LEARNER

Human-Robot Colaboration performance is highly dependent on the robots' comprehension capabilities. Hence, social robots need to understand humans' states, predict their behavior, and act accordingly. Such a property can considerably reinforce the effectiveness of interaction and raise trust, especially within laborious tasks or tense situations, which are currently limited. We intend to introduce social skills in the PP module by considering the humans' presence, dynamics, actions, and emotional states in the robot's internal map representation.

The efficient introduction of suitable perception capabilities in robotics platforms, that improve the way robots comprehend their environment and interpret humans, remains an open and highly use-case dependent challenge. Due to their application scope, such systems must effectively operate under non-deterministic events, considering environmental variations and dynamic human behavior, while maintaining autonomous and real-time performance. Within LEARNER, we will develop robust SLAM and Path Planning modules that will enhance the existing techniques for coping with structural and conditional changes in the environment, as well as handling the human's dynamic presence.

To prove the package's effectiveness and portability, the system’s capablities need to be evaluated within a schenario and performance metrics set by end-users. Therefore, apart from simulations, we will integrate and assess the developed framework on a mobile robot under challenging conditions that resemble a physical environment defined by user requirements. The prototype will also be tested concerning its real-time capabilities.