One particularly fertile field of research using these techniques will be investigations into situations where behavior of individuals (humans, vehicles, animals, etc.) have impacts on their environment, and the impacted (or changed) environment in turn affects the behavior of the individuals. In such situations, knowledge of processes at one level cannot predict what will happen at a higher level: for example, knowledge of a particular cow's physiology and eating habits will not be enough to explain which part of a field or rangeland will suffer from overgrazing. A similar analogy can be made for human use of the environment.
One problem of the recent advances in artificial life is that they are bringing in issues that are not really "new". Many questions being studied via these methods have already been modeled using established analytical techniques, though often at the expense of impossibly unrealistic assumptions. While the actual contributions of these modeling frameworks to spatial analysis is not clear, there seems to be great promise and considerable excitement.
I am especially interested in attending a forum where people with knowledge and experience in this field can meet and disagree with me.
Before my appointment here at USU I worked as a graduate assistant at University of Florida, where I worked with GIS for coastal and recreational boat traffic management for Florida. The research there, funded by Florida Sea Grant, was primarily extension-based; spatial databases were developed for Florida coastal counties, and a series of extension manuals and technical reports were published that described implementations of waterway management schemes. I worked on simulations of recreational boat traffic to aid in management decisions, and developed many techniques for integrating agent-based simulations into GIS datasets.
My duties with my appointment here at USU have been to continue research in integration of agent-based simulations and spatial datasets. I have designed a simulation environment where GIS layers are read into a "landscape object". In this landscape, individual agents can be created to roam about the landscape. They can read information from the landscape as they pass over pixels or regions, and can modify pixels as they pass over them, and modify their behavior based on the environment as it changes. In addition, the pixels themselves are objects, which contain information about the various data layers occupying their space. The pixel objects can also communicate with neighboring pixels, allowing for cellular automata and/or spatial monte carlo simulations in the landscape. By allowing for simultaneous execution of free-roaming, autonomous agents, and a cellular-automata representation of the landscape, a wide variety of human/environment or organism/environment are created with relative ease. I currently collaborating with colleagues in a variety of disciplines to create simulations of grazing animals, coyote packs, boat campers, subsistence pastoralists, and hydrologists ;-) using this simulation environment.
All work regarding the simulation techniques, simulation environment, and their applications are either being written or under submission; to date the only publications regarding this work have been at conference presentations. The software relies on the Swarm simulation libraries developed at the Santa Fe Institute for creation and synchronous execution of objects.