Search results for "Autonomous agent"
showing 10 items of 25 documents
Agent's actions as a classification criteria for the state space in a learning from rewards system
2008
We focus in this paper on the problem of learning an autonomous agent's policy when the state space is very large and the set of actions available is comparatively short. To this end, we use a non-parametric decision rule (concretely, a nearest-neighbour strategy) in order to cluster the state space by means of the action that leads to a successful situation. Using an exploration strategy to avoid greedy behaviour, the agent builds clusters of positively-classified states through trial and error learning. In this paper, we implement a 3D synthetic agent which plays an 'avoid the asteroid' game that suits our assumptions. Using as the state space a feature vector space extracted from a visua…
Ethics of Artificial Intelligence : Research Challenges and Potential Solutions
2020
Artificial Intelligence (AI) is a rapidly emerging paradigm with many applications in healthcare, industries, and smart cities. However, this rise of global interest in AI has fueled a renewed interest from the public sector and global policymakers. As AI networks (e.g., chatbots, automation systems, and helping agents) are paving their way as interactive household items, a critically important research issue is understanding the ethical impact of these autonomous agents. What is the explanation of the AI decision-making process? What are the legal, societal, and moral consequences of these decisions and actions? Should these AI systems be allowed to make decisions for human beings and to w…
Interpretable Option Discovery Using Deep Q-Learning and Variational Autoencoders
2021
Deep Reinforcement Learning (RL) is unquestionably a robust framework to train autonomous agents in a wide variety of disciplines. However, traditional deep and shallow model-free RL algorithms suffer from low sample efficiency and inadequate generalization for sparse state spaces. The options framework with temporal abstractions [18] is perhaps the most promising method to solve these problems, but it still has noticeable shortcomings. It only guarantees local convergence, and it is challenging to automate initiation and termination conditions, which in practice are commonly hand-crafted.
Evolution and Learning: Evolving Sensors in a Simple MDP Environment
2003
Natural intelligence and autonomous agents face difficulties when acting in information-dense environments. Assailed by a multitude of stimuli they have to make sense of the inflow of information, filtering and processing what is necessary, but discarding that which is unimportant. This paper aims at investigating the interactions between evolution of the sensorial channel extracting the information from the environment and the simultaneous individual adaptation of agent-control. Our particular goal is to study the influence of learning on the evolution of sensors, with learning duration being the tunable parameter. A genetic algorithm governs the evolution of sensors appropriate for the a…
A Binary Particle Swarm Optimization Algorithm for a Double Auction Market
2007
In this paper, we shall show the design of a multi-unit double auction (MDA) market. It should be enough robust, flexible and sufficiently efficient in facilitating exchanges. In a MDA market, sellers and buyers submit respectively asks and bids. A trade is made if a buyers bid exceeds a sellers ask. A sellers ask may match several buyers bids and a buyers bid may satisfy several sellers asks. The trading rule of a market defines the organization, information exchange process, trading procedure and clearance rules of the market. The mechanism is announced before the opening of the market so that every agent knows how the market will operate in advance. These autonomous agents pursue their o…
A heuristic for problem formalization in agent based simulation studies
2015
Agent Based Modeling and Simulation (ABMS) is considered an effective approach for conducting simulation studies in many fields. In order to develop high quality simulation models, methodological approaches are demanded. In such direction we are moving by proposing a heuristic for the formalization of agent based simulation problems. The proposed heuristic is based on some guidelines developed for identifying the main elements of the problem domain description by analysing verbs and their common taxonomy in grammar.
A Distributed Framework for Scalable Large-Scale Crowd Simulation
2007
Emerging applications in the area of Emergency Response and Disaster Management are increasingly demanding interactive capabilities to allow for the quick understanding of a critical situation, in particular in urban environments. A key component of these interactive simulations is how to recreate the behavior of a crowd in real- time while supporting individual behaviors. Crowds can often be unpredictable and present mixed behaviors such as panic or aggression, that can very rapidly change based on unexpected new elements introduced into the environment. We present preliminary research specifically oriented towards the simulation of large crowds for emergency response and rescue planning s…
Autonomous agent system using dispatching rules in the negotiation protocol
2002
In this paper, the most important results obtained by the simulated application of autonomous agent paradigms to a. real factory are presented. The classical rules of dispatching are compared with the autonomous agents approach. In particular, the possibility of redesigning the negotiation rules in terms of currency in order to take into account even non-time-related costs is considered. Finally, a new project on the effective application of the autonomous agent system to a test bed, modelling a simplified firm, is proposed.
Intelligent Energy Management System
2009
Energy management is nowadays a subject of great importance and complexity. It consists in choosing among a set of sources able to produce energy that will give energy to a set of loads by minimising losses and costs. The sources and loads are heterogeneous, distributed and the reaction of the system, the choice of sources, must be done in real-time to avoid power outage. The goal of this paper is to present a system able to self-regulate a heterogeneous set of power sources and loads organised as a coherent group of entities that is called micro-grid, in order to optimize several criteria such as: cost and efficiency. This system is based upon the Multi-Agent Systems paradigm. Each micro-g…
Perceptual Social Dimensions of Human - Humanoid Robot Interaction
2013
The present paper aims at a descriptive analysis of the main perceptual and social features of natural conditions of agent interaction, which can be specified by agent in human-humanoid robot interaction. A principled approach to human-robot interaction may be assumed to comply with the natural conditions of agents overt perceptual and social behaviour. To validate our research we used the minimalistic humanoid robot Telenoid. We have conducted human-robot interactions test with people with no prior interaction experience with robot. By administrating our questionnaire to subject after well defined experimental conditions, an analysis of significant variance correlation among dimensions in …