As the number of people with access to commercial, and disposable, products continues to increase with the overall increase in population, methods to recycle or reuse broken objects must also be investigated more thoroughly to enable a sustainable future. Most common household objects, e.g., mugs, plates, vases, utensils, furniture, fixtures, etc., are considered to be disposable, and when an object is fractured during daily use consumers are encouraged to discard the object instead of attempting to repair it. No tools are available to consumers to repair objects with complex fractures even if the object has value, e.g., an expensive vase, a family heirloom, or an object with cultural significance. Environments with limited resources, e.g., remote research facilities or the space station, offer few alternatives to repair fractured objects other than reassembly, which is not feasible if parts of the object have been lost or destroyed in the fracture process. This thesis presents six approaches to automate the repair of fractured objects comprising (a) an approach using high resolution 3D scans of the fractured object and a corresponding complete, i.e., unfractured, object, (b) an approach using a database of complete objects when a 3D scan of the complete object is not available, © three approaches using deep neural networks to automatically infer the repair object from a high resolution 3D scan of the fractured object, and (d) an approach using deep neural networks to automatically infer the repair object from an image of the fractured object. We also introduce the first dataset of real, physically fractured 3D models to enable evaluation of repair approaches, and the first large-scale dataset of images depicting real, physically fractured objects to enable training and evaluation of methods to perform automated repair from images.