A reactive logical agent

A system for developing and executing reactive, autonomous, rule-based agents is presented. The rules consist of sentences of a temporal logic connecting the present and past history of the system to its future states, and are interpreted at run time. The design and the implementation of the system are illustrated, along with some experimental performance measures derived in a real-time setting, where the agent has been put at work. These measures show that, even is such severely constrained context, the system can efficiently handle agents embodying up to a few hundreds rules.