The effects of soft somatosensory system on the execution of robotic tasks

Nowadays there is a growing interest in the design and implementation of robots capable of exhibiting sophisticated behaviours, taking part in social interactions, expressing and manifesting their "feelings". These kinds of robots can be employed as human-like collaborators or companions. At the heart of this sophisticated behaviour, there should be the robot's ability of effectively being aware of its internal status. The feeling of pleasant or unpleasant sensations comes before higher-level mechanisms that determine a robot's emotions and motivations in the accomplishment of a task. Furthermore, negative sensations can represent an alarm indicating dangerous situations. In this work, we have modelled a bio-inspired somatosensory system embedded in a cognitive model for a humanoid robot. The system has been designed by analysing the internal physical structure of the robot and it relies on different soft sensors built on top of the robot's physical sensors. The soft sensors allow the robot to interpret its body sensations using a suitable classification of the somatosensory signals. They influence the motivation of the robot, modulating its final behaviour, as shown in an experimental setup.