Formation conditions of mutual adaptation in human-agent collaborative interaction

When an adaptive agent works with a human user in a collaborative task, in order to enable flexible instructions to be issued by ordinary people, it is believed that a mutual adaptation phenomenon can enable the agent to handle flexible mapping relations between the human user’s instructions and the agent’s actions. To elucidate the conditions required to induce the mutual adaptation phenomenon, we designed an appropriate experimental environment called “WAITER” (Waiter Agent Interactive Training Experimental Restaurant) and conducted two experiments in this environment. The experimental results suggest that the proposed conditions can induce the mutual adaptation phenomenon.     

Facility layout design using virtual multi-agent system

This paper presents a novel approach to the facility layout design problem based on multi-agent society where agents’ interactions form the facility layout design. Each agent corresponds to a facility with inherent characteristics, emotions, and a certain amount of money, forming its utility function. An agent’s money is adjusted during the learning period by a manager agent while each agent tries to tune the parameters of its utility function in such a way that its total layout cost can be minimized in competition with others. The agents’ interactions are formed based on market mechanism. In each step, an unoccupied location is presented to all applicant agents, for which each agent proposes a price proportionate to its utility function. The agent proposing a higher price is selected as the winner and assigned to that location by an appropriate space-filling curve. The proposed method utilizes the fuzzy theory to establish each agent’s utility function. In addition, it provides a simulation environment using an evolutionary algorithm to form different interactions among the agents and makes it possible for them to experience various strategies. The experimental results show that the proposed approach achieves a lower total layout cost compared with state of the art methods.     

A virtual human agent model with behaviour based on feeling exhaustion

A computational agent model for monitoring and control of a virtual human agent’s resources and exhaustion is presented. It models a physically grounded intelligent decision making process within the agent model for physical effort to be spent. Simulation results are discussed, and a formal analysis is presented on conditions under which the agent model functions properly, for example, such that it can be used to avoid running out of resources. The model is related to a model for monitoring or simulating a person’s heart rate. Finally some validation experiments are briefly discussed.     

Reinforcement learning technique using agent state occurrence frequency with analysis of knowledge sharing on the agent’s learning process in multiagent environments

Reinforcement learning techniques like the Q-Learning one as well as the Multiple-Lookahead-Levels one that we introduced in our prior work require the agent to complete an initial exploratory path followed by as many hypothetical and physical paths as necessary to find the optimal path to the goal. This paper introduces a reinforcement learning technique that uses a distance measure to the goal as a primary gauge for an autonomous agent’s action selection. In this paper, we take advantage of the first random walk to acquire initial information about the goal. Once the agent’s goal is reached, the agent’s first perceived internal model of the environment is updated to reflect and include said goal. This is done by the agent tracing back its steps to its origin starting point. We show in this paper, no exploratory or hypothetical paths are required after the goal is initially reached or detected, and the agent requires a maximum of two physical paths to find the optimal path to the goal. The agent’s state occurrence frequency is introduced as well and used to support the proposed Distance-Only technique. A computation speed performance analysis is carried out, and the Distance-and-Frequency technique is shown to require less computation time than the Q-Learning one. Furthermore, we present and demonstrate how multiple agents using the Distance-and-Frequency technique can share knowledge of the environment and study the effect of that knowledge sharing on the agents’ learning process.     

Design of Information Content and Layout for Process Control Based on Goal–Means Domain Analysis

With regard to the design of information content in information display, it is often claimed that the abstraction hierarchy (AH) of the work domain should be considered as a basis for identifying and structuring the information content. The primary advantage of AH-based analysis and design is that functionally abstracted information can systematically be identified and provided to the operator, which has rarely been presented in traditional displays. This study evaluated the effectiveness of providing functional information, which was abstracted and represented based on goal–means analysis along the AH, to the operator in two task situations (fault diagnosis and operation). The results showed that the operator’s performance was improved with the high-level information, and the latter’s utility became greater when the goal–means relations between information at different abstraction levels were exhibited. From the results, three design principles for information display can be drawn. First, information should be identified and displayed at multiple abstraction levels. Second, the goal–means relations among the abstraction levels should be explicitly presented, especially for analytical cognitive tasks. Third, information layout should support information integration along decomposition structure within an abstraction level as well as along abstraction levels.     

Temporal Languages for Epistemic Programs

This paper adds temporal logic to public announcement logic (PAL) and dynamic epistemic logic (DEL). By adding a previous-time operator to PAL, we express in the language statements concerning the muddy children puzzle and sum and product. We also express a true statement that an agent’s beliefs about another agent’s knowledge flipped twice, and use a sound proof system to prove this statement. Adding a next-time operator to PAL, we provide formulas that express that belief revision does not take place in PAL. We also discuss relationships between announcements and the new knowledge agents thus acquire; such relationships are related to learning and to Fitch’s paradox. We also show how inverse programs and hybrid logic each can be used to help determine whether or not an arbitrary structure represents the play of a game. We then add a past-time operator to DEL, and discuss the importance of adding yet another component to the language in order to prove completeness.     

Genomic control of agent morphogenesis in a physically correct virtual environment

Based on the analysis of the morphogenesis process of real biological organisms, the basic principles are formulated for the simulation of genomic control of the morphogenesis of a virtual agent in a physically correct environment. A model of the genome of a compound agent’s body composed of different functional subsystems is developed. A timing algorithm for genetically conditioned processes of regeneration and division of agent’s microcells is developed. The problem of forming agents with a morphology that is locally optimal for a given environment is formulated as a multi-generation optimization problem in a genetic algorithm. The fitness function is defined as the life duration of agents. Constraints are connected with the physical correctness of an environment and an energy deficit in the environment. Agent genomes are considered as chromosomes in the genetic algorithm. __________ Translated from Kibernetika i Sistemnyi Analiz, No. 2, pp. 42–54, March–April 2008.     

Towards an environment interface standard for agent platforms

We introduce an interface for connecting agent platforms to environments. This interface provides generic functionality for executing actions and for perceiving changes in an agent’s environment. It also provides support for managing an environment, e.g., for starting, pausing and terminating it. Among the benefits of such an interface are (1) standard functionality is provided by the interface implementation itself, and (2) agent platforms that support the interface can connect to any environment that implements the interface. This significantly reduces effort required from agent and environment programmers as the environment code needed to implement the interface needs to be written only once. We propose that the interface presented may be used as a standard that enables agents to control entities in environments. Our starting point for designing such a generic interface is based on a careful study of the various interfaces used by different agent programming languages to connect agent programs to environments. We discuss several case studies that use our interface (an elevator simulator, the well-known agent contest, and an implementation of the interface to connect agents to bots in Unreal Tournament 2004).     

Active adaptation in human-agent collaborative interaction

When a human user interacts with an adaptive agent to achieve human-agent collaboration, active adaptation is considered to be one of the critical characteristics of the agent. In order to investigate the principal features of active adaptation, we developed a human-agent collaborative experimental environment called WAITER (waiter agent interactive training experimental restaurant) and conducted two types of experiments, a Wizard of OZ (WOZ) agent experiment and an autonomous agent experiment. The objective of these experiments is to observe how human users perceive the agents and change their instructions when interacting with adaptive agents. The results indicate that humans can recognize changes in the agent’s actions and change their instruction methods accordingly. This implies that active adaptation of the agents may encourage the adaptation of the human users and help to build an adaptation loop between them. The experimental results also suggest that active adaptation may play an important role in a human-agent collaborative task.     

The Cortexionist architecture: behavioural intelligence of artificial creatures

Traditionally, producing intelligent behaviours for artificial creatures involves modelling their cognitive abilities. This approach raises two problems. On the one hand, defining manually the agent’s knowledge is a heavy and error-prone task that implies the intervention of the animator. On the other hand, the relationship between cognition and intelligence has not been theoretically nor experimentally proven so far. The ecological approaches provide a solution for these problems, by exploring the links between the creature, its body and its environment. Using an artificial life approach, we propose an original model of memory based on the synthesis of several neuroscience theories. The Cortexionist controller integrates cortex-like structure into a connectionist architecture in order to enhance the agent’s adaptation in a dynamic environment, ultimately leading to the emergence of intelligent behaviour. Initial experiments presented in this paper prove the validity of the model.     

Agent-Based Infrastructure and an Application to Internet Information Gathering

Pragmatic applications and studies of agent-based software engineering have evolved over the last decade. In order to explore how an agent is organized and applied, in this paper an agent framework is presented and applied to Internet information gathering. Agent systems are classified as micro or macro perspectives and agent applications are characterized by the four feature dimensions: behavior (user), knowledge (task), safety (time), and cooperation (social). An agent itself can be modeled according to the information, behavior, and organization aspects of the agent’s functional modules as proposed in this paper. A three-tier multi-agent and JAVA-implemented system, which coordinates information-gathering activities using KQML for inter-agent communication, is developed on the basis of the proposed architectural modules. Finally, we explore possible areas for future study. Received 1 February 1999 / Revised 30 June 2000 / Accepted in revised form 5 November 2000     

Normative Communication Models for Agent

An agent message is an attempted action upon the information state of the receiver that, if successful, would cause the receiver to move to a new information state. A model of normative communication can define when messages are not merely unsuccessful but instead are illegal or impossible actions upon the receiver’s internal state. The model uses the preconditions of the other core message types, coupled with a model of task interdependencies, agent roles, and belief-desire-intention elements, to define the preconditions for sending a canonical not-understood error message. By defining the space of messages that are legal actions on an agent’s internal state, a normative communication model also defines a set of ‘reasons’ that can accompany the error message. A not-understood error message signals a mismatch between agent interaction models and the accompanying reason opens the possibility for agents to realign their respective models. The paper discusses the matters arising from this possibility. This approach assumes that normative communication behavior reflects normative domain behavior. It also assumes that each agent accesses the normative model, in contrast with more centralized frameworks for defining normative interaction among agents and identifying interaction errors.     

From logic programming towards multi‐agent systems

In this paper we present an extension of logic programming (LP) that is suitable not only for the “rational” component of a single agent but also for the “reactive” component and that can encompass multi‐agent systems. We modify an earlier abductive proof procedure and embed it within an agent cycle. The proof procedure incorporates abduction, definitions and integrity constraints within a dynamic environment, where changes can be observed as inputs. The definitions allow rational planning behaviour and the integrity constraints allow reactive, condition‐action type behaviour. The agent cycle provides a resource‐bounded mechanism that allows the agent’s thinking to be interrupted for the agent to record and assimilate observations as input and execute actions as output, before resuming further thinking. We argue that these extensions of LP, accommodating multi‐theories embedded in a shared environment, provide the necessary multi‐agent functionality. We argue also that our work extends Shoham’s Agent0 and the BDI architecture. This revised version was published online in June 2006 with corrections to the Cover Date.     

RoboCup Rescue as multiagent task allocation among teams: experiments with task interdependencies

This paper addresses distributed task allocation among teams of agents in a RoboCup Rescue scenario. We are primarily concerned with testing different mechanisms that formalize issues underlying implicit coordination among teams of agents. These mechanisms are developed, implemented, and evaluated using two algorithms: Swarm-GAP and LA-DCOP. The latter bases task allocation on a comparison between an agent’s capability to perform a task and the capability demanded by this task. Swarm-GAP is a probabilistic approach in which an agent selects a task using a model inspired by task allocation among social insects. Both algorithms were also compared to another one that allocates tasks in a greedy way. Departing from previous works that tackle task allocation in the rescue scenario only among fire brigades, here we consider the various actors in the RoboCup Rescue, a step forward in the direction of realizing the concept of extreme teams. Tasks are allocated to teams of agents without explicit negotiation and using only local information. Our results show that the performance of Swarm-GAP and LA-DCOP are similar and that they outperform a greedy strategy. Also, it is possible to see that using more sophisticated mechanisms for task selection does pay off in terms of score.     

Learning to trust in the competence and commitment of agents

For agents to collaborate in open multi-agent systems, each agent must trust in the other agents’ ability to complete tasks and willingness to cooperate. Agents need to decide between cooperative and opportunistic behavior based on their assessment of another agents’ trustworthiness. In particular, an agent can have two beliefs about a potential partner that tend to indicate trustworthiness: that the partner is competent and that the partner expects to engage in future interactions. This paper explores an approach that models competence as an agent’s probability of successfully performing an action, and models belief in future interactions as a discount factor. We evaluate the underlying decision framework’s performance given accurate knowledge of the model’s parameters in an evolutionary game setting. We then introduce a game-theoretic framework in which an agent can learn a model of another agent online, using the Harsanyi transformation. The learning agents evaluate a set of competing hypotheses about another agent during the simulated play of an indefinitely repeated game. The Harsanyi strategy is shown to demonstrate robust and successful online play against a variety of static, classic, and learning strategies in a variable-payoff Iterated Prisoner’s Dilemma setting.     

Exploring agent support at the user interface in e-commerce applications

In this paper, we present some of the results from our ongoing research work in the area of ‘agent support’ for electronic commerce, particularly at the user interface level. Our goal is to provide intelligent agents to assist both the consumers and the vendors in an electronic shopping environment. Users with a wide variety of different needs are expected to use the electronic shopping application and their expectations about the interface could vary a lot. Traditional studies of user interface technology have shown the existence of a ‘gap’ between what the user interface actually lets the users do and the users’ expectations. Agent technology, in the form of personalized user interface agents, can help to narrow this gap. Such agents can be used to give a personalized service to the user by knowing the user’s preferences. By doing so, they can assist in the various stages of the users’ shopping process, provide tailored product recommendations by filtering information on behalf of their users and reduce the information overload. From a vendor’s perspective, a software sales agent could be used for price negotiation with the consumer. Such agents would give the flexibility offered by negotiation without the burden of having to provide human presence to an online store to handle such negotiations. Published online: 25 July 2001     

The case of sculpting atmospheres: towards design principles for expressive tangible interaction in control of ambient systems

According to the vision of Ambient Intelligence, technology will seamlessly merge into people’s everyday activities and environments. A challenge facing designers of such systems is to create interfaces that fit in people’s everyday contexts and incorporate the values of daily life. This paper focuses on tangible expressive interaction as one possible approach towards linking everyday experiences to intuitive forms of interaction and presents a number of principles for expressive interaction design in this field. A case study of a tangible expressive interface to control a living room atmosphere projection system (orchestrating living room lighting, audio and video-art) is presented to illustrate and reflect upon the design principles. Furthermore, the case study describes possible techniques towards integrating the design principles into a design method.