Object-Based Probabilistic Place Recognition for Indoor Human Environments

Giving a robot autonomy and independence in a human environment implies not only having to move safely but also the ability to understand the environment where it is located. Scene understanding is one of the most challenging tasks in robotics because the design, the objects and the arrangement of them in the scene varies considerably. In this paper we present a Probabilistic Place Recognition Model applied to mobile robots and able to work in indoor human environments. A model of uncertainties is proposed based on the information about the objects in the scene and the relationships between them. This information can influence the final decision about the probability of the presence of a robot in a place. The experimental results obtained of common indoor human environments demonstrate the ability of the model to predict place categories considering the information of the objects and the relations between them. Using more information in the prediction process makes the model more descriptive, scalable and better adapted for human-robot and robot-environment interaction.