Navigation and Grasping with a Mobile Manipulator: from Simulation to Experimental Results

Cobot is the name for collaborative robots. This kind of robot is intended to work in close contact with the human being and to collaborate, by increasing the production rate and by reducing the human onerous tasks, in terms of repetitiveness and precision. At the state of the art, Cobots are often fixed on a support platform, static in their workstation. The aim of this thesis is, hence, to explore, test and validate navigation algorithms for a holonomic mobile robot and in a second moment, to study its behavior with a Cobot mounted on it, in a pick-move-place application. To this purpose, the first part of the thesis addresses the mobile navigation, while the second part the mobile manipulation. Concerning mobile robotics, in the first place, a theoretical background is given and the kinematic model of a holonomic robot is derived. Then, the problem of simultaneous localization and mapping (SLAM) is addressed, i.e. how the robot is able to build a map while localizing itself. Finally, a dedicated chapter will explain the algorithms responsible for exploration and navigation: planners, exploration of frontiers and Monte Carlo localization. Once the necessary theoretical background has been given, these algorithms will be tested both in simulation and in practice on a real robot. In the second part, some theoretical knowledge about manipulators is given and also the kinematic model of the Cobot is derived, together with the algorithm used for a collision free trajectory planning. To conclude, the results of the complete task are shown, first of all in simulation and then on the real robotic system.