Analytics, Transportation Safety And Extreme Weather


transportation safety landslide 01

Extreme weather conditions such as floods, landslides, sinking water tables, droughts, ice storms and snow drifts takes a toll on the landscape. Those weather-driven changes to the landscape directly impact rail beds, roadways and bridges, rendering them unsafe to use until conditions change and repairs are made.

For instance, heavy rains trigger landslides that block roadways and railways, stopping traffic until the debris is removed and the damage repaired.

Analytics can highlight where and when disasters might strike. Predicting where damage may occur and taking precautions prevents damage to people and property.

Factors that signal a landslide or flood hazard include

  • Transportation corridors in proximity to sloped terrain or bodies of water
  • Vegetation, or lack thereof, on slopes and banks
  • Moisture levels in the soil
  • Past, current and projected weather conditions
  • Condition of the road or railway itself
  • Traffic metrics for the corridor including overall volume of traffic, patterns of traffic flow, and criticality of the corridor in connecting valuable locations

Imagine a situational intelligence approach to transportation safety. An analytics application could correlate, analyze and visualize the factors listed above. That analysis and visualization would enable government transportation officials and railroad operations and planning professionals to see the probability of landslide on a slope adjacent to a right of way. Knowing the probability of a landslide and the magnitude of its impact informs decisions about operations, crisis response and mitigation.

A similar situational intelligence approach applies to floods damaging transportation corridors that are adjacent to bodies of water. Some of the measurements and algorithms would be different, but the much of the output and outcomes from such a system would resemble that from landslides.

When public transportation departments and private transportation companies know that a corridor is at risk, they can take action to avert disasters. Traffic and shipments can be redirected to minimize delays and remain safe. Customers can be notified if their shipments will be delayed and new arrival times. Repair crews can be positioned for faster response to the most critical areas.

Analytics for transportation and weather is especially important in the United States given the current condition of transportation infrastructure. The American Society of Civil Engineers give these systems low marks:

Infrastructure that is in barely passable condition is less resilient to the impact of extreme weather. If we can’t change the weather and don’t have the money to improve the infrastructure, at least we can be smart about how we plan and respond.

(Image courtesy of fotokostic / 123RF Stock Photo)


Does This ‘Hologram’ Fly?



An Augmented Reality Headset Interface
An augmented reality headset interface

Holograms are awesome. Their power lies in allowing humans to share imaginations, which is why they work great in movies. But what are holograms well suited for, other than for entertainment purposes like Star Wars special effects  or resurrecting Tupac at Coachella in 2012? Well, while what we are calling ‘Holograms’ are technically just an optical illusion called Pepper’s Ghost, they are very useful for other applications. From here on out we will refer to ‘pseudo’ holograms as Augmented Reality.

An optical illusion of 2Pac in 2012 using Pepper’s Ghost

As Tupac rapped, “Reality is wrong. Dreams are for real.” I think he meant that dreams allow us to simulate all our hopes and fears and consider the best outcomes. Augmented Reality allows us to ‘dream’ into the future and make really smart decisions affecting our present.

Can AR be used for business application? Sure! AR is powerful for visually exploring predictive models that are registered to our physical surroundings.

For example, take an air traffic controller who just started a position in the all new, $126 million control tower that goes online July 28 at San Francisco International (SFO).

The new control tower coming online in July 2016

Did you know that SFO as well as many other air traffic control units have been operating by paper strips until last year?

paper strips
A paper flight ticketing system still in use today

The FAA is currently implementing an upgraded air traffic control system called NexGen in ‘pilot’ airports (pun intended). But this is still limited to a screen. Air traffic controllers will still need switch between looking at information regarding the plane and then the plane itself. They also have to look into the future.

As part of NexGen, why not simplify the air traffic control job by melding information about the plane with a hologram of the plane itself and allow for direct manipulation of the scenario through modern UI? Enter Holograms.

The Use Case:

SFO air traffic controllers are directing planes during the holidays. They need the ability to look into the future and see where issues may arise. For instance a storm might be affecting arrival times for planes coming in from the east. Our controller needs to see where margins are thin for collisions and potentially reroute.

What Is On The Horizon?

Orange outlines represent closer planes, white dots show planes further away
Orange outlines represent closer planes, white dots show planes further away

Planes that are far off in the distance can be represented by white dots of varying size. Size will represent spatial distance. When planes come within a predetermined distance they can be rendered as low-fidelity holograms. This plane is orange since it is triggering a warning. Other planes are white since they are all speculative.

Radar Projections

With an augment reality interface, a controller can reach out to a series of radars showing simulated or real-time inbound and outbound planes by time of day. There is no clicking to do, just grabbing and moving hologram-like objects.

radar projects
Radar time machine for controllers

For example, here is a radar timeline that projects all the events and planes happening in real time. They are arranged in a circular dial that spins and expands to show you the radar at the specific time you are interested. The white box corresponds to the largest circle with two orange dots. If you dialed the radar back to the blue circle (current-real time) then the box would move to the ‘Now’ indicator. Obviously the present is always moving so the word ‘Now’ would move with time. Anything projected or in the past takes on a white color. Once controllers select a radar matching the busiest time of day, they can verbally tell their computer to, “Show projected warnings” or grab the two orange dots that would slightly expand as their arm approaches.

Flight Simulations

Flight simluation
A traditional pop up window in AR showing a warning

Once they interactively activate the warning dots, controllers would see low fidelity planes landing, departing and taxing along the runway. Gazing or moving an arm towards a plane would populate an information window displaying information and a warning for that asset, where relevant.

Closer Examination

closer examination

Controllers can grab the virtual plane by reaching out, at which point its display quality would increase to a CAD model with more detail. The philosophy behind this design is that you start at a high level and then are offered more detail as you interact with items. This alleviates the need to have an explicit file structure. You can spin the plane, open it to see section, or even view pilot and system information.


Augmented reality software engineer Tyler Lindell says, “Using augmented reality and interacting with real-world objects will take us beyond the barrier we have known for the last 40 years, the personal computer.”

Computers should feel natural. You should be able to use any interface and feel like you have always used it. No more opening applications. No more going through file systems. No more explicit metaphors trapped in 2D ‘planes’ like icons and spreadsheets trapped in a PC.

As imaginative as this prototype AR interface is, it is likely to have myriad shortcomings and flaws. That is why I’m asking you, the reader, to complete this survey regarding this UI. Please be brutally honest.



Why We Need Situational Intelligence, Part 2


Big Picture 01

In my previous blog I addressed the need for situational intelligence (SI) as an approach to decision-making that combines insights with relevant context to create the big picture we need to make the best possible decisions with the lowest risk. I concluded that blog by promising to explain why and how various technologies such as data access and fusion, analytics with machine learning, artificial intelligence and visual analytics come together to support situational intelligence.

Situational Intelligence itself is not a technology, nor can you use just one technology to create it. Rather, a situational intelligence approach requires a combination of integrated technologies. The main types of technologies are listed below. Seamlessly integrating these technologies creates actionable insights that are especially applicable for real-time operational decision-making.

  • Live connections to data, both at rest and in motion, in a variety of formats and structures (including no structure at all). Access to multiple, disparate sources of data provides the context for new, deeper insights. Connecting directly to data creates great efficiency and savings savings because data access and preparation often consume as much as 80% of the effort of making data-driven decisions.
  • Analytics, big data, and streaming foundational technologies (such as Spark, Hadoop, SAP HANA) that are inherently scalable and enable high-performance execution of analytics and processing of large datasets. These foundations are typically distributed and use in-memory processing so that complex software executes and generates answers and insights as quickly as possible. Streaming message brokers such as Kafka and Internet of Things (IoT) platforms are also necessary to manage streams of data that can be passed through streaming analytics for real-time insights and/or to be stored for inclusion in subsequent applications of advanced analytics that derive deeper insights.
  • Advanced analytics and streaming analytics that derive insights from the data at rest and data in motion, respectively. Because situations inherently occur at specific times and locations, the ability to correlate spatial and temporal relationships increases the insights that can be derived. Similarly, the ability to correlate entity-to-entity relationships increases the insights by revealing actual and likely ripple effects. Altogether these analytics make it possible to identify the what, when, where, why and how of situations that happened or may happen. In addition, machine learning allows the analytics to adapt to your data and to your use cases.
  • Visual analytics is essential to complete the transformation of data into actionable insights. Intuitive renderings of the relevant data and resultant insights derived by analytics helps users comprehend and acted on data at-a-glance. Output from visual analytics included in alerts via email and SMS is a powerful way of notifying people about critical matters and focusing their attention on acute situations and the decisions to be made.

In summary, situational intelligence is an approach that combines data and analytics, including visual analytics, to aid human decision-making. Insights from advanced analytics and streaming analytics are combined with relevant underlying data to create context so that decision-makers have a complete understanding of each situation and make decisions that lead to the best possible outcome.


Saving Lives On The Job With Analytics – It’s Easier Than Curing Cancer


Construction workers blog

A University of California, Davis study calculates that workplace accidents and illnesses cost the United States economy $250 billion annually. That’s more than the direct and indirect costs of cancer. That’s more than diabetes and strokes, combined.

The good news is that workplace accidents and illnesses are easier to prevent than cancer, diabetes and strokes. That’s in part because accidents and illnesses happen at a known place and time: on the job.

That sounds obvious, I realize, and a little trite given the seriousness of the topic. But as we know with situational intelligence, understanding the place and time of an actual or predicted event is powerful.

For example: according to the U.S. Bureau of Labor Statistics, 4,679 workers in the US died from workplace accidents in 2014 (the most recent year for which we have data). That’s an awful number, but also a very general number. It doesn’t tell us who, where or when. Can a situational intelligence approach to safety help detect accidents more quickly and even prevent accidents?

To be more specific, 20 percent of those workplace deaths happened in the construction industry. That tells us a lot more about who, where and when. Furthermore, the BLS reports that the nearly half of all construction deaths are the result of falling or being electrocuted.

Now we have two specific and preventable accident scenarios within a single industry.

Imagine that you’re the safety director for a construction company. Receiving a report every morning of the day’s planned elevated and electrical work gives you insight into where accidents might happen later that day. You’d have time to review safety equipment and procedures with workers before they start their tasks. That’s a good first step towards a safer workplace.

Since we’re now in the Internet of Things era, it’s easy to envision that your workers wear safety vests equipped with GPS and other location technology. Knowing the location of all your workers in real time is another step towards protecting them. The vest tells you when a worker is more than, say, three floors off the ground, or within 10 feet of high-voltage equipment.

Warnings about these potentially dangerous situations are pushed to your laptop and also to your mobile device as you’re out on the site. This information helps you respond faster to accidents, since you know exactly where people are working

Analytics can go further, to help prevent accidents. Knowing that falls and electrocutions are top priorities, you can design new measures to restrict access to dangerous areas. By correlating personnel records with scheduled tasks, you know which workers have the certification and experience to safely work at heights and with electrical systems. If a worker is potentially distracted by daydreams about an upcoming vacation, the personnel records flag that potential hazard for you to address during a safety briefing.

These are just a couple of high-value use cases from the construction industry. Similar use cases exist in nearly any industry where workplace safety can be an issue such as utilities, mining, transportation, agriculture and manufacturing.

Analytics is broadly applicable to reducing the impacts of workplace accidents and illnesses. Those impacts include hard costs associated with insurance premium and claims, litigation, compliance and other expenses that contribute to that $250 billion annual figure. There are also soft costs associated with employee morale, customer dissatisfaction and company reputation.

Improving safety with analytics may be more mundane than curing cancer, but also more achievable.

Image Copyright: hxdbzxy / 123RF Stock Photo