A Construction Robot for Autonomous Plastering of Walls and Ceilings

The problem addressed is plastering of walls and ceilings during the construction of apartment and office buildings. The concrete walls are to be covered by a smooth layer of plaster, a few millimetres thick. The workspace is structured enough that map generation, navigation and motion planning can be performed autonomously. The robot must reliably avoid plastering windows and doors. Typical requirements on the navigation accuracy are standard deviations no greater than 1 cm. Two prototypes are described. For navigation the robot uses a range measuring sensors scanning in one (or two) planes. The observations are used to create, and continuously update, a mainly planar map with some vertical information i. e. door and window positions. The map generation and sensor fusion algorithms are based on a Bayesian association algorithm and an extended Kalman filter. Tests have been successfully performed at an actual construction site. The time used for plastering the walls and ceiling in a room is expected to be less than 50% of that required by manual work The amount of plaster used is also greatly reduced due to more even spraying. However the reasons for developing the robot is mainly ergonomic.