Abstract
In this thesis, an agile robotic workcell that uses a mobile robotic platform mounted on a scaffold structure was designed for the automated welding of large structures such as ships, or oil rig legs and platforms. In these cases, the robot must be moved to the weld joint as the structure (workpiece) is considered to be immovable. Scaffolding systems are commonly constructed around such structures during their construction to allow workers to access the weld task, which is often at elevated heights. The mobile platform was designed to be highly portable – the platform, including the lightweight robot arm (UR10), weighs 50 kg and can be installed onto any two standard parallel scaffold pipes in less than 60 seconds without any specialized tooling. This allows workers to easily transport and deploy the robotic workcell around a 3D work site to perform a variety of tasks. An analytical model was developed based on Euler Bernoulli beam theory that calculates the expected flexion of the supporting rails based on the pose-specific static loading of the robot, and predicts the resulting error at the robot tool tip as compared to a rigid robot base. The model was used to optimize the design of the platform to minimize the tool tip error caused by rail flexion. Experimental results show that the proposed model shows appreciable accuracy, and can predict the x, y, and z components of the tool tip location to be within 1.2 mm when the tool tip error caused by the flexion of the supporting rails is upwards of 9 mm when welding a segment of an X-Joint. The model can therefore be used for predictive error mapping as part of a tool tip compensation strategy during robot control.