Distilling Insights from Tomorrow’s Digital Exhaust
The robotics industry is at a turning point. We’re leaving the industry’s open, exploratory infancy, and entering the more rigid realm of maturity. During this period, we will see some hard lines start to form—exploration will yield to well-worn paths, and experimentation will take a back seat to best practices. And while there will always be space for innovation, this landscape will be the terrain on which our industry’s titans take root.
As someone on the “front lines” of this industry, I can feel this sea change coming. And the companies poised to “come out on top” after this seismic shift aren’t the types of companies most imagine. In the near future, the titans of robotics will be those who deal in data, not devices.
Until recently, robots were largely confined to the safe, predictable confines of the factory floor—bolted down to fixed positions along assembly lines. However, for today’s generation of robotics, one of its most recognizable characteristics is mobility. In addition to its literal interpretation, “mobility” also speaks to robots moving into entirely new sectors of our economy. Increasingly sophisticated technologies are enabling robots to perform increasingly complex tasks, in increasingly complex environments. And, as a result, managing these robots is growing increasingly complicated.
Meanwhile, as they begin to work with and around humans, the need to monitor, analyze and observe their actions grows. Whereas metrics like up-time and down-time were once substantial parts of monitoring robots, today’s applications demand much more. Today, we must monitor processes, not variables. And as applications grow even more sophisticated and diverse, the need for (and the complexity of) observability skyrockets.
SENSORS, SENSORS, SENSORS
When we talk about robots occupying novel spaces, it’s important that we understand “how” they’ve made this shift. The short answer is: by ingesting, processing and transmitting lots of complicated data. The transformative nature of modern robotics data isn’t merely a question of total data volume. Nor is it simply the rate at which data must be ingested and processed. It is also a matter of the types and variety of data a single machine must work with in order to reliably perform its role.
To help illustrate this idea, let’s look at a real-world use-case. Amazon (among others) has recently started using robots in their distribution centers to perform “pick & pack” labor. These robots are given product codes and tasked with locating items, retrieving them and delivering them to packaging station.
In order to perform even such a small number of discrete tasks, a robot must be equipped with a litany of sensors, and be able to rapidly process countless data types. In this case, we can assume that the robot will require a variety of motion sensors—such as an accelerometer and odometer—plus various geometric sensors to determine physical orientation.
Already, we’ve identified over 4 sensors. And yet they address only a single aspect of the robot’s functionality—movement. It isn’t at all uncommon for robots to employ twice as many sensor types, monitoring variables such as temperature, location, video monitoring and 3D data—for example lidar and depth cameras.
GOOGLE AND DATA
When we look at the history of digital data, there is one organization whose import is so massive that its shadow often blots out every other actor in the field—namely, Google. In 1998, Google was born as a small-scale search engine company. However, it wasn’t long before it became the most popular search engine in the world. 21 years later, Google retains its position as the #1 search engine in the world. But that status is no longer its greatest claim to fame. Today, Google is also the world’s most powerful advertising company. Currently, advertisements account for roughly 90% of Google’s total annual revenue, which amounts to roughly one third of the world’s digital ad revenue.
The nature of Google’s transformational ascent is better left for another time. However, the question of how Google managed to achieve such market supremacy has never been much of a secret. In fact, Google indirectly identifies it in its public mission statement: “Google’s mission is to organize the world’s information and make it universally accessible and useful.”
While the “universally accessible” clause is debatable, the mission statement is remarkably direct. It can also be used to cast light on robotics’ current data problem. First, one must ask:” Why would a search engine startup remake itself into an advertising megalith?” Well, in their efforts to “organize information and make it accessible and useful,” they realized something extremely important: information has the potential to be priceless. The operative word there, however, is “potential.”
The true value of Google’s analytics operations doesn’t lie in the data itself. It lies in its processing. It resides in their efforts towards “organizing” that data and making it “accessible” and “useful” to its customers.
DATA, DETAIL, VALUE
The next time you’re on a bus or a train, take a moment to look around you. Take into consideration that nearly 96% of the people around you have cellphones in their pockets (or in their hands), and about 79% of those phones are smartphones. If you happen to be in New York and riding on a busy metro line, that’s roughly 50 smartphones in just your train car alone—all of them churning through untold amounts of data.
Now, try to visualize all that data streaming out of those phones like exhaust. The sheer volume would choke the car with an impenetrable fog of zeros and ones. Now, consider the fact that the average robot transmits 10 times as much data as a smartphone (Figure 1). There’s more of it, more metrics are involved, more complex data types are used (for example, streaming video) and the robot operates continuously.
If each smartphone is a single tailpipe, then, each robot is like a billowing smokestack. Left alone, that digital exhaust is simply lost—left to vanish into the ether. But, if we’ve learned anything from recent history, it’s that such “exhaust” should not be left on the table. And despite being practically ubiquitous, this exhaust has the potential to be more precious than gold. Like crude oil or iron ore, digital exhaust is a raw material. When collected, processed, and stored effectively, it becomes something else entirely. It becomes insight.
Although Google is a useful analogue for our discussion, robotics needn’t reinvent itself as an advertising mechanism. The insights gained from robotic data could be used to answer countless questions; including everything from customer attitudes to optimal speed in low pressure environments.
In the future, robotics will play a role in virtually every industry imaginable. For each of those industries, there is an immeasurable amount of value to gain from the distillation of insight from digital exhaust. And as the role of robotics matures, both the stakes and the challenges are bound to rise. The time to start tackling this problem is already upon us. Any company considering employing robotics in their operations should already have a plan in place regarding how best to make use of their robotic data. Because, those who don’t will invariably be left behind.
Formant CEO Jeff Linnell brings an unorthodox approach to robotics. Part Cinematographer, part Creative Director and part self-taught Engineer, Jeff is a serial robotics entrepreneur and a thought leader.
Formant | www.formant.io
PUBLISHED IN CIRCUIT CELLAR MAGAZINE • MARCH 2020 #356 – Get a PDF of the issueSponsor this Article