Kinematics models for human robot collaboration

The collaborative robots, also known as cobots, are nowadays involved in several industrial processes to speed up and streamline the process for greater productivity. The cobots and the operators are implicated in the cooperation of activities combining robotic and human skills: the robot's precision and repeatability is synchronous with the human intelligence and flexibility. Since these robotic systems are safe and reliable, they are adopted in the medical field, especially in the rehabilitation sector. Some traditional rehabilitation devices already exist on the market (MOTORE, PABLO), but the cobots application with handle systems have grown only over the last few years, as in the project of Universal RoboTrainer or ROBERT Life Science. Unlike traditional robots, the new generation of cobots is designed to be flexible, lightweight and easy to program. They typically have a rounded shape without any pinch points and internalized wires. Standards for collaborative operations are needed to ensure human safety and the proper machine operation. In this thesis, mathematical models of human and robotic arm will be studied for the kinematic analysis of Human-Robot Collaboration (HRC). First, the motion planning is accomplished to generate a desired motion through a sequence of points. Then, the computation of the direct and inverse kinematics for a human and a robotic arm has been performed. In particular, the model chosen for the upper limb is the one with 7 degrees of freedom (DoF) represented by three rigid segments connected by frictionless joints. The robotic arm considered, instead, is the UR5 by Universal Robot company, which is characterized by 6 DoF. The aim of this project is to realize a framework for the human-robot motion planning. Starting from the anthropometric measurements and the joint excursions of the upper limb, the kinematic analysis allows the execution of robot joint motion perfectly in line with the human one. Therefore, the proposed algorithm is designed to be suitable for each patient, and in the meantime, the physiotherapist is responsible only for the assessment of the upper limb motion. Simulations for HRC in rehabilitation processes will be executed with the aim of realizing a virtual prototype model for human and robotic arm. OpenSim software is adopted to model only the human system based on anthropometric measures. ADAMSsoftware, instead, is exploited for combining motions of human and robot models. Then, a control simulation has been performed in Simulink-ADAMS environment. Three different human-robot motions are analyzed: one-way motion, round-trip motion and 3-points motion. In conclusion, the virtual model can execute the predefined movements without computational errors. The proposed methodology is a starting point of cobots integration into current rehabilitation practice improving the quality of physical rehabilitation. Future research could involve the analysis of different human-robot working modality.