Planning of indoor construction tasks for mobile manipulators
- Planung von Bauaufgaben im Gebäudeinneren für mobile Manipulatoren
Haschke, Tobias; Corves, Burkhard (Thesis advisor); Schmitt, Robert H. (Thesis advisor)
Aachen : RWTH Aachen University (2023)
Dissertation / PhD Thesis
Dissertation, Rheinisch-Westfälische Technische Hochschule Aachen, 2023
The challenges of climate change, labour shortages and urbanisation require increased automation in the construction industry. In this context, mobile manipulators are used as indoor construction robots. The flexibility of these robots allows them to work on different building surfaces with numerous types of processes. Currently, these construction robots are operated as single systems, but in the future the formation of heterogeneous fleets will be indispensable. To enable the effective operation of such fleets of mobile manipulators, this thesis presents an approach for planning tasks in the field of indoor construction applications. First, a planning-oriented modelling of the properties of the building, the construction robots and the processes to be executed is introduced. The basis is provided by semantic maps, which are available through Building Information Modelling or 3D mapping. The construction robots are represented on the basis of function-oriented modules, which include manipulation, mobility, as well as process and auxiliary actions. A general process model is provided, which becomes executable by assigning process-specific constraints. For a holistic problem description, the models of the building, the construction robots and the processes are related to each other in a high-level planning model. Based on the problem description, a robot-level and a fleet-level planning problem are then formulated. The problem at robot-level describes how a construction robot processes a building surface. For this, mobility, manipulation and auxiliary costs are defined, which are derived from the individual robot modules and the process. The fleet-level problem assigns the building surfaces to the robots as tasks and determines the fleet performance. For this purpose, additional mobility and auxiliary costs are introduced, which result from the reciprocal influence of the robots within the respective building. Task allocations are made on the basis of all introduced cost forms through a sequential single-object auction. Finally, different solution methods for the presented planning problems are evaluated. At robot-level, the problem is solved for different test sets using a branch and bound method. For the fleet-level problem, four construction-specific heuristics and an evolutionary metaheuristic are applied. The results of the five solution methods are compared for twenty use cases of automated asbestos removal and their general applicability is investigated.
- Chair and Institute of Mechanism Theory, Machine Dynamics and Robotics