Optical Windows for Robotic Computer Vision Systems and other SLAM Based Sensors: Glass + Plastics

As an engineer and professional photographer, I have spent thousands of dollars on optical glass, filters, and other elements that operate within the imaging optical stack between my target (subject) and sensor (camera). So when someone wants to put plastic in my system I often reference the following excerpt from a well-respected optical engineering book.

7.5 Plastic Optical Materials

Plastics are rarely used for precision optical elements. A great deal of effort was made to develop plastics for optical systems during the Second World War, and a few systems incorporating plastics were produced.

Today, the use of plastics is largely restricted to novelty items such as magnifiers and toys, with the notable exceptions of low-priced camera leases and some Schmidt aspheric corrector plates. In these applications, the fact that plastic may be conveniently molded gives it a great advantage over glass. For the toys, novelties, and camera lenses, the advantage is mass production at extremely low cost. For the Schmidt corrector plate (and the camera lens) the advantage is the ease of production of a nonspherical surface, which is essential to the design and is expensive to execute in glass.

The obvious advantages of plastic- that it is light and relatively shatterproof-are offset by a number of disadvantages. It is soft and scratches easily. Except by molding, it is difficult to fabricate. Styrene plastic is frequently hazy, scatters light, and is occasionally yellowish.

Plastics tend to soften at 60 to 80°C. In some plastics the index will change as much as 0.0005 over a period of time. Most plastics will absorb water and change dimensionally; almost all are subject to cold flow under pressure. The thermal expansion coefficient is almost 10 times that of glass, being 7 or 8 x 10-5/°C.

The change of index with temperature for plastics is very large and negative. Thus, maintaining focus over a range of temperatures is a significant problem for plastic optics. Often they must be ather-malized as well as achromatized. The density of plastics is low, usually to the order of 1.0 to 1.2. The characteristics of some of the most widely used optical plastics are summarized in Fig. 7.9.

Another optical application for plastics is in replication. In this process, a precisely made master mold is vacuum-coated with a release, or parting layer, plus any required high- or low-reflection coatings. (The nature of the release layer is usually considered proprietary, but very thin layers of silver, salt, silicone, or plastic have been publicly mentioned.) Next, a few drops of low-shrinkage epoxy are pressed out into a thin ideally about 0.001 or 0.002 in thick) layer between the master and a closely matching substrate. The substrate may be Pyrex, ce-amic, or very stable aluminum (for reflector optics), or glass (for refracting optics.) When the epoxy has cured, the master is removed and a reasonably precise (negative) replica is left on the substrate. This process has several advantages. For example, any surface including aspherics) for which a master can be made can be replicated relatively inexpensively, since the master can be used over and over. Other advantages are that a mirror can be made an integral part of its mount, the bottom of a blind hole can have an optical polish and figure, and extremely thin and lightweight parts can be produced. In many cases these things are effectively impossible with standard optical fabrication techniques. The limitations to replicated parts are the inherent softness of the epoxy and the change in the surface figure from that of the mold.

Source: Modern Optical Engineering - The Design of Optical Systems (Second Edition), Warren J. Smith

The goal is to capture the best analog sensor data possible before a computer digitizes that information into a compressive storage-friendly format (which even at 4K is still a data loss-based process) and starts making computational decisions based on that digitized information. 4 months before the first iPhone shipped, Steve Jobs contacted Corning to develop a thin glass screen to replace the plastic screen cover that was originally used for the iPhone public demo. Steve felt, and rightly so, that the plastic scratched too easily. This special glass would come to be known as Gorilla Glass, which was already developed by Corning but had no market until Apple came along. The equation runs in the other direction when trying to communicate lossless analog information. Always represent the best optical stack possible, capture the best data possible, and then process that information.

Polycarbonate manufacturers like Lexan will claim things like “optical grade” polycarbonate, which for the Lockheed Martin F-22 Raptor is perfectly fine, or even an observational “glass” window that needs to protect the human eye in a lab. At the photonic level, the well-studied characteristics (and parasitics) of the material still exist. Thin film depositions, scratch-resistant treatments, or UV stabilizers will not change the electromagnetic interaction of the core polymer material with light. So use glass whenever possible in your imaging optical stack.