The development of a fully autonomous ground vehicle (FAGV)

Abstract
As a first step toward the creation of a fully autonomous vehicle that operates in a real world environment, we are currently developing a prototype autonomous ground vehicle (AGV) for use in factories and other industrial/business sites based on behavior-based artificial intelligence (AI) control. This flexible and fully autonomous AGV (FAGV) is expected to operate efficiently in a normal industrial environment without any external guidance. The crucial technique employed is a non-Cartesian way of organizing software agents for the creation of a highly responsive control program. The resulting software is considerably reduced in size. Through numerous experiments using mobile robots we confirmed that these new control programs excel in functionality, efficiency, flexibility and robustness. The second key technique in the planning stage is evolutionary computation, of which genetic algorithms are a principal technique. An online, real-time evolution of the control program will be incorporated in later phases of the project to make FAGVs adaptable to any given operational environment after deployment. The first prototype FAGV has an active vision and behaviour-based control system.

This publication has 7 references indexed in Scilit: