Sensing and real-time expert system for a masonry building robot

The construction industry is striving to eliminate dangerous and repetitive work, as well as increase quality and productivity in the various tasks. For this reason, there is growing interest in the use of automation and robotics. However, the requirements of robots for construction are different from those of industrial robots, due to the characteristics of the construction tasks and the relatively unstructured working environment.

The main objective of this research is to investigate the enabling technology for a masonry tasking robot, utilising an experimental robot cell. To truly automate the masonry construction tasks, there is need to utilise the advancement in robotic technology, especially to deal with the unstructured environment. This view is in-line with this research, which attempts to solve part of the complex problems of automating the building task, by using forms of sensing and intelligence. Concentration on this is the main distinguishing difference between this work and the few other attempts at physical realisation and experimentation in masonry automation. In terms of research and development of masonry tasking machines and robots, there is much activity on an international scale. Concerning the provisions for machine intelligence in this, it appears that the work reported has the most advanced provisions for computer intelligence. This work is of general relevance to construction robots because imprecision, dynamic performance, unplanned events and cell component relocation are considered.

The experimental robot cell, built at City University, is used in the research. Standard construction materials have been adopted with imprecise dimensions. Using a CAD/CAM facility, building project designs are translated into robot's 'theoretical task'. However, because the masonry material is unpredictable, this can not be directly implemented without real-time adjustments derived from sensing. Notwithstanding, advantage is taken of pre-processing, with real-time accommodation of discrepancies, obstacle avoidance and un-planned events.