic operations in orbit. In fact, several agencies and organizations around the world are investigating satellite proximity operations as an enabling technology for future space missions such as on-orbit satellite inspection, health monitoring, surveillance, servicing, refueling, and optical interferometry, to name a few. Contrary to more traditional satellite applications, robotic servicing requires addressing both the translational and the rotational motion of the satellite at the same time. One of the biggest challenges for these applications is the need to simultaneously and accurately estimate – and track – both relative position and attitude reference trajectories in order to avoid collisions between the satellites and achieve stringent mission objectives. Motivated by our desire to control spacecraft motion during proximity operations for robotic in-orbit servicing missions which do not depend on the artificial separation of translational and rotational motion, we have recently developed a complete theory to describe the 6-DOF motion of the spacecraft using dual quaternions. Dual quaternions emerge as a powerful tool to model the pose (that is, both attitude and position) of the spacecraft during all phases of the mission under a unified framework. In this paper, we revisit the basic theory behind dual quaternions, the associated Clifford algebras, and compare quaternion-based attitude rigid-body control laws and estimation algorithms, to their dual quaternion-based pose counterparts. We also show that the resulting mathematical structure lends itself to the straightforward incorporation of an adaptive estimation scheme known as concurrent learning, which allows us to also estimate on-the-fly the mass properties of the spacecraft.

Gradient-Based Discrete-Time Concurrent Learning for Standalone Function Approximation

When using standard learning techniques for uncertainty approximation, persistently exciting inputs are necessary in order to achieve good parameter estimation. It has been shown in recent years that, through concurrent learning (CL), it is possible to achieve better learning without requiring persistency of excitation. We define good learning here by how well an uncertainty can be reconstructed given the estimated parameters. Most studies concerning CL have however been done in the continuous-time (CT) framework. While working with discrete-time (DT) structured uncertainties, we have shown in an earlier study that the concept of CL could be used to solve the parameter identification problem granted, much like in the CT domain, a less restrictive condition compared to that of persistency of excitation is verified. This paper furthers our fundamental study of CL in the DT framework while drawing comparisons with the traditional but fundamental gradient descent technique. As a main contribution, via formal derivations, we present a generalized gradient-based CL motivated DT algorithm for online approximation of both DT structured and unstructured uncertainties. Numerical simulations are provided to show how well the designed algorithm leverages memory usage to achieve better learning.

Concurrent Learning of Adjacent and Nonadjacent Dependencies in Visuo-Spatial and Visuo-Verbal Sequences.

Both adjacent and non-adjacent dependencies (AD and NAD) are present in natural language and other domains, yet the learning of non-adjacent sequential dependencies generally only occurs under favorable circumstances. It is currently unknown to what extent adults can learn AD and NAD, presented concurrently in spatial and verbal sequences during a single session, and whether a second session improves performance. In addition, the relationship between AD and NAD learning and other theoretically related cognitive and language processes has not yet been fully established. In this study, participants reproduced two types of sequences generated from an artificial grammar: visuo-spatial sequences with stimuli presented in four spatial locations, and visuo-verbal sequences with printed syllables. Participants were tested for incidental learning by reproducing novel sequences, half consistent with the grammar and half containing violations of either AD or NAD. The procedure was repeated on a second day. Results showed that both AD and NAD were learned in both visuo-spatial and visuo-verbal tasks, although AD learning was better than NAD and learning of NAD decreased over time. Furthermore, NAD learning for both spatial and verbal tasks was positively correlated with a language measure, whereas AD learning for both spatial and verbal tasks was negatively associated with working memory measures in the opposite domain. These results demonstrate that adults can learn both AD and NAD within a single session, but NAD learning is more easily disrupted than AD and both types of learning are sub-served by partially distinct cognitive processes. These findings increase our understanding of the processes governing the learning of AD and NAD in verbal and spatial domains.

Competence for the Unforeseen: Social Support and Concurrent Learning as Basic Components of Interaction Under Risk

This study examines the importance of basic educational, organizational and operational structures of interaction under risk, and how these vary with competence level within an organization. The purpose of this study is to examine whether there are any basic components that can provide further insight into how competence for interaction under risk can be developed and implemented. The first aim was, therefore, to examine the relationship between components of the basic educational, organizational and operational structures and interaction. The second was to assess whether the perceived level of interaction varies due to competence level, controlling for gender, age, and professional experience. The third aim was to determine whether competence level group membership could be predicted by interaction, social support and the specified educational and organizational components. A questionnaire survey was carried out in the autumn of 2017. The respondents were male or female employees of the Norwegian Armed Forces (n =917). A purposive expert sample of 20 different units with different levels of competence were selected and included commissioned and non-commissioned officers, officer cadets, and conscripts. A total of 1050 personnel were employed by these units. All the employees were asked to participate. The response rate was 87%. The results showed that social support and concurrent learning were the most important predictors of interaction. Social support and concurrent learning combined with basic capabilities, organizational improvisation, training on decision-making, flexibility, general preparedness, and contingency plans accounted for a considerable proportion of the variance in interaction. Interaction, social support, and the specified educational, organizational and operational structure components were also significantly associated with competence level and competence group membership. The results showed that it could be possible to prepare for unforeseen events by implementing in particular social and educational measures that improve interaction. This study should be especially relevant to those involved in handling and stabilizing unforeseen events and emergency preparedness management.

Concurrent concept-cognitive learning model for classification

Concept-cognitive learning (CCL) concerns the manner in which to incorporate new information into itself by mimicking human cognitive processes. Numerous CCL systems have been proposed to meet different requirements based on formal concept analysis. However, most of these can only operate effectively on small-scale datasets and lack classification ability. To overcome these two challenges, based on a regular formal decision context and multi-thread technique, a new concurrent CCL model is proposed in this study as an extension of classical CCL. More precisely, to enhance computational efficiency, a new concurrent learning framework is designed and its corresponding learning algorithms at the initial concept construction and CCL stages are developed. We propose a concurrent incremental learning technique by continuously accommodating newly added data to meet the requirements of classification tasks. Finally, experiments on various datasets, including real-world and synthetic data, demonstrate that the proposed concurrent method can achieve comparable classification effectiveness and significantly improve the concept learning performance simultaneously.

A video-based, flipped classroom, simulation curriculum for dermatologic surgery: A prospective, multi-institution study

Medical education is evolving to emphasize trainee engagement. The impact of a flipped classroom curriculum and surgical simulation on dermatology resident education has not been evaluated. To assess the impact of video education and surgical simulation on dermatology resident procedural skills. We created a curriculum on foundational surgical skills for 31 first- and second-year dermatology residents at 3 institutions. The flipped classroom approach replaces traditional in-person lectures with at-home viewing of instructional videos. After this self-directed learning, trainees had 3 hands-on sessions using simulated skin models. The Objective Structured Assessment of Technical Skills (OSATS) instrument was used to assess residents performing a simulated elliptical excision with intermediate repair before and after the curriculum. Residents completed precurriculum and postcurriculum surveys evaluating operative confidence and perceived value of the curriculum. Residents' total OSATS score increased from a median of 27 (interquartile range, 22-38.5) before the curriculum to 46 (interquartile range, 39.5-51.5) after the curriculum (P<.001). Self-reported confidence in surgical performance significantly improved, and residents were highly satisfied. Limitations include the small sample size and potential influence from concurrent learning on surgical rotations. Video education and simulation are effective for improving dermatology residents' procedural skills. We hope to serve as a template for other institutions and nondermatology trainees hoping to improve procedural skills.

Parameters concurrent learning and reactionless control in post-capture of unknown targets by space manipulators

This paper studies parameters identification and minimizing base disturbances problems after the space manipulator capturing an unknown target. A concurrent learning algorithm that concurrently uses past motion data points and instantaneous motion data of the system is proposed for the parameters identification. Given a condition for selecting the used past data points as well as a scaling technique to make the parameters have the same magnitude, the concurrent learning algorithm guarantees that parameters identification errors can globally converge to zero at an exponential rate and without the need for satisfying the persistent excitation (PE) condition. An adaptive reactionless control method is proposed based on the passivity theorem and Task-priority method, which ensures that the base attitude is stationary and joint motions satisfy the limits during the system generating excitation motions for the parameters identification. Simulation results verify the effectiveness of the proposed method.

Performance Improvement of Finite Time Parameter Estimation with Relaxed Persistence of Excitation Condition

In this paper, a novel finite time parameter estimation method is proposed to solve the parameter estimation problem for a class of linearly parameterized nonlinear systems. The main feature of the proposed method is that the existing method is modified via concurrent learning technique such that the strict persistence of excitation (PE) condition on the regression matrix is relaxed to a rank condition on the recorded data. This makes the presented method more practical. Furthermore, the convergence rate is improved significantly by sliding mode technique in finite time sense. The simulation results of the existing general nonlinear system illustrate the aforementioned features. Comparison with existing methods from literature proves the effectiveness of the proposed method.

DEES: A Class of Data-Enabled Robust Feedback Algorithms for Real-Time Optimization

The increasing availability of information-rich data sets offers an invaluable opportunity to complement and improve the performance of existing model-based feedback algorithms. Following this principle, in this paper we present a novel class of Data-Enabled Extremum Seeking (DEES) algorithms for static maps, which make use of current and recorded data in order to solve a convex optimization problem characterized by a variational inequality. The optimization dynamics synergistically combine ideas from concurrent learning and classic neuro-adaptive extremum seeking in order to dispense with the assumption of requiring a persistence of excitation condition in the closed-loop system. Using analytical tools for nonlinear systems we show that for a general class of optimization dynamics it is possible to tune the parameters of the controller to guarantee convergence in finite time to an arbitrarily small neighborhood of the set of optimizers. The results are illustrated in a scalar constrained minimization problem.

Real-Time Analysis Method and Application of Engagement in Online Independent Learning

Engagement is an important factor influencing the effectiveness and quality of online learning programs and how satisfied online learners are with the online learning experience. Therefore, when developing online learning programs, ways to increase engagement should be one of the top priorities. To determine the parameters that increase engagement of independent online learners, this paper selects three types of quantitative data by which to assess online learning engagement: video playing options, video lecture viewing time, and concurrent learning behaviors. This data, collected from real-time observation and analysis of authentic online learning of 14,000 adult learners, was used to run a real-time growth algorithm that determined a few key parameters that increase independent-learning engagement. A few primary results of this study are: 1) real-time, dynamically calculated data representing general or individual engagement parameters of independent learning can help both teachers and learners be aware of, recognize, and adjust learning status accurately and effectively; 2) the algorithm can identify optimal parameters for online learning by analyzing numerical values of engagement, rules, and characteristics of online learning that are difficult to be observed directly; and 3) it is necessary to dynamically analyze learners' engagement based on their learning processes and behaviors in response to different video lectures to know the effectiveness and feasibility of such materials and to support the design, production, and modification of optimal learning materials.

Conventional analysis of trial-by-trial adaptation is biased: Empirical and theoretical support using a Bayesian estimator.

Research on human motor adaptation has often focused on how people adapt to self-generated or externally-influenced errors. Trial-by-trial adaptation is a person’s response to self-generated errors. Externally-influenced errors applied as catch-trial perturbations are used to calculate a person’s perturbation adaptation rate. Although these adaptation rates are sometimes compared to one another, we show through simulation and empirical data that the two metrics are distinct. We demonstrate that the trial-by-trial adaptation rate, often calculated as a coefficient in a linear regression, is biased under typical conditions. We tested 12 able-bodied subjects moving a cursor on a screen using a computer mouse. Statistically different adaptation rates arise when sub-sets of trials from different phases of learning are analyzed from within a sequence of movement results. We propose a new approach to identify when a person’s learning has stabilized in order to identify steady-state movement trials from which to calculate a more reliable trial-by-trial adaptation rate. Using a Bayesian model of human movement, we show that this analysis approach is more consistent and provides a more confident estimate than alternative approaches. Constraining analyses to steady-state conditions will allow researchers to better decouple the multiple concurrent learning processes that occur while a person makes goal-directed movements. Streamlining this analysis may help broaden the impact of motor adaptation studies, perhaps even enhancing their clinical usefulness.

Integrating forests and biodiversity in Nepal's National Adaptation Plan: A review and synthesis of knowledge stock on opportunities and way forward

Climate change brings lasting changes in forests and biodiversity together with the ecosystem services altering its ability to support present and future economic activities. Current forest utilization and preservation is based on how forests developed under past climatic conditions. Policy-makers and forest managers must accept that climate change is inevitable and from which forests and forest communities are significantly impacted globally and in Nepal also, sustainable forest management (SFM) is already based on many measures to adapt to climate change as planned adaptation will reduce vulnerability at intervened sites and will have long term impacts. However, many forest species will be adapting autonomously and society will have to adjust to the result. Adaptation requires planning for change so that a suite of options for the future but based on the present practice and knowledge is to be available whenever needed. On the foundation of concurrent learning, knowledge and experiences of National Adaption Program of Action (NAPA) process, the National Adaptation Plan (NAP) process for forests and biodiversity will build medium and long-term adaptation strategies and plans with widely accepted objectives of future forests and biodiversity management. Banko JanakariA Journal of Forestry Information for NepalVol. 27, No. 2, 2017, page: 21-31

Scalable Multilabel Learning Based on Feature and Label Dimensionality Reduction

The data-driven management of real-life systems based on a trained model, which in turn is based on the data gathered from its daily usage, has attracted a lot of attention because it realizes scalable control for large-scale and complex systems. To obtain a model within an acceptable computational cost that is restricted by practical constraints, the learning algorithm may need to identify essential data that carries important knowledge on the relation between the observed features representing the measurement value and labels encoding the multiple target concepts. This results in an increased computational burden owing to the concurrent learning of multiple labels. A straightforward approach to address this issue is feature selection; however, it may be insufficient to satisfy the practical constraints because the computational cost for feature selection can be impractical when the number of labels is large. In this study, we propose an efficient multilabel feature selection method to achieve scalable multilabel learning when the number of labels is large. The empirical experiments on several multilabel datasets show that the multilabel learning process can be boosted without deteriorating the discriminating power of the multilabel classifier.

Directed-Hypergraph-Based Channel Allocation for Ultradense Cloud D2D Communications With Asymmetric Interference

In this paper, we investigate channel allocation for ultradense device-to-device (D2D) communications. Different from both binary graph and undirected hypergraph interference model, an improved directed hypergraph is applied to simultaneously represent cumulative and asymmetric interference aspects in the context of ultradense communications. We formulate the channel access problem in cloud D2D communication networks as a directed-hypergraph-based local altruistic game, which is proved to be an exact potential game. Then, a multiagent concurrent learning scheme in centralized-distributed fashion is proposed to search the optimal pure Nash equilibrium, which can also maximize the normalized network capacity. Finally, simulation results are presented to validate the proposed learning scheme.

Current Core Competencies Trend of Small to Medium Enterprises (SMEs) in China—A Concurrent Comprehensive Evaluation and Active Learning Study of Newly Listed Chinese Stocks from 2015 through 2017

With plenty of stocks newly listed in the Chinese stock market everyday, it becomes more and more important for managers and governess to examine the trend of core competencies for these companies. Since most companies of newly listed stocks are small to medium-sized enterprises, existing methods are not capable enough to evaluate their competitiveness. To provide an understanding for the trend of core competencies in the Chinese market, this article conducts a concurrent comprehensive evaluation and active learning methodology to analyze the newly listed stocks in SSE (Shanghai Stock Exchange Composite Index) and SZSE (Shenzhen Stock Exchange Component Index) from 2015 through 2017. There is an evidence that Number of Market Makers, Equity Financing Frequency and Executive Replacement Frequency are three main core competencies from 2015 through 2017. Authors contend that their findings in this paper question the quo of core competencies for small to medium-sized enterprises in the Chinese market.

Visual Servoing of Wheeled Mobile Robots Without Desired Images.

This paper proposes a novel monocular visual servoing strategy, which can drive a wheeled mobile robot to the desired pose without a prerecorded desired image. Compared with existing methods that adopt the teaching pattern for visual regulation, this scheme can still work well in the situation that the desired image has not been previously acquired. Thus, with the aid of this method, it is more convenient for mobile robots to execute visual servoing tasks. Specifically, to deal with nonexistence of the desired image, the reference frame is craftily defined by taking advantage of visual targets and the planar motion constraint, and the pose estimation algorithm is designed for the mobile robot with respect to the reference frame. Then, an adaptive visual regulation controller is developed to drive the mobile robot to the intermediate frame, where the parameter updating law is constructed for the unknown feature height based on the concurrent learning framework. Stability analysis shows that regulation errors and height identification error can converge simultaneously. Afterwards, the mobile robot is driven to the metric desired pose with the identified feature height. Both simulation and experimental results are provided to validate the performance of this strategy.

Optimal guaranteed cost control of uncertain non‐linear systems using adaptive dynamic programming with concurrent learning

In this study, the authors study the optimal guaranteed cost control problem for a class of non-linear uncertain systems based on adaptive dynamic programming (ADP) with concurrent learning. A neural network-based approximate optimal guaranteed cost control design is developed not only to ensure the system stability for all admissible uncertainties but also to achieve a minimal guaranteed cost. First, the optimal guaranteed cost control problem is transformed into an optimal control problem of the nominal system by properly modifying the cost function to account for all possible uncertainties. Then based on ADP, an adaptive optimal learning algorithm is proposed for the nominal system by using a single critic network with concurrent learning to approximate the solution of Hamilton–Jacobi–Bellman equation. To relax the demands of the persistence of excitation condition, the recorded past data are used simultaneously with the current data for the adaptation of the critic network weights. Besides, an additional adjusting term is employed to stabilise the system and relax the requirement for an initial stabilising control. Moreover, uniform ultimate boundedness of the closed-loop system is guaranteed by using the Lyapunov approach. Finally, three simulation examples are provided to demonstrate the effectiveness of the proposed approach.

Introducing Tina Spagnola

Introducing Tina Spagnola

Fault Detection for Dynamical Systems using Differential Geometric and Concurrent Learning Approach

A fault identification and estimation approach that brings together the differential geometric concept of observability codistribution with data-driven concurrent learning, is presented in this paper. In order to identify faults in the presence of unknown disturbances, we use the differential geometric approach to design a coordinate transformation, in order to find a subspace in which the effects of disturbances and system faults can be segregated. We then use concurrent learning to estimate the magnitude of the constant fault. We illustrate the approach to fault isolation for a spherical pendulum dynamics. We use Lyapunov stability analysis to show that the fault estimate by concurrent learning converges to the actual fault value, and then illustrate the design of a recovery controller.