par Ambrosino, Michele 
Président du jury Innocenti, Bernardo
Promoteur Garone, Emanuele
Publication Non publié, 2023-11-06
Président du jury Innocenti, Bernardo
Promoteur Garone, Emanuele
Publication Non publié, 2023-11-06
Thèse de doctorat
| Résumé : | With the constant technological advancements, we are assisting to the progressive robotization of many labor-intensive and repetitive human activities. In their first historical phase, robots were mostly confined in the production lines of factories. Currently, robots are “leaving” the factories to assist humans in an increasing number of activities, ranging from logistics and transportation to house services (eg cleaning and mowing lawns), and agriculture, just to mention a few. In this thesis, we focus on the use of robotics in civil construction. Surprisingly, although the construction of buildings is a very relevant human activity, robots are still a very rare sight on construction sites, and if present, their role is typically limited to logistics (eg transportation of pallets) or collection of information (eg drones for topographical surveys). At the current stage, only a few working robotic platforms for automatic construction have been developed around the world and their diffusion is quite limited. We believe that the current scarce presence and success of robots in construction are mainly due to the specificity of the construction work environment and actions (eg adverse and unstructured environment, need of maneuvering in limited space, complexity of the construction actions and their low level of standardization, high weight of the construction material, etc.). This context is very different from the industrial context in which robots have been developed and requires the development of methodologies that are beyond “classical robotics”. The main goal of this thesis is to develop a novel methodology able to make robots more common in construction sites. To this end, we will design and develop an innovative multi-robot concept for automatic bricklaying with sand-lime bricks. Sand-lime bricks for large civil construction are typically in the 100 kg range. The use of these blocks is very common in many civil constructions thanks to their mechanical and insolation characteristics. Currently, they are placed and glued manually by teams of two or three masons with the aid of a small crane. The roboticization of this activity poses a series of very interesting methodological challenges that are quite general for robotics in the construction context. First, the weight of the blocks and the typical distance at which the blocks must be placed makes it unreasonable the classical robotic assumption of “rigid-manipulator”. Second, in order to correctly place the blocks, the system should be composed of two heterogeneous sub-robots. The correct cooperation between the two sub-units poses a series of coordination challenges that we believe must be tackled within the context of multi-agent constrained control. Finally, the correct placement of each block includes the combination of data coming from a set of very heterogeneous sensors.To meet all these challenges, in this thesis we divide the bricklaying activity into different phases. Initially, we focus on the macro movement of the block from the pallet close to its final position. This operation is performed by the lifting mechanism (ie the crane). Mathematical modelling and development of control laws for cranes are then provided and explained in detail. Next, the mico-movement phase to position the block in its final position is analysed. In this perspective, the mathematical modelling and control of the proposed multi-robot system are addressed and experimentally validated. At the end of this thesis, some preliminary results on the Situational Awareness aspects of the project will be shown. We believe this development will be not only innovative per se but it is also a very representative case study for the field of construction robotics at large which we believe can inspire a number of other initiatives. |
