Augmented Reality Is The Future Of Work

LinkedInEvernoteFacebook

VR 01

Your day-to-day computational work is about to change. Why? Because you don’t like sitting in a chair in order to interact with digital aspects of the world around you. Your mouse annoys you (albeit subconsciously), you don’t like opening apps and it is difficult for you to pay attention.

Imagine for a moment that your computer vanishes. The room goes white and suddenly you are in SpaceX headquarters doing what you always wanted: designing rockets. Yes my friend, you are now a rocket scientist. Impressive! The company has recently completed a successful vertical landing, bringing reusable rockets closer to reality and significantly reducing the cost of space exploration.

Elon Musk comes to you and asks you to further reduce the cost and increase the reliability of the vertical landings for future missions to Mars.

VR 02

From a wobble projected by predictive models simulating vertical landing on Mars, you have a hunch that your Merlin Engines  are over correcting in the final 35 seconds of descent, meaning that more energy is being ‘thrust’ into the system than necessary (pardon the pun).

Context-based User Interface

VR 03

As a SpaceX employee, you carry with you an Augmented Reality Headset, a pair of glasses allowing projected hologram rendering in three dimensions. You don your headset on and see a menu based on your location, context and calendar. Your schedule contains associated files, programs and tasks meaning that you don’t need to open anything else. It is all there, waiting for you.

Eye Sight Path Prediction

VR 04

You can either use eye gaze or hand motion to explore your schedule. When you start working with the Mars simulator information, your attention is drawn to an analysis the company’s data science team performed the other day. You engage it.

As the chart comes up the green brackets draw your focus to two 270 degree thrusters that are both near the upper sigma limits, indicating that they could be outliers compared to previous computational predictions.

Accounting for Randomness

VR 05

Recall that sigma limits are three standard deviations from a mean. If a measurement lies outside of a sigma limit then it may not be random. Otherwise most measurements that deviate from the mean but remain within sigma bounds can be attributed to randomness.

The chart you see in the SpaceX demo shows how our measurement now compares to rocket thrusts from previous descents. One could assume that the 270 degree engines vary too much in their individual outputs at approximately 22 seconds in the descent of the rocket, causing the wobble

Thrust Variance by Time

It’s time to roll up your sleeves and get to work. You acquaint yourself with a polar graph showing current rocket thrust variance from the mean of past launches. You can now see where in the descent groups of thrusters are firing and how that impacts over correction and fuel consumption.

VR 06

Your novel idea is to run various scenarios through your flight simulator to see if total fuel expended decreases and reliability increases. After several trials you find a combination of timing and thrust that improves trajectory and conserves fuel. That will make Mr. Musk and the Board of Directors happy. Maybe they’ll move you to the Mars mission planning team.

Working with Augmented Reality

So why is Augmented Reality going to replace your PC? Because the world is not two-dimensional and static. Even if you aren’t designing rockets you need to see your data and analytics in the context of the problem you’re trying to solve.

Often, the data, the analytics, and the problem itself is three or dimensions. We need to be up, discovering, inspecting, and interacting with the world to inform our decisions and analyses. Sitting at desks working in two dimensions and asynchronously pinging static sources of data gives us too limited real-time inputs to be effective. It is the reason why even if you have three monitors at your workstation it still isn’t enough.

Still not convinced? You don’t need to be. You brain is largely dedicated to visual processing. Like it or not you reason through visual metaphor. Even Nikoli Tesla, physics god and namesake of Musk’s electric cars, said he could visualize all the moving parts of his machines in his mind.

You can check out a prototype of my AR interface design.

Special thanks:

  • UX Designers Ouliana Love and Virgilio Guevara for helping me iterate through the radial graph.
  • SpaceX for putting up the video using the leap motion. I used it as a background.
  • Physicist Paul Hofmann for helping me understand my chart does not approximate reentry to earth and may be better adapted to far less dense atmosphere.
  • Zaid Tashman, Data Scientist who found some fallacies in my original chart.
  • Authors Patrick Blau and Am Morgenberg for their break down of rocket re-entry.
LinkedInEvernoteFacebook

Predicting and Managing Urban Storm Runoff to Protect Roads

LinkedInEvernoteFacebook

 1200px-FEMA_-_37361_-_Flooded_road_in_Texas

It’s El Nino time in the West, which means our cities see storm water runoff from heavy winter rains. Large pools of standing water form on roadways where there’s no drainage, drains are clogged, or the storm water system overflows. So much water on and around roads leads to accidents and traffic jams. Water also degrades road surfaces and erodes road beds, creating more hazards and eventually requiring road repair work.

Viewed as a challenge for situational intelligence, storm water and traffic becomes a compelling example of what’s possible. Situational intelligence features analytics of networks and space / terrain across time. Storm water and traffic is a problem of two networks–drains and roads–during the time when heavy rain is moving through the terrain.

Analytics can pinpoint crucial aspects of storm water runoff, such as:

  • Places where roads and drains intersect
  • Road and drain segments that need repair or replacing
  • Bottlenecks in the flow of water and cars
  • Times and locations where rainfall will be heaviest
  • Locations of road and drain repair crews and police

Of course, meteorology continues to benefit from more sensors and computer power, making weather forecasts more accurate and providing earlier, more specific warnings of storm activity.

Once you have all this data and analysis in a single situational intelligence application, you can start to improve storm water operations and planning.

In operations, you could notify police and crews before the storm hits, so that they can be in place to help. You could also notify news agencies and social media outlets to announce what streets to avoid to prevent accidents and traffic jams.

In planning, you can predict areas where runoff will be heaviest. By overlaying this data with data about road and traffic conditions, you can target your storm system investments to places of highest risk for road damage and traffic problems. Taking a proactive approach helps you stretch your operations budgets further while also positioning you well for the next storm.

The West needs the rain to counter the past four years of drought. We’re not complaining. We just need a way to better manage the impact of heavy rains on our heavily traveled roads.

+++

LinkedInEvernoteFacebook

Three Rules For User-Centered IoT Analytics

LinkedInEvernoteFacebook

In a recent TechTarget article, Maribel Lopez of Lopez Research says that manufacturing may have a head start in implementing Internet of Things (IoT) solutions, but she still sounds skeptical about IoT in general.

IoT “is a lot of talk and not a lot of action,” she says. “First of all, the phrase ‘IoT’ is meaningless because it doesn’t talk about anybody doing anything that’s useful. Just connecting your stuff is not enough.”

How do you make IoT solutions that are more action than talk? Lopez cites three rules for user-centered analytics in the Internet of Things:

  1. Be relevant to users:  Pushing data to users just because it’s possible is not helpful. Information presented to users needs to be relevant to a task or situation that needs attention. For instance, reporting that vibration in manufacturing equipment is within acceptable limits is of little use. Such information requires no action from the user.
  2. Do the work for users: Performing analysis for users is more useful than equipping users to perform their own analysis. Business intelligence tools may make a table of vibration data  easier to manipulate and visualize, but that manipulation and visualization work takes users away from their main task of operating the equipment. As Lopez says, “Saying that the vibration [of manufacturing equipment] is out of range is interesting yet not sufficient.”
  3. Be timely for users: Presenting users with exception data in context with time to react has far more value to users. That keeps users on task and ahead of potential issues. As Lopez says, “Saying the vibration is out of range and if it continues for the next two hours, it’s going to shut down the plant — that’s more interesting.”

Situational Intelligence abides by these rules by turning big data to little data, focusing users on events or conditions that require attention. It’s not looking at all the data that counts; it’s looking at the right data at the right time.

LinkedInEvernoteFacebook

Demystifying Analytics

LinkedInEvernoteFacebook

black box

Unlike many other enterprise data processing solutions, analytics is viewed as a mysterious black box. Historically some analytics and models were so complex that only experienced IT professionals could execute analytics jobs on costly large computing platforms. IT personnel, database administrators, software developers, and other skilled personnel were necessary to interpret the output of analytics to arrive at the proverbial “answer.” In such cases the time-to-answer could be hours, days and even weeks.

The air of mystery especially envelopes analytics that derive likely outcomes (a/k/a “predictive analytics”). Predictive results are actually likely outcomes derived by processing large volumes of data using specific mathematical and statistical methods. Nevertheless some say colloquially that predictive analytics can predict the future. Statements such as this add to the mysticism about analytics.

Today’s vocabulary of analytics increases the aura of mysticism: Hadoop, big data, artificial intelligence, machine learning, data science, stochastic optimization, etc.

Because of this mysticism and the seeming ability to predict the future, people also have a notion that analytics is an elite category of software available only to large enterprises with large budgets, extensive IT infrastructures and dedicated teams.

The good news is that specialized terminology, rarified skill sets and expensive machinery no longer confine analytics to elite glass houses. As the simplicity of analytics becomes more commonplace, the aura of mysticism evaporates. The image of an esoteric technology for an elite few fades, giving way to adoption by a broad range of workers and end-users.

Powerful yet affordable commodity hardware and other technological advances make it possible for organizations regardless of size, budget and personnel to obtain and run analytics, consume and act on the results, and realize the many benefits. Software advances such as distributed high performance computing platforms and alternatives to traditional relational databases, to name a few, bring analytics within the grasp of any organization. Advances in data visualization remove the need to post-process analytics results into readily consumable and actionable answers, whether the answers are recommendations, predictions or other forms of insight. Ubiquitous communications, modern browsers and applications that take advantage of HTML5 greatly simplify the ability to deploy software solutions of any type and complexity, including analytics.

This new demystification and accessibility comes just in time. For organization to benefit from their data, personnel must be able to receive, comprehend and act on alerts, recommendations, predictions and all other insights generated by analytics. Without this operationalization of analytics, data threatens to flood the organization without providing any value.

As analytics become more accessible, their use and results will be embedded in many applications, which in turn hides complexity and helps make analytics ubiquitous. I eagerly look forward to many new and innovative uses of analytics, and the resulting business and societal benefits.

LinkedInEvernoteFacebook