Leveraging the human example for planning and controlling robotic manipulators and advancing human-machine interaction

The introduction of the concept of collaborative robot (or "cobot") has overturned the paradigm of the robotic manipulator isolated from human operators inside a cage, allowing the creation of agile production sites where humans and robots can coexist and cooperate. In the last decades, a great effort has been put into improving human-robot interaction and making these systems capable of managing increasingly less structured environments. Despite major improvements in the state of the art recently produced, these robotic systems are still far from being able to be deployed to perform daily living tasks beside humans in a safe and efficient manner. In this sense, looking at how humans move and interact with everyday environment could be beneficial to improve human-machine interaction from different points of view, from the generation of motion similar to the human one to the estimation of the biomechanical state of workers during physical human-robot interaction. This thesis proposes to leverage the human example, i.e. human upper limb muscular-skeletal state modeling, for the definition of novel solutions, ranging from planning and control of artificial manipulator motion to the design of undersensorized systems to estimate the biomechanical state of the human body, to finally advance human-machine interaction. More precisely, the contributions of this work can be split into two major research lines: 1) the characterization of human motion for improving human-machine interaction and optimal sensing of the biomechanical state of the body, and 2) the development of novel human-inspired motion planning and control algorithms for robotic manipulators.