Implementação de agente inteligente, modelado como um veículo autônomo, para desvio de obstáculos estáticos

Abstract

Autonomous cars are a revolution for the automotive industry and are becoming closer to reality, already being developed by big companies like Google, Uber and Tesla. The area of autonomous vehicles only tends to grown in the coming years and comes to modify and broaden the structure of society in terms of how cars are used and how are the interactions with users.The development requires the joining of several areas, one of which is the area of intelligent agents, within intelligent agents it is possible to incorporate several techniques to the agents, one of these techniques is so that it can avoid obstacles. There are several ways to implement an intelligent agent, this work uses the BDI (Belief, Desire and Intention) architecture, where the agent has beliefs, desires and intentions, a framework that implements this architecture is JaCaMo. In order for the agent to be able to act, an environment is necessary. For this purpose, Simbad is used, which corresponds to a simulator that allows the construction of an environment with static and dynamic objects (robots). The environment defined for this work has only static obstacles, which are defined within tracks, and the agent transits between them to avoid the obstacles. The simulator is also responsible for providing the information to the agent, so that the agent is able to control its representation within the simulator. In order for the information to be exchanged a communication protocol was defined, and allows the agent to receive information about the environment and sent the actions that must be performed for its representation within the simulator. After the integration of the simulator tests were performed with the agent to obtain information of their capabilities, being possible to note that within the defined scenario the agent can perform the proposed maneuvers.