Using Genetic Algorithms to Optimize Social Robot Behavior for Improved Pedestrian Flow

Abstract

This paper expands on previous research on the effect of introducing social robots into crowded situations in order to improve pedestrian flow. In this case, a genetic algorithm is applied to find the optimal parameters for the interaction model between the robots and the people. Preliminary results indicate that adding social robots to a crowded situation can result in significant improvement in pedestrian flow. Using the optimized values of the model parameters as a guide, these robots can be designed to be more effective at improving the pedestrian flow. While this work only applies to one situation, the technique presented can be applied to a wide variety of scenarios.