Projects tagged ‘agent’ and ‘simulation’

Repast is a free and open source agent-based modeling toolkit that offers users a rich variety of features.

Regional Multi Agent Simulator (RegMAS) is a spatially explicit, multi-agent representation of rural systems with an emphasis on the agriculture. It is a flexible model framework (with GUI) that ... [More] allows simulation of any regions, once needed data is available. We use as input data land-use information as Corine Land Cover maps and, optionally, a Digital Terrain Model (DTM) of the region under simulation. [Less]

In the simulation world, there are two well known types of simulation frameworks: Discrete Event Simulation (or DES) and Multi-Agent Simulation (or MAS). Both have valuable properties on their own: ... [More] DES has a good notion of time progress and MAS is a good way to think in role separation, locality and asynchronously. However, If you want to simulate a real world operation, you need both at once: time progress and separate agents communicating with each other. For example, a manufacturing process, where the manufacturing agents do a planning step, then execute their plan. The planning is done via communication, and the execution is done (skipped) via virtual time elapse. However, it might be that the virtual time elapse is not uniform among the participants: some do things faster, some slower. There are tons of scientific papers that imply the existence of such of a combined framework, however, there are very little to none about the implementation details of those. The most typical way to implement this to do a discrete time advancement and a real time limit how long the agents can communicate with each other: The time advances by 1 minute, and the agents have 100ms to do their communication per simulation step. This way, achieving a 600x speedup. However, this might seem a waste of time if the agents are quick, or worse: the agents run out of the time slice. There should be a better way to do this. I'm sure there are lots of non-free environments which might do this nicely, but they are non-free, and probably not written in Java 7. I decided to create a framework to do just that cheaply and efficiently. I was inspired by the scientific paper Multi-Agent System Simulation Framework by Bryan Horling et al. (2000). The TiDrA - Time Driven Agent Framework was born. The TiDrA framework is (currently) rather small, but effective. It utilizes the excellent concurrency primitives of Java 6 and beyond, but the implementation is aimed at Java 7. You can download just the framework JAR, or the version, where the FindBugs and JSR 305 annotations for software defect detection (latest as of October 2. 2009) annotations are also included. [Less]

Trata-se de um trabalho realizado por Heliomar Kann e Roberto Weidmann, alunos da UFF (Universidade Federal Fluminense) na disciplina Introdução a Sistemas Multi-Agentes. A idéia consiste em ... [More] simular o comportamento de residências com necessidades de de faxina. Os moradores solicitam faxinas para uma Central de Serviços, que por sua vez possui uma secretária que atende as solicitações e as coloca em um "mural" no qual as agentes empregadas periodicamente verificam se há solicitações pendentes. Caso positivo a empregada que verificou primeiro a solicitação desloca-se para a residência e faz seu serviço dentro de um tempo pré-determinado. O programa também gera gráficos do aprendizado dos agentes moradores. Esse aprendizado ocorre quando o agente consegue atingir seu objetivo. [Less]

SimPatrol is a software simulator of multiagent systems constructed for the task of patrolling. References about the task of patrolling (temporary): http://www.cin.ufpe.br/~glr/patrolling Sponsored ... [More] byUFPE/CIn (http://www.cin.ufpe.br) Powered byJava (http://www.sun.com/java) Eclipse (http://www.eclipse.org) Colt (http://dsd.lbl.gov/~hoschek/colt/) [Less]

visit www.nevu.de too! Browse the source!

Framework zum Simulieren von intelligenten Robotic Agenten in einer virtuellen Umgebung, mit besonderem Augenmerk auf die Maschinelle Bildverarbeitung

For academics purposes in AI teaching and researches.

Kdot is an agent-based simulation of a market economy as a complex adaptive system. Although computational modeling of economies seems to be an excellent avenue of research, little work has been done. ... [More] Perhaps the most sophisticated models are present in online video games, using human game-players instead of autonomous simulated market-participants. We seek to implement a robust and extensible tool to enable experimentation with a simulated economy. We will start small. The Solow model of exogenous growth is perhaps the simplest macroeconomic model. During each period of time, homogeneous households work, consume, and save. Production in each period is a function of the number of workers and the amount of capital in the economy. The households consume a portion of production and save the rest as capital for the next period. In our simulation, the households will use reinforcement learning techniques to pick a saving rate that maximizes net present utility--in this case consumption. A successful implementation of the Solow model will serve as a stepping stone for more sophisticated models. Why is this interesting? Economists have few ways to test their theories, relying on sparse data collected over decades. There are millions of confounding factors. Most importantly, in the real world, we have no way of creating the same situation twice. We want to do just that--enable experimentation in economics. ... Actually, right now I'm focusing on a model described by Caucutt, et al. (2006). The economy has one firm/entrepreneur and two types of workers--skilled and unskilled. The workers live for one period as an adult, working and raising a child. The adults choose to invest a portion of their income in the child's education. The more money spent, the better the chance of that child being skilled when it becomes an adult the next period. Each adult's utility is a function of its own consumption as an adult and its child's consumption as an adult. Caucutt's paper does some balanced growth path analysis and concludes that a progressive tax is likely to hinder economic growth and is likely to increase inequality. However, the paper does not discuss the dynamics of the model. The simulation I am building is an attempt to study the dynamics of this model. [Less]

Simulates 2D terrain water flow, weather, plant growth as well as animal senses and behavior. Uses Chipmunk physics library and Clanlib for GUI. This is not your typical "matrix of colorful pixels" ... [More] alife simulation. More like aiming to be environment for persistent and meaninful video game world. [Less]