dc.description.abstract
The current challenge in flexible production with highly automated production lines is the ongoing trend toward customization of products up to full individualization. Consequently, path-planning algorithms for industrial robots have to keep pace with this trend. This thesis presents flexible planning algorithms to support the automatic generation of robot programs in flexible automation to solve complex path-planning problems in industrial processes on freeform 3D surfaces. In industry, offline robot programming approaches using computer-aided workflows, where manufacturing paths are generated manually or semi-automatically, are state-of-the-art. This work investigates a fully automatic generation of robot programs for 3D workpieces based on user-generated 2D input patterns. For this, two projection methods from 2D to 3D, i.e., a simple parallel projection and a least-squares conformal mapping, are evaluated. In order to show the accuracy of the projection approaches, a drawing process with an industrial robot is demonstrated in an experimental setup with two task-space control concepts, i.e., a motion control and a hybrid force/motion control. With the hybrid force/motion control, the normal contact force of the pen with the workpiece's surface is controlled, yielding the most accurate drawing result. In flexible production, a robotic work cell must be able to execute an industrial process on a wide range of products, including ones not known during the design phase of the work cell. If specific manufacturing paths are not executable, laborious adaptions of the robot placement or manufacturing path are necessary. In some instances, adapting the tool mounting on the end-effector can also lead to executable robot trajectories, as shown in this work. Therefore, an optimization algorithm is developed and combined with a joint-space path planner to compute the optimal tool mounting while maximizing, e.g., the number of joint-space path solutions and distance to the mechanical joint limits. This algorithm is also applicable for finding the optimal robot base placement. It is validated in an industrial trimming process from the shoe industry, where a set of manufacturing paths must be executed. To further increase the flexibility of given work cells, the distinct properties of the manufacturing process, i.e., redundant degrees of freedom and allowed deviations from the manufacturing path (tolerances and process windows), can significantly enlarge the path planning search space. Hence, an optimization-based joint-space path planner is developed, which systematically includes these process properties and a collision avoidance strategy. Multiple joint-space paths are computed in parallel to find the optimal path. The proposed algorithm optimizes two distinct processes, i.e., a drawing process and a spray-painting process. It is shown that complex path-planning problems can be solved on freeform 3D surfaces where state-of-the-art concepts fail. In this thesis, the developed algorithms are demonstrated experimentally for a drawing task and in simulation for a trimming and spray-painting task, which are manufacturing processes representative of industrial processes on freeform 3D surfaces. Additionally, the proposed methods are formulated in a general way such that they can be easily applied to other processes, e.g., welding, spray painting, milling, polishing, or textile fabrication processes like cutting, sewing, and gluing.
en