pyAutonomousAgent: An Academic Tool for Modeling Autonomous Agent Behaviors Using Behavior Trees

Abstract

Computer simulations have been applied in several areas. Non-player characters in computer simulations are autonomous agents. An autonomous agent receives information from the simulated environment, processes this information, and estimates a situation awareness (state), makes a decision based on the estimated state, and performs an action related to the decision. The decision-making process of an autonomous agent can be efficiently modeled through behavior trees. However, teaching the construction of behavior trees for autonomous agents can be a very complex task, especially if students have little knowledge of computer programming. The objective of this work is to simplify this task. Therefore, I proposed the pyAutonomousAgent, an academic tool for modeling autonomous agent behaviors through behavior trees. The tool is an open-source set of computer routines, developed with the aim of being easy to use and learn. As a case study, the tool provides a simulation scenario with a swarm of drones. Results of the drone behavior modeling process are presented.Computer simulations have been applied in several areas. Non-player characters in computer simulations are autonomous agents. An autonomous agent receives information from the simulated environment, processes this information, and estimates a situation awareness (state), makes a decision based on the estimated state, and performs an action related to the decision. The decision-making process of an autonomous agent can be efficiently modeled through behavior trees. However, teaching the construction of behavior trees for autonomous agents can be a very complex task, especially if students have little knowledge of computer programming. The objective of this work is to simplify this task. Therefore, I proposed the pyAutonomousAgent, an academic tool for modeling autonomous agent behaviors through behavior trees. The tool is an open-source set of computer routines, developed with the aim of being easy to use and learn. As a case study, the tool provides a simulation scenario with a swarm of drones. Results of the drone behavior modeling process are presented.