Simulating Intelligent Human Agents for Intricate Social Robot Navigation
Résumé
Simulators are very effective tools in testing and debugging robotic systems before real-world deployment. In the context of social robot navigation, simulated human agents are required to test various settings. However, simple and reactive human agents may not be sufficient to simulate some complex indoor scenarios. Such scenarios require intelligent human agents that can analyse the situation and decide a navigation behavior. Our current work [1] presents an Intelligent Human Simulator (InHuS) that provides a generic architecture for simulating intelligent human agents and its application in social robot navigation. In this work, we present an extension of InHuS, which enhances the navigation of the human agents and also enables the simulation of different navigation behaviors. This is achieved by integrating InHuS with a modified human-aware planning system. The results in various simulated scenarios show clear improvements in the human agent's navigation. Furthermore, results also present some interesting navigation behaviors of the human agent that are simulated by simple parameter tuning.
Domaines
Robotique [cs.RO]Origine | Fichiers produits par l'(les) auteur(s) |
---|