16Jan 2016

Our new website presents our suite of products called Adaptive Charging Software. It is fair to say that everyone understands and recognizes the meaning of “Charging” and “Software”…but “Adaptive”? What does it really mean? The purpose of this post is to give the reader an intuitive feeling of the meaning of (and consequently the need for) “adaptive” as it relates to technology.

Let’s first start with the classical definition of adaptive:

a•dapt•ive (ə-dăpˈtĭv)  adj. Relating to or exhibiting adaptation.

Ok, it relates to adaptation, but adapting to what? and why? For that, let’s illustrate with an example at how adaptive algorithms and software became instrumental to modern photography.

Let’s look at two photographs of Liberty Cap from a recent trip I took to Yosemite National Park. Can you tell by looking at the photographs what camera(s) were used in taking the shots? I doubt it. They both offer plenty of resolution, richness of color and great image quality (you may click on each photo to enlarge it).

                         D7K_1310_1200px

 IMG_2484_1200px

The top photograph was taken by a Nikon D7000 DSLR with a 24-mm f/2.8 prime lens. Total weight: 1,042 g (2.3 lb). Total cost when new: about $1,500.

The bottom photograph was taken by an iPhone 6 Plus. Total weight:172 g (0.38 lb). Total cost when new: about $600 for the smartphone. The camera component is less than $15 and only a few grams.

So why are the two photographs so similar, and what is the purpose of using DSLRs over a smartphone if the differences are so minuscule if not inexistent?

From an optical standpoint, the camera optics of the iPhone 6 Plus are absolutely no match to the superb optics of the Nikon lens. The iPhone 6 Plus camera sensor is also no match to the one in the Nikon D7000 — though both are manufactured by Sony but to vastly different requirement standards.

Today’s cameras, both DSLRs and smartphone cameras included, incorporate very sophisticated computational electronics on board. The iPhone 6 Plus boasts a powerful Apple ARM processor, and the Nikon camera includes a sophisticated Expeed processor. Both of these processors perform corrections on the fly before, during and after the photograph is taken. For instance, they both incorporate algorithms that assess the nature of the scene (e.g., is it a landscape, or does it include faces?) to determine the exposure parameters. Same for the focusing. Additionally, they both make corrections on the fly for the optical errors coming from the optics…and this is just the beginning.

Now, the Nikon 24mm prime lens is a superb lens and has excellent optics. In contrast, the lens used in the iPhone 6 Plus is no match. It suffers from significant optical errors called aberrations. For instance, one of these errors is called distortion: the photograph, uncorrected, looks distorted. Another error is chromatic aberration: different colors have different focus points. Guess what? Both camera processors correct for all of these errors: this is what “adaptive” does. It adapts and corrects. In other words, there are algorithms (and intelligence) that measure and recognize errors in the system (here, the camera and the optics) that may vary depending on the device and circumstances, then make the proper corrections in real time such that the end product is nearly free of problems. The smartphone industry cleverly shifted the burden of camera performance from expensive and sophisticated lens manufacturing (what it used to be in the past decades) to inexpensive computation. Brilliant!

It becomes immediately obvious to the reader that the biggest beneficiary from this “adaptive” performance is the inexpensive plastic lens used in the iPhone 6 Plus. In other words, the benefit of shifting the burden to computation is the use of lower cost components, in this case, a lower cost sensor and lens, albeit with worse optical specifications. And I mean much lower cost: in this example here, it is about 100X less expensive.

Adaptive systems are not new by any stretch of the imagination. They were initially proposed and used in complex systems — for example, correcting the optical errors in large telescopes as a result of variations in the upper atmosphere. However, the rapid decline in the cost of computing over the past decade has made the implementation of “adaptivity” accessible across a broad range of applications.

So, now you can begin to imagine what adaptive solutions can do to improve the performance of batteries where materials and manufacturing can have significant variability and associated costs. This will be the topic of a future post.