The keywords for the project's approach are reincarnation and co-evolution. Everything starts with a human's question about an unknown environment. This question is then put to the test in two loops. In the first 'outer' loop extremely small autonomous physical agents explore the unknown environment. The first iteration of the loop is initiated based on a “best guess” model for the environment and agent configuration derived from (possibly uncertain) knowledge elicited from experts that is stored in a database.

Based on the information provided by the agents recovered from this environment, a virtual model of the environment is constructed - this is the second loop. This virtual model is repeatedly explored by virtual representations of the physical agents using an evolutionary algorithm, co-evolving both the virtual model and the virtual agents, helping to refine the next incarnation of the actual physical agents.

These reincarnated agents will again explore the unknown environment to gather more data. To achieve this goal the project brings together researchers from multiple areas including computer science, mathematics, physics, linguistics, electrical engineering, communication technology and systems architecture.

The process, which also benefits from statistical data and any information previously known, will generate as precise information as possible about spaces that are inaccessible to humans. Questions regarding natural resources or environmental conditions will be much easier to answer.

More information is available on the website of this EU-funded project.