Conformal Additive Manufacturing and Cooperative Robotic Repair and Diagnosis

In the past several years exponential growth has occurred in many industries, including additive manufacturing (AM) and robotics, enabling fascinating new technologies and capabilities. As these technologies mature, the need for higher-level abilities becomes more apparent. For instance, even with current, commercial state-of-the-art technology in AM it is impossible to deposit material onto a nonplanar surface. This limitation prevents the ability to fully encase objects for packaging, to create objects with hollow features or voids, and even to retrofit or repair preexisting items. These limitations can be addressed by the introduction of a conformal AM (CAM) process or more concretely the process in which material is deposited normal to the surface of an object as opposed to solely planar layers. Therefore, one of the main contributions of this work is the development of two novel methods to generate layers from an initial object to a desired object for use in two- and three-dimensional CAM processes. The first method is based on variable offset curves and subject to mild convexity conditions for both the initial and desired object. The second method reparametrizes solutions to Laplace's equation and does not suffer from these limitations. A third method is then presented that alters solutions from the previous methods to incorporate hollow features or voids into the layer generation process. Although these hollow features must obey mild convexity conditions, the location and number of said features is not limited. Examples of all three layering methods are provided in both two- and three-dimensions. Interestingly, these same methods can also be applied to determine the collision-free configuration space in certain robot motion planning applications. However, ultimately, the most compelling application may be in the repair of damaged items. Given an accurate model of a damaged item, these techniques, in conjunction with fused deposition modeling devices embedded on robotic arms, can be leveraged to restore a damaged item to its original condition. In a separate but similar vein, although robotic systems are becoming more capable each day, their designs still lack almost any semblance of a repair mechanism. This issue is increasingly important in situations where robotic systems are deployed to isolated or even hostile environments as human intervention is limited or impossible. The second half of this work focuses on solving this issue by introducing the Hexagonal Distributed Modular Robot (HexDMR) System which is capable of autonomous team repair and diagnosis. In particular, agents of the HexDMR system are composed of heterogeneous modules with different capabilities that may be replaced when damaged. The remainder of this work discusses the design of each of these modules in detail. Additionally, all possible non-isomorphic functional representations of a single agent are enumerated and a case study is provided to compare the performance between two possible iterations. Then, the repair procedures for an agent in the system are outlined and verified through experiments. Finally, a two-step diagnosis procedure based on both qualitative and quantitative measures is introduced. The particle filter based quantitative portion of this procedure is verified through simulation for two separate robot configurations, while the entire procedure is validated through experiments.
conformal additive manufacturing, cooperative robots, robotic diagnosis and repair