Reducing Computational Cost During Robot Navigation and Human-Robot Interaction with a Human-Inspired Reinforcement Learning Architecture | Litlas