Stochastic Simulation with Human Interaction in Remotely Piloted Aircraft Systems (SARP)

Abstract

Due to advances in artificial intelligence, the robot may have some level of autonomy to decide what it considers ideal for the mission or objective it has to fulfill. In this way, the decision maker that is contained in the SARP can, at some point, choose whether to obey the human operator or to continue executing its planned mission. The objective of this research is to propose a stochastic simulation model that evaluates the influence of operator commands on robot decision mechanisms. Under an exploratory, methodological research of applied nature and a quantitative approach, simulations will be carried out using the Markov chain through the PRISM tool, simulating semi-autonomous agents in which human interaction is performed without restrictions. In this model, human interaction obviously affects the overall activities of the SARP operation. The objective is to obtain data while this interaction affects the robot's plans to propose a model in which the human will influences the robot's decisions in cases that either the survival or the mission is not compromised. This is important because human decisions are slow and can be delayed or truncated due to communication channel problems. An air reconnaissance scenario was considered through the use of the SARP with some degree of autonomy and receiving remote human commands at the same time. The results show that depending on the type of human interaction and its frequency, it is possible to make human interaction compatible without compromising the robot or the mission.