What Is Next with SORDI

This chapter describes the challenges and next development steps with SORDI to perform effective and efficient object recognition and transfer learning. It discusses the integration of real-world assets and third party assets into SORDI to make it more realistic. It also covers the so-called reality gap between real and synthetic assets, and describe different solutions to breach the gap and bring synthetic images and environments closer to their real counterparts. The chapter also covers the augmentation and partial automation of real data capture, and the integration of both real and synthetic datasets to perform deep learning model training. In addition, This chapter highlights the impact of using SORDI in smart manufacturing and logistics applications in order to reduce the industry’s carbon footprint, and allow for more sustainable production pipelines. This chapter attempts to answer the following questions: What is next for SORDI? How can we make SORDI more realistic, and its usage more practical? How can SORDI help BMW Group’s manufacturing process become more sustainable and help reduce its carbon footprint?