• intuitive interfaces
• smart robot programming

Today, robotics offers robust and mature hardware solutions, by ensuring flexible manufacturing and set-up investment longevity. Furthermore, the advent of cobots on the market has provided a sense of safety, reliability, efficiency and affordability. Besides, more and more anthropomorphic robots begin to be used in contexts that go beyond industrial sector, like disaster-scenarios, nuclear power plants, hospitals and home settings. This widespread diffusion in different fields and among non-expert users requires accessible and intuitive robotics. In this direction, Human-Robot Interface and Human-Robot Interaction fields of study are dedicated to understanding, designing, and evaluating robotics systems for use by or with human. In addition, Learning from Demonstration (LfD) is established as a promising method to transfer skills from humans to robots. The latest developed approaches in this field make use of machine learning techniques, which have shown successful results although they require a large dataset of demonstrations. This thesis proposes an intuitive human-robot interface with different levels of autonomy, ranging from teleoperation to autonomous execution, depending on the context and the task. I developed methods, tools and interfaces to enable non-experts to use and program robots with a single demonstration and without any coding. Neither knowledge in robotics nor programming is required, thereby bridging the gap between humans and robots, making the latter available for everyone. These systems have been tested both in the laboratory and in different real contexts in the field of industry, inspection&maintenance, telemedicine and in home-setting, especially in the nuclear field (RACE - Oxford - UK), natural disasters scenario (post-earthquake Amatrice - Italy), I.C.T. infrastructures (Fastweb Test Plant Milan -Italy) and during the recent pandemic event (Covid-19 Pisa and Massa Hospitals - Italy).