Multi-agent Interaction Research Articles

Microscopic Road Traffic Scene Analysis Using Computer Vision and Traffic Flow Modelling

This paper presents the development of a vision-based system for microscopic road traffic scene analysis and understanding using computer vision and computational intelligence techniques. The traffic flow model is calibrated using the information obtained from the road-side cameras. It aims to demonstrate an understanding of different levels of traffic scene analysis from simple detection, tracking, and classification of traffic agents to a higher level of vehicular and pedestrian dynamics, traffic congestion build-up, and multi-agent interactions. The study used a video dataset suitable for analysis of a T-intersection. Vehicle detection and tracking have 88.84% accuracy and 88.20% precision. The system can classify private cars, public utility vehicles, buses, and motorcycles. Vehicular flow of every detected vehicles from origin to destination are also monitored for traffic volume estimation, and volume distribution analysis. Lastly, a microscopic traffic model for a T-intersection was developed to simulate a traffic response based on actual road scenarios.

An Investment Decision Model of Distribution Network Planning Based on Correlation Mining of Reconstruction Measures and Loss Load Index

In recent years, how to build a reliable, high-efficient, high-tech and environmental-friendly distribution network infrastructure and service system, especially to accurately assess the reliability impact of various technical measures and establish optimal investment decision model for the future distribution network planning, has great theoretical and practical significance. However, the traditional analysis of investment decision model of distribution network is based on complex power flow calculation, which is not suitable for current construction of large and smart distribution network with multi-agent interaction under electrical market environment. Therefore, in this paper, based on the sample data, the correlation mining methods of the reconstruction measures and loss load index, such as neural network method, are applied to analyse the relationship between different types of reconstruction measures and loss load index. Based on the correlation mining methods, the relationship between the reconstruction measures and loss load index can be obtained and the value of loss load can be fast predicted. Through the comparison the relationship between reconstruction measures and loss load index, the optimal reconstruction measures can be chosen, and the data relationship based accurate investment decision model can be built. Experimental result shows the accuracy and effectiveness of the presented methodology.

INTELLECTUAL CAPITAL DEVELOPMENT WITHIN THE CONDITIONS OF THE INFORMATION SOCIETY

In the era of rapid development of the information society, there is a problem of accumulation and rational use of intellectual capital, as well as the need to ensure its development. Now it is impossible to carry out the effective activity of enterprises without technologies, know-how, inventions, knowledge and experience of workers. Since world globalization and rapid transformation of the domestic economy led to some kind of economic and financial instability, which leads to an increase in interest in the research and use of information, it becomes necessary to study the possibilities of providing intellectual capital development in entrepreneurship and in the country. It should also be noted that sufficient funding for the development of intellectual potential leads to increased competitiveness of the country as a whole. Therefore, intellectual capital in a society with information as the main source holds key positions and is at the centre of a number of issues related to the development of entrepreneurship in the country, the development of the information society, the knowledge society, and the digital economy. Taking into account the cultural and value aspect and the explicit and implicit problems of multi-agent interaction in the information society, we believe that the approaches to the definition of the “information society” should not be antagonistic to each other, but symbiotic, and integrate the workings of each area of research of scientists, taking into account rapid development, globalization and informational synergistic effects that arise in recent years, it can be argued that the concept will be repeatedly transformed. It is investigated that, in the context of the information society, the concept of “intellectual capital” appears. We determine that intellectual capital is an aggregate of knowledge, experience, skills, creativity, abilities, relationships, accumulated in the process of intellectual activity, that have economic value and are used in the process of production and exchange for the purpose of obtaining income. The conducted research allows generalizing the conclusion that exactly the development of information and communication technologies has led to the emergence of an information society, which was an impetus for the development of e-economy, which precisely determined the emergence of e-commerce as its inherent part. It is proven that financing for the development of intellectual capital is required for many reasons; however, one of the keys is to raise the level of competitiveness not only of enterprises but of the state as a whole. Thus, the intellectual potential of citizens is aimed at the economic, technical, and cultural development of enterprises and their own state. Therefore, the support and development of the information society and its main good, namely, information and knowledge, should be actively implemented in enterprises for further development of entrepreneurship, the economy of the country and the state. That is why the research of models of financial provision of intellectual capital, which was carried out in the article (investment, crowdfunding, innovative), are defined as perspective directions of development of science and practice.

A Layered-Coevolution-Based Attribute-Boosted Reduction Using Adaptive Quantum-Behavior PSO and Its Consistent Segmentation for Neonates Brain Tissue

The main challenge of attribute reduction in large data applications is to develop a new algorithm to deal with large, noisy, and uncertain large data linking multiple relevant data sources, structured or unstructured. This paper proposes a new and efficient layered-coevolution-based attribute-boosted reduction algorithm (LCQ-ABR*) using adaptive quantum-behavior particle swarm optimization (PSO). First, the quantum rotation angle of an evolutionary particle is updated by a dynamic change of self-adapting step size. Second, a self-adaptive partitioning strategy is employed to group particles into different memeplexes, and the quantum-behavior mechanism with the particles’ states depicted by the wave function cooperates to achieve superior performance in their respective memeplexes. Third, a new layered coevolutionary model with multiagent interaction is constructed to decompose a complex attribute set, and it can self-adapt the attribute sizes among different layers and produce the reasonable decompositions by exploiting any interdependence among multiple relevant attribute subsets. Fourth, the decomposed attribute subsets are evolved to compute the positive region and discernibility matrix by using their best quantum particles, and the global optimal reduction set is induced successfully. Finally, extensive comparative experiments are provided to illustrate that LCQ-ABR* has better feasibility and effectiveness of attribute reduction on large-scale and uncertain dataset problems with complex noise as compared with representative algorithms. Moreover, LCQ-ABR* can be successfully applied in the consistent segmentation for neonatal brain three-dimensional MRI, and the consistent segmentation results further demonstrate its stronger applicability.

Sequential Decision Making in Artificial Musical Intelligence

My main research motivation is to develop complete autonomous agents that interact with people socially. For an agent to be social with respect to humans, it needs to be able to parse and process the multitude of aspects that comprise the human cultural experience. That in itself gives rise to many fascinating learning problems. I am interested in tackling these fundamental problems from an empirical as well as a theoretical perspective. Music, as a general target domain, serves as an excellent testbed for these research ideas. Musical skills---playing music (alone or in a group), analyzing music or composing it---all involve extremely advanced knowledge representation and problem solving tools. Creating "musical agents"---agents that can interact richly with people in the music domain---is a challenge that holds the potential of advancing social agents research, and contributing important and broadly applicable AI knowledge. This belief is fueled not just by my background in computer science and artificial intelligence, but also by my deep passion for music as well as my extensive musical training. One key aspect of musical intelligence which hasn’t been sufficiently studied is that of sequential decision-making. My thesis strives to answer the following question: How can a sequential decision making perspective guide us in the creation of better music agents, and social agents in general? More specifically, this thesis focuses on two aspects of musical intelligence: music recommendation and multiagent interaction in the context of music.

Video Event Recognition and Anomaly Detection by Combining Gaussian Process and Hierarchical Dirichlet Process Models

In this paper, we present an unsupervised learning framework for analyzing activities and interactions in surveillance videos. In our framework, three levels of video events are connected by Hierarchical Dirichlet Process (<small>HDP</small>) model: low-level visual features, simple atomic activities, and multi-agent interactions. Atomic activities are represented as distribution of low-level features, while complicated interactions are represented as distribution of atomic activities. This learning process is unsupervised. Given a training video sequence, low-level visual features are extracted based on optic flow and then clustered into different atomic activities and video clips are clustered into different interactions. The <small>HDP</small> model automatically decides the number of clusters, i.e., the categories of atomic activities and interactions. Based on the learned atomic activities and interactions, a training dataset is generated to train the Gaussian Process (<small>GP</small>) classifier. Then, the trained <small>GP</small> models work in newly captured video to classify interactions and detect abnormal events in real time. Furthermore, the temporal dependencies between video events learned by HDP-Hidden Markov Models (<small>HMM</small>) are effectively integrated into <small>GP</small> classifier to enhance the accuracy of the classification in newly captured videos. Our framework couples the benefits of the generative model (<small>HDP</small>) with the discriminant model (<small>GP</small>). We provide detailed experiments showing that our framework enjoys favorable performance in video event classification in real-time in a crowded traffic scene.

Construction of multi-agent mobile robots control system in the problem of persecution with using a modified reinforcement learning method based on neural networks

A method for constructing a multi-agent control system for mobile robots based on training with reinforcement using deep neural networks is considered. Synthesis of the management system is proposed to be carried out with reinforcement training and the modified Actor-Critic method, in which the Actor module is divided into Action Actor and Communication Actor in order to simultaneously manage mobile robots and communicate with partners. Communication is carried out by sending partners at each step a vector of real numbers that are added to the observation vector and affect the behaviour. Functions of Actors and Critic are approximated by deep neural networks. The Critics value function is trained by using the TD-error method and the Actor’s function by using DDPG. The Communication Actor’s neural network is trained through gradients received from partner agents. An environment in which a cooperative multi-agent interaction is present was developed, computer simulation of the application of this method in the control problem of two robots pursuing two goals was carried out.

An ant-inspired model for multi-agent interaction networks without stigmergy

The aim of this work is to construct a microscopic model of multi-agent interaction networks inspired by foraging ants that do not use pheromone trails or stigmergic traces for communications. The heading and speed of each agent is influenced by direct interactions or encounters with other agents. Each agent moves in a plane using a correlated random walk whose probability distribution for heading change is made adaptable to these interactions and is superimposed with probability distributions that emulate how ants remember nest and food source locations. The speed of each agent is likewise influenced by a superposition of impetus and resistance effects that arise from its recent interactions. Additionally, the agents use a quorum sensing mechanism to trigger a non-deterministic decentralized congestion avoidance scheme. A discrete-time non-deterministic recruitment model is adopted and incorporated to regulate the population of foraging agents based on the amount of food perceived to exist in the environment. Simulation experiments were conducted to evaluate and demonstrate how agents employ the interaction network when foraging in open and closed environments as well as in scenarios with narrow pathways that trigger congestion.

Efficient and Robust Emergence of Norms through Heuristic Collective Learning

In multiagent systems, social norms serves as an important technique in regulating agents’ behaviors to ensure effective coordination among agents without a centralized controlling mechanism. In such a distributed environment, it is important to investigate how a desirable social norm can be synthesized in a bottom-up manner among agents through repeated local interactions and learning techniques. In this article, we propose two novel learning strategies under the collective learning framework, collective learning EV-l and collective learning EV-g , to efficiently facilitate the emergence of social norms. Extensive simulations results show that both learning strategies can support the emergence of desirable social norms more efficiently and be applicable in a wider range of multiagent interaction scenarios compared with previous work. The influence of different topologies is investigated, which shows that the performance of all strategies is robust across different network topologies. The influences of a number of key factors (neighborhood size, actions space, population size, fixed agents and isolated subpopulations) on norm emergence performance are investigated as well.

Formal modelling of multi-agent interaction dynamics in biological systems

Formal modelling of multi-agent interaction dynamics in biological systems

To Pass or Not to Pass: Modeling the Movement and Affordance Dynamics of a Pick and Place Task.

Humans commonly engage in tasks that require or are made more efficient by coordinating with other humans. In this paper we introduce a task dynamics approach for modeling multi-agent interaction and decision making in a pick and place task where an agent must move an object from one location to another and decide whether to act alone or with a partner. Our aims were to identify and model (1) the affordance related dynamics that define an actor's choice to move an object alone or to pass it to their co-actor and (2) the trajectory dynamics of an actor's hand movements when moving to grasp, relocate, or pass the object. Using a virtual reality pick and place task, we demonstrate that both the decision to pass or not pass an object and the movement trajectories of the participants can be characterized in terms of a behavioral dynamics model. Simulations suggest that the proposed behavioral dynamics model exhibits features observed in human participants including hysteresis in decision making, non-straight line trajectories, and non-constant velocity profiles. The proposed model highlights how the same low-dimensional behavioral dynamics can operate to constrain multiple (and often nested) levels of human activity and suggests that knowledge of what, when, where and how to move or act during pick and place behavior may be defined by these low dimensional task dynamics and, thus, can emerge spontaneously and in real-time with little a priori planning.

Communicative knowledge pervasively influences sensorimotor computations

Referential pointing is a characteristically human behavior, which involves moving a finger through space to direct an addressee towards a desired mental state. Planning this type of action requires an interface between sensorimotor and conceptual abilities. A simple interface could supplement spatially-guided motor routines with communicative-ostensive cues. For instance, a pointing finger held still for an extended period of time could aid the addressee’s understanding, without altering the movement’s trajectory. A more complex interface would entail communicative knowledge penetrating the sensorimotor system and directly affecting pointing trajectories. We compare these two possibilities using motion analyses of referential pointing during multi-agent interactions. We observed that communicators produced ostensive cues that were sensitive to the communicative context. Crucially, we also observed pervasive adaptations to the pointing trajectories: they were tailored to the communicative context and to partner-specific information. These findings indicate that human referential pointing is planned and controlled on the basis of partner-specific knowledge, over and above the tagging of motor routines with ostensive cues.

Punishment in the form of shared cost promotes altruism in the cooperative dilemma games

One phenomenon or social institution often observed in multi-agent interactions is the altruistic punishment, i.e. the punishment of unfair behavior by others at a personal cost. Inspired by the works focusing on punishment and the intricate mechanism behind it, we theoretically study the strategy evolution in the framework of two-strategy game models with the punishment on defectors, moreover, the cost of punishing will be evenly shared among the cooperators. Theoretical computations suggest that larger punishment on defectors or smaller punishment cost incurred by cooperators will enhance the fixation of altruistic cooperation in the population. Through the replicate dynamics, the group size of the randomly selected individuals from the sufficiently large population will notably affect the strategy evolution in populations nested within a dilemma. By theoretical modeling the concept of shared cost for punishment from one point of view, our findings underscore the importance of punishment with shared cost as a factor in real-life decisions in an evolutionary game context.

Reciprocal Social Strategy in Social Repeated Games and Emergence of Social Norms

In an artificial society where agents repeatedly interact with one another, effective coordination among agents is generally a challenge. This is especially true when the participating agents are self-interested, and that there is no central authority to coordinate, and direct communication or negotiation are not possible. Recently, the problem was studied in a paper by Hao and Leung, where a new repeated game mechanism for modeling multi-agent interactions as well as a new reinforcement learning based agent learning method were proposed. In particular, the game mechanism differs from traditional repeated games in that the agents are anonymous, and the agents interact with randomly chosen opponents during each iteration. Their learning mechanism allows agents to coordinate without negotiations. The initial results had been promising. However, extended simulation also reveals that the outcomes are not stable in the long run in some cases, as the high level of cooperation is eventually not sustainable. In this work, we revisit he problem and propose a new learning mechanism as follows. First, we propose an enhanced Q-learning-based framework that allows the agents to better capture both the individual and social utilities that they have learned through observations. Second, we propose a new concept of \social attitude" for determining the action of the agents throughout the game. Simulation results reveal that this approach can achieve higher social utility, including close-to-optimal results in some scenarios, and more importantly, the results are sustainable with social norms emerging.

Special issue on multiagent interaction without prior coordination: guest editorial

This special issue of the Journal of Autonomous Agents and Multi-Agent Systems sought research articles on the emerging topic of multiagent interaction without prior coordination. Topics of interest included empirical and theoretical investigations of issues arising from assumptions of prior coordination, as well as solutions in the form of novel models and algorithms for effective multiagent interaction without prior coordination.

Time-informed task planning in multi-agent collaboration

Human-robot collaboration requires the two sides to coordinate their actions in order to better accomplish common goals. In such setups, the timing of actions may significantly affect collaborative performance. The present work proposes a new framework for planning multi-agent interaction that is based on the representation of tasks sharing a common starting and ending point, as petals in a composite daisy graph. Coordination is accomplished through temporal constraints linking the execution of tasks. The planner distributes tasks to the involved parties sequentially. In particular, by considering the properties of the available options at the given moment, the planner accomplishes locally optimal task assignments to agents. Optimality is supported by a fuzzy theoretic representation of time intervals which enables fusing temporal information with other quantitative HRI aspects, therefore accomplishing a ranking of the available options. The current work aims at a systematic experimental assessment of the proposed framework is pursued, verifying that it can successfully cope with a wide range of HRI scenarios.

A Compact Autonomous Underwater Vehicle With Cephalopod-Inspired Propulsion

AbstractIn this paper, a bioinspired, compact, cost-effective autonomous underwater vehicle system is presented. Designed to operate in a heterogeneous, multivehicle collaboration hierarchy, the presented vehicle design features 3D printing technology to enable fast fabrication with a complex internal structure. Similar to a previous vehicle prototype, this system generates propulsive forces by expelling unsteady, pulsed jets, inspired by the locomotion of cephalopods and jellyfish. The novel thrusters enable the vehicle to be fully actuated in horizontal plane motions, without sacrificing the low-forward-drag, slender vehicle profile. By successively ingesting water and expelling finite water jets, periodic actuation forces are generated at all possible vehicle velocities, eliminating the need for control surfaces used in many conventional underwater vehicle designs. A semiactive buoyancy control system, inspired by the nautilus, adjusts the vehicle depth by passively allowing water flowing into and actively expelling water out of an internal bladder. A compact embedded system is developed to achieve the control and sensing capabilities necessary for multiagent interactions with the minimum required processing power and at a low energy cost. The new vehicle design also showcases an underwater optical communication system for short-range, high-speed data transmission, supplementing the conventional acoustic communication system. Experimental results show that, with the thruster motors powered at a 60% duty-cycle, the new vehicle is able to achieve a 1/4 zero-radius turn in 3.5 s and one-body-width sway translation in 2.5 s.

Interdisciplinarity and multiagent interactions for innovations in horticulture – paradigms beyond the words

The progression towards sustainable horticulture is usually associated with scientific and methodological breakthroughs. As a consequence, the innovation processes in the fields of biology and ecology and their agronomic implementation are increasingly recognized as main drivers to improve horticultural systems. At the same time, a horticultural system cannot be designed without good knowledge of the social and economic contexts in which it is embedded; it has to be considered as an integrated social-ecological-agrosystem. For example, the cross-knowledge derived from interdisciplinary research constitutes the backbone for building ecological lybased crop ideotypes and cropping systems, including genetics, agronomy and social environmental levers and constraints. Designing new horticultural systems in such a systemic framework is also relevant to develop and better implement knowledge exchange among agents. This strong intertwining of the different research fields poses new challenges. There is a need for more participatory research and scaling-up to achieve adoption of innovations along the horticultural chain. Thus, the former linear and top-down scheme, i.e., from basic science to applied science and fieldwork, is now reconsidered and the design of innovative agricultural production systems is viewed as an integrated, interactive and participatory process where agents are dynamically interacting. The new challenge is to better combine the detailed knowledge typical of the dominant reductionist paradigm generally oriented towards the one-size-fits-all objective and the paradigm of complexity where the custom-fit approach predominates. Progressing towards these frameworks of knowledge and relationships among agents poses epistemological questions about inter-disciplinarity and hybridization between scientific and non-scientific (advisor, grower) knowledge. (Resume d'auteur)

Editorial: How Humans Recognize Objects: Segmentation, Categorization and Individual Identification.

Editorial: How Humans Recognize Objects: Segmentation, Categorization and Individual Identification.

Modelling complex systems of heterogeneous agents to better design sustainability transitions policy

This article proposes a fundamental methodological shift in the modelling of policy interventions for sustainability transitions in order to account for complexity (e.g. self-reinforcing mechanisms, such as technology lock-ins, arising from multi-agent interactions) and agent heterogeneity (e.g. differences in consumer and investment behaviour arising from income stratification). We first characterise the uncertainty faced by climate policy-makers and its implications for investment decision-makers. We then identify five shortcomings in the equilibrium and optimisation-based approaches most frequently used to inform sustainability policy: (i) their normative, optimisation-based nature, (ii) their unrealistic reliance on the full-rationality of agents, (iii) their inability to account for mutual influences among agents (multi-agent interactions) and capture related self-reinforcing (positive feedback) processes, (iv) their inability to represent multiple solutions and path-dependency, and (v) their inability to properly account for agent heterogeneity. The aim of this article is to introduce an alternative modelling approach based on complexity dynamics and agent heterogeneity, and explore its use in four key areas of sustainability policy, namely (1) technology adoption and diffusion, (2) macroeconomic impacts of low-carbon policies, (3) interactions between the socio-economic system and the natural environment, and (4) the anticipation of policy outcomes. The practical relevance of the proposed methodology is subsequently discussed by reference to four specific applications relating to each of the above areas: the diffusion of transport technology, the impact of low-carbon investment on income and employment, the management of cascading uncertainties, and the cross-sectoral impact of biofuels policies. In conclusion, the article calls for a fundamental methodological shift aligning the modelling of the socio-economic system with that of the climatic system, for a combined and realistic understanding of the impact of sustainability policies.