Abstract
Smart City indeed become a vision of most of the countries. The city will have robots, drone, smart car, the house entirely operated Intelligently. However, managing how the human crowd and machine can live together is a challenging task; social and emotional interaction between human may affect the crowd behaviour. Therefore, Crowd simulation has the potential to demonstrate the behaviour of a massive people that gathering on a particular location during a specified period. The size of the crowd, geographical site condition and agent's personality potentially make crowd simulation more believable. This paper aims to observe the potential of reinforcement learning for controlling Socio-emotional crowd behaviour by adding the parameter of emotion towards the crowds. The Tree algorithm more superior compare to other machine learning algorithm that capable of predicting the female agent with accuracy 88.1% and male agent around 85.3%. The simulation performed with the railway station scene that has four platform and six lanes to accommodate a passenger. Simulation is initiated with passengers walking toward the main entrance and went to the desired platform. The train is set up to arrive every minute, and the on-board passenger will move toward the entrance door and exit the central station. The reinforcement learning with socio-emotional control expected to bring the crowd simulation to provide realistic and human mimicked behaviour. (C) 2021 The Authors. Published by Elsevier B.V.