Trade-Off: Balancing Accuracy and Usability for the FutureForge Manipulator
Mechanical and Aerospace Engineering
Figure 1: The FutureForge manipulator installed at the ARFC near the end of 2021.
Traditionally speaking, the modern forge takes its origins from blacksmiths' workshops. However - because of the scale of materials being handled in industrial processes today, the extreme temperatures being used, and the forces at play - modern forges are dangerous environments for people to be working in. Additionally, while there is charm and artistry to hand-made items in this general process, the accuracy and precision demanded in modern industry makes a strong argument for removing people from the equation as far as possible. With these factors in mind, Strathclyde's Advanced Forming Research Centre (AFRC) has been working on a project entitled The FutureForge. The FutureForge is intended to be a world-leading forging platform and consists of two major components: a 2,000-tonne hydraulic press; and a similarly massive robotic arm. These are very much analogous to the blacksmith's hammer and hand manipulating the material. This piece of research focuses on the robotic arm (or "manipulator") in this environment and on how we can safely control this next generation of industrial robot.
Previous work derived the equations that describe the manipulator's motion, which produced two very large, unwieldy differential equations. Taking off from this previous work, our task is to get this system of equations into a solvable state. The reason that we want to solve this is two-fold. Firstly, having a reliable analytical solution to the system enables us to accurately simulate the machine's behaviour. In turn, this allows us to make observations and predictions that can help guide safety practices and operational procedures. The second motivation we have is to design a suitable control system for the manipulator: ensuring its behaviour is true to what the operator has specified down to the sub-millimetric scale. While this present work focuses only on getting this system into a solvable form, it is important to note that these motivations are what guide our decisions along the way.
The overview of our approach begins with the reduction of the governing equations into an accepted standard form - grouping large expressions of geometric terms into more manageable constants. After this, a significant step was to convert the trigonometric terms (sines and cosines) into algebraic expressions. We went about this step by making pragmatic observations about how/where we could afford to reduce the accuracy of our model: the result being an approximate set of equations that trades some degree of accuracy fro solvability. Making further pragmatic observations (regarding how the manipulator would be operated), we were able to reduce the system's complexity by showing how the gravitational terms can be removed from the consideration entirely. The result of these efforts is an algorithm that adequately approximates the manipulator's behaviour under the operating conditions described to us by our industrial partners.
Author retains copyright to text and images.