New usage models and applications for technology products, such as machine learning and artificial intelligence, created a need for computers to “see” the outside, physical world. We refer to this as “computer vision”. As these usage models and markets evolve, higher quality imaging systems are needed to enable better optical performance. Consumer expectations for thinner devices and better-quality images contribute to the increased demand for high quality imaging and drive up the complexity of manufacturing these optical systems. As complexity increases, so do manufacturing challenges. For example, the use of lower cost materials adds to part-to-part variation, and high failure rates (scrap) make it difficult for manufacturers to deliver these complex optics systems at scale.
You can download the paper here