This paper presents a robot and algorithms for autonomous spray plastering of walls and ceilings during the construction of apartment houses. Successful experimental tests are described where the robot measures the size of the room and the location of doors and windows. The work sequence is then planned and executed autonomously. To sense the environment the robot uses a range measuring scanning laser. The range weighted Hough transform is used to observe the walls and the resulting observations are used to update estimates in an extended Kalman filter based map. Association of observations to estimates is performed using the estimated probability of a correct match, taking correlations between estimates into account when needed.