Embodied Computation for Uncertain Environments Automated construction in unstructured environments Abstract: The transfer of industrial robots from the assembly line to the fabrication of architecture has recently been shown to offer great flexibility and opportunity for custom fabrication. At present, robots are typically controlled with strict offline programing to produce controlled results similar to those in more traditional manufacturing industries. In architectural construction, however, robots have not managed to replace humans in on site operations, in large part because of the need for adaptability required. In the bricklaying profession, the craftsman uses his intuition to adapt to the material he is working with and to the changing work environment. As an example, this craft presents many of the problems inherent in building a control system for a robot that can begin to be of use in the unstructured environment of the construction site. The research focuses on the robotic performance of two main aspects of bricklaying: the craft of laying brick and mortar itself and the ability to keep track of the work on a larger scale. In order to replicate these aspects an adaptive program was built with reliance on sensors and feedback to control the robotic arms. The robotic arm managed to achieve the tasks in question using two software tools: a new plugin to control the movements of the robots and a server that is capable of dealing with feedback. Robotic Bricklaying With recent emphasis in the field of architecture towards fabrication technologies, the use of industrial robots has surfaced as a replacement of custom fabrication machines. Their versatility allows different tasks to be executed on the same machine. Robotics is a relatively new field of technology, the term ‘robot’ itself only dating to the 1920s. The technology has come a long way since then, growing more complex and capable, however the limitations associated with the idea of a simple, unthinking automaton have often remained in practice. In manufacturing, robots are typically controlled with strict offline program that is developed in detail in advance, runs for the lifespan of the machine, and is situated in a controlled, human free environments with limited sensory input. Although robots are now a crucial part of the manufacturing industry, they have not managed to replace humans in on site operations or in construction due to these limitations. Apparently simple tasks like avoiding obstacles or adapting to the location of work become a difficult task for robots. Construction still relies mainly on human labour, and crucially, for many crafts performed on site, humans rely on their tuition and senses to achieve their work. Bricklaying is one of those professions. It hasn’t changed dramatically through millennia of practice, and the tools themselves are as old as the profession. The craftsman uses his sense to shift between different building materials and to deal with relatively unpredictable ones like mortar. The difficulties inherent in these actions are taken as the basis of research into the ability to build a system that can respond to changes in order to replicate this craft using industrial robots. For this research one of the new Bartlett UR10 robotic arms was used. In order to test the viability of such a machine for this type of activity, the research focused on two main aspects of the brick laying craft: the ability to work with multiple different materials, both solid bricks and fluid mortar, and the ability to keep track of the overall execution of the task of wall building. A special end-effector was built for the task around an industrial gripper. This allowed shifting between picking and placement of bricks and application of mortar without the need for changing the tool. A traditional trowel was used for the application of the mortar, the low-tech device allowing for and necessitating the creation of a more sophisticated controlling movements rather than creating a motor printing system. The end-effector also contained the limit switch and a depth camera needed for the feedback and sensing part of the program. The first requirement of the research was the development of a control system for the robots themselves. Upon commencement of the research, no software was publicly available to control the novel UR10 either by the manufacturers, Universal Robots, or by third party developers. A new open source platform under the name Onix was created in grasshopper to be used as the controlling platform. The platform included an inverse kinematics solver, program generation and end effector tools with the ability to directly stream to the robots. The simple interface of the platform allowed the development the programs for the first part of the research dealing with the craft of bricklaying itself. Several small programs were created for experimentation with basic subtasks, including picking and placement of bricks, picking and leaving trowel, checking the mortar surface, checking the brick height and mortar laying. The breakdown of the actions in this way allowed for concentration of research on the particularly difficult and interesting areas of the craft, while removing those initially considered more trivial. As pick and place actions have been taken for granted in the field of robotics for years, achieving this for bricks or the trowel was straightforward.
Top: Building a sample of a rotated brick wall as a test of the program Middle top Right: The Brick laying server running Middle top left: running a program to build a four raw wall sample Middle bottom: System diagram of the logical sequence of the process Bottom left: Onix grasshopper plugin developed during the project to control the universal robots Bottom right: The inverse kinematic solutions for the universal robots for the IK solver in Onix
Khaled ElAshry khaled.elashry.12@ucl.ac.uk / thinkingrobotics.tumblr.com
MSc/MRes Adaptive Architecture & Computation Bartlett School of Graduate Studies