Challenging Human-Aware Robot Navigation with an Intelligent Human Simulation System

The final validation of human-aware social robot navigation schemes requires testing and evaluation with real-life humans. During the prototyping and developmental phase, however, experiments with real humans for debugging and testing may not be a feasible option. Simulators can help overcome this issue, but the lack of available intelligent human avatars with rational behaviors restricts the testing to simple scenarios. In this regard, we propose a system for simulating an autonomous intelligent human agent specifically designed to act and interact with the robot navigating in a simulated environment. Further, the proposed system provides interactive GUI modules to run repeatable scenarios and visualize the interaction data. We conduct a series of experiments that show how the proposed Intelligent Human Simulation (InHuS) system can help tune and debug social navigation better.