In recent years research on Aerial Physical Interaction has been accelerated by improvements and innovations in Aerial platform design. Additionally, the Robotics community at large is taking rapid steps in Human-Robot Interaction (HRI), allowing for complex collaborative actions, including robotic handovers. Consequently, these advancements in both the aerial and the HRI domain, lead to the natural progression of studying the problem of human-aerial robot interaction and its particularities.
As part of the EU project Aerial-CORE, this thesis develops a method that relies on onboard perception in performing human-aware control of an aerial robot. We consider the problem of human-aerial robot handovers, particularly the approach phase in which the aerial robot moves towards the human. Additionally, our proposed method is able to consider human safety and comfort which are paramount for HRI. The method constitutes the formulation of an optimal control problem within the Non-linear Model Predictive Control framework that takes into account, perception goals, actuator limitation and the collaborating human’s safety and comfort with respect to the robot motion.
The proposed method is tested within a realistic simulation using software in the loop approach. The results show the applicability of the approach in handling multiple objectives that are necessary for HRI.
To follow the presentation via Microsoft Teams click here