Title Image

Driverless Cars Hit the Streets — How Do We Prepare for the Network Effect? – a BoldIQ byline in Wired.com

Remember the 2054-model Lexus that Tom Cruise drove in the film “Minority Report”? Or who could forget Kit, the Knight Rider’s ultimate ride. Well it seems that Google is showing us that autonomous driverless cars are now a reality… at least from a technological perspective.

And Google is not alone in this race as it seems many of the major auto manufacturers are putting in the effort to be the first to market with an “autobot.” If we consider some of the actions taken by the United Nations recently to enable the use of driverless vehicles on the roads of Europe, it seems that European car makers believe they might be ready to lead the way.

I wonder if auto manufacturers and regulators alike are looking beyond the single vehicle technologies. There is much more that needs to be considered beyond the ability of a vehicle to swerve away from a pedestrian, or brake before hitting the car in front of it.

We need to anticipate and be ready for a highly complex dynamic network of vehicles – fleets of unmanned vehicles in aggregate. If we plan and execute the network correctly, and in a dynamic fashion, we will in fact need fewer vehicles (be they manned or unmanned) to perform the demands of our day to day activities, be they deliveries, public transport, and the like. To that end, I propose regulators and auto manufacturers take some key factors into account:

Anticipate Dynamic Complexity and Unpredictability:

Consider aerospace and what is happening with UAVs (otherwise known as drones). These highly technological pilotless planes are the subject of a raging battle between the FAA and the prospective (and in some cases already operational) users. Legitimate concerns around safety, security and more, are all part of the debate going on right now. And all of this is in one of the most controlled environments around – airspace. Known flights and flight paths, constants speeds, overly conservative separations between aircraft, air traffic control, etc.

If this is what is happening in aviation — where the National Safety Council calculated the odds of dying in a plane as 1 in 7,178 for a lifetime — can you imagine what will happen in our streets where the odds of dying in a car crash are 1 in 98 for a lifetime?

The movement of vehicles on the roads are unpredictable and not very controllable even without adding disruptions into the mix. This drives a need for real-time dynamic decision-making technologies that will match the complexity and speed of this system.

Leverage the Network of Vehicles:

There are two layers of the network that should be addressed combining both safety and efficiency. The first layer is the individual vehicles themselves. I am confident that companies like Google and Mercedes-Benz are doing a phenomenal job addressing this layer. The second layer, the aggregation of vehicles be they a fleet of delivery trucks, a fleet of buses, or a fleet of on-demand taxis, is where the biggest impact will be on traffic, congestion, and waste. By leveraging the network using sophisticated optimization tools, the number of vehicles per fleet, and in aggregate, can be significantly reduced creating positive ripple effects throughout the operational, environmental, and financial aspects of this network.

Prepare to be Adaptable in Real-Time:

This network needs to be structured, and yet rapidly adoptable and adaptable to meet the needs of the public from both the efficiency aspect as well as the safety one. It is one thing for a single vehicle to sense a jaywalker and apply the brakes. It is another for the network to adapt itself in real-time to the “butterfly effect” of that car braking — the other cars behind it braking in succession; the manned car tailgating the unmanned car not braking in time and colliding into it; the traffic jam created due to this incident; the ensuing delays; the delivery service needing to adjust its plans to meet its same day food delivery commitments, or the bus service needing to somehow get commuters to work on time. Using adaptable and adoptable optimization software in real-time, will enable a transportation system that operates effectively and efficiently.

As we progress with great excitement and anticipation into an on-demand world, one which blends manned and autonomous resources to meet our needs, we must think bigger and wider than the single car, a single package or a single bus route. I look forward to the not-too-distant future in which optimizing real-time data being streamed in from the ‘autobots’ can make responding and adjusting their usage in the real-world possible, and give the public and regulators some of the comfort needed as we explore the realm of possible.

Click to see story online

The Modern John Henry: Why Executives Are Wrong to Favor Intuition over Analytics – by Loraine Lawson in IT Business Edge

Ironically, one of the most common barriers to companies adopting analytics isn’t the business case, but executives’ faith in their own memory.

Roei Ganzarski, CEO of the predictive analytics firm BoldIQ, told Information Week that he daily encounters leaders who reject the idea of using analytics over their own business intuition.

Ah, man versus the machine — it’s an old, old debate for literature, philosophy, and more recently, brain researchers. To be honest, the more I read, the more I think the gig is up for business intuition.

As least, that’s what the data says, according to Andrew McFee, co-director of the Initiative on the Digital Economy in the MIT Sloan School of Management. According to McFee, “…there have been a raftload of studies comparing the predictions of human experts vs. those of algorithms, and …in the great majority of them the algorithms have been at least as good as or significantly better than the humans,” McFee wrote in a Harvard Business Review blog post. “In a meta-analysis conducted by William Grove and colleagues of 136 research studies, for example, expert judgments were clearly better than their purely data-driven equivalents in only eight cases.”

In fact, with what we now know about the brain, the harder it gets to justify intuition over algorithms. From memory formation to mental fatigue brought on by too many choices, research tends to favor the idea that we are much more fallible than we would ever want to believe.

Marcelo Gleiser, a theoretical physicist, natural philosopher and professor at Dartmouth College, expressed this problem eloquently in a recent NPR post.

“We can thank the brain for tricking us into building a sense of the ‘real,’” Gleiser writes. “What we call reality is the result of our brain’s very complex integration of external stimuli: sights, sounds, tastes, touch and smells. We perceive nothing in the actual present.”

To be fair, the same is somewhat true about analytics, despite all the hype about real-time analytics. The difference is that analytics is based on a collective sense of reality — not one person’s reality. Or as Barrett Thompson, GM of pricing excellence solutions for Zilliant, told Information Week, predictive analytics is “…the distilled wisdom and experience of five hundred salespeople who encountered tens of thousands, or hundreds of thousands, of unique selling circumstances.”

The pro-intuition crowd isn’t without a champion, however. Analytics expert Tom Davenport has written about the important role of intuition in Big Data projects.

I checked out Davenport’s full HBR post. It’s worth noting that his examples aren’t so much about decisions business leaders made, but rather about new ideas or theories they had that were confirmed or, in many cases, actually made possible by Big Data.

To me, Davenport’s examples are creative acts — which actually fall outside the domain of many executive decisions, if you think about it — even the strategic one.

As Information Week notes, what we’re talking about is significant, but daily, decisions that are largely based on guessing—not facts—even though you have access to the data.

For example: Which route is the most cost-effective for our delivery drivers? Where are we making overpayments or fraudulent charges?

Businesses often have this data, but struggle to analyze it. This is where executives play John Henry, stubbornly fighting the machine despite what’s best for business. But in such cases, Big Data analytics outplays intuition.

Click to see story online

Big Data Debate: Do Analytics Trump Intuition? – by Jeff Bertolucci in Information Week

C-suite types often scoff at the notion that software can outperform insights acquired from years in the trenches. Fear not: Analytics will complement, not replace us.

Will the next evolution in big data remove human intuition from key business decisions and rely exclusively on data-driven analytics?

Probably not, but organizations will increasingly rely on analytics to make real-time decisions based on a rising tide of big data, predicts Roei Ganzarski, CEO of BoldIQ, a Seattle-based optimization company. This doesn’t mean, however, that managers warmly embrace analytics over intuition. In fact, the opposite is often true. C-suite types often scoff at the notion that software can outperform insights acquired from years in the trenches.

“We run into that every day,” says Ganzarski in a phone interview with InformationWeek. “Our response is, ‘It’s not that we think you’re doing something wrong.'” Rather, he says, analytics provide “an additional tool that enables you to do things the human mind simply can’t do. We’re not here to replace you. We’re here to enhance your ability to make decisions.”

Obviously, the human brain still trumps computers at myriad tasks. But software is significantly better at split-second analysis, he says, providing a transportation industry example. “Let’s talk about next-day and same-day deliveries. People talk about that as the Holy Grail of distribution and supply chain — where you make an order and it shows up on your door that very day.” This presents the sort of big-data challenge suited for analytics “Once the order has been put in place, how does [the delivery company] make sure its vehicle-and-driver network is set up in such a way that, within milliseconds, it can tell you which vehicle and driver should do the pickup and delivery so that everything is done on time and in a profitable manner?”

The Human Algorithm: Barrett Thompson is general manager of pricing excellence solutions for Zilliant, an optimization company that helps businesses use in-house data to make better sales and pricing decisions. In a January interview with InformationWeek on the algorithm-versus-intuition topic, Thompson pointed out that the algorithmic approach to decision-making is based on the collective experiences of multiple individuals within a business.

“Let’s say I’m looking at set of sales transactions completed over the past year, and I have five million of them sitting in the data warehouse,” says Thompson. “What I have captured in that data, what I have to find a way to unlock, is the distilled wisdom and experience of five hundred salespeople who encountered tens of thousands, or hundreds of thousands, of unique selling circumstances.”

A predictive algorithm, he says, doesn’t create knowledge out of thin air. Rather it’s an “accelerator” of what an organization already knows. It’s driven by human experience, by “data points you’ve lived through.”

“I can’t see what the other 499 salespeople did, and I can’t even remember what I did in March of last year. I make so many decisions that they become lost in memory,” Thompson said. “If I had a software tool, an algorithm that could remind me of what I’ve learned, and reveal to me what everyone else has learned… I could look at the guidance coming out of the algorithm as a distilled and refined experience from myself and people just like me.”

Of course, the big caveat here is that Thompson and Ganzarski are heavily vested in the analytics camp. Neither is an unbiased observer of the intuition-versus-algorithm debate.

In a December 2013 Harvard Business Review blog on intuition’s role in a data-driven organization, analytics expert and author Tom Davenport argued that major big-data projects are often driven by intuition:

Major big data projects to create new products and services are often driven by intuition as well. Google’s self-driving car, for example, is described by its leaders as a big data project. Sebastian Thrun, a Google Fellow and Stanford professor, leads the project. He had an intuition that self-driving cars were possible well before all the necessary data, maps, and infrastructure were available. Motivated in part by the death of a friend in a traffic accident, he said in an interview that he formed a team to address the problem at Stanford without knowing what he was doing.

Fair enough, chalk one up for intuition. But in the enterprise, analytics will play an increasingly influential role in the intuitive process.

Jeff Bertolucci is a technology journalist in Los Angeles who writes mostly for Kiplinger’s Personal Finance, The Saturday Evening Post, and InformationWeek

Click to see story online

BoldIQ Team Driverless cars are cool, but the network layer that will run them is even cooler – a BoldIQ byline in Venture Beat

Google has shown us that autonomous, driverless cars are now a reality, at least from a technological perspective. These vehicles are capable of reacting to unforeseen disruptions such as jaywalkers, roving bicyclists, or slowing traffic.

It is exciting and scary to think of the world ten years from now replete with these technological advancements. Our world will need to adjust.

Consider the air, where unmanned aerial vehicles or drones are already being used – and where a fierce battle is being waged between the FAA and prospective and current drone users. Concerns around safety, security, and more are all part of the huge debate in one of the more controlled environments we have: our airspace.

If this is what is happening in aviation, can you imagine what will happen when “autobots” hit our streets?

The environment of ground transportation is much larger and more dynamic than aviation, yet it has significantly fewer (if any) regulations in place. Drivers can go anywhere using any route they chose. Travel plans are not known in advance; and speed, while supposedly controlled by rules of the road, is often unpredictable and inconsistent.

All this causes near-collisions every day with licensed and trained humans behind the wheel. Imagine what will happen when we throw driverless vehicles into the mix. This is a highly complex dynamic network that requires real-time tools to manage it in order to be safe and efficient.

To do this, there are two layers of the network that should be addressed, combining both safety and efficiency. The first layer is the individual vehicles themselves. Google is doing a phenomenal job addressing this layer. The second layer is what I will call the network layer — the fleets of unmanned vehicles in aggregate. If we plan and execute the network layer correctly and in an adaptive fashion, we will actually need fewer vehicles (human-driven or driverless) to meet the demands of our day-to-day lives, whether that entails deliveries or public transportation.

And with fewer vehicles on the road, there is less of a safety risk at the individual vehicle level and the added benefit of less traffic, congestion, and waste. The ripple effects of this can, of course, be extended to a price reduction for deliveries and public transportation, etc.

This network needs to be structured yet rapidly adoptable and adaptable to meet the needs of the public, from both transportation and safety aspects. It is one thing for a single vehicle to sense a jaywalker and apply the brakes. It is another for the network to adapt itself in real-time to the butterfly effect of that car braking: the other cars braking; the human-driven car tailgating and not braking in time and colliding into the autonomous vehicle in front of it; the traffic jam created due to this incident; the ensuing delays; the delivery company needing to adjust its plans to meet its same-day delivery commitments; and the public transport authority needing to get the commuters to work on time.

As we move steadily and surely into an on-demand world, one which blends human and artifcially intelligent resources, we must think beyond a single resource, a single delivery or bus route. Using already available big data to streamline operations in real time is a highly effective way to manage a constantly changing environment, all while using the vehicles to the best of their capabilities. Optimizing real-time data streamed in from the vehicles can make responding and adjusting their usage in the real world possible, and it can give the public and regulators some of the comfort needed as we explore the possibilities.

Click to see story online

‘Send in the Drones’ featuring BoldIQ – by Zach Noble in FCW

Drones, UAVs, UAS — call them what you will, remotely piloted aircraft are poised to make huge inroads in the national airspace. And although publ0ic perception might link unmanned aircraft systems with intelligence agencies and the military, the federal government’s UAS user base extends well beyond spies and soldiers.

Agencies as diverse as NASA, Customs and Border Protection, and the Forest Service are all experimenting with UAS and deploying the systems in novel ways. Drones “can reach hard-to-fl y areas and maneuver well at low altitudes,”said Jeff Sloan, a UAS operator at the U.S. Geological Survey. “They give us data there’s no way you could get with a manned aircraft.” NASA is sending drones through hurricanes and volcanic plumes to collect data, while USGS is using the technology to map changing landscapes. The Border Patrol is scanning for lawbreakers from above, and the Forest Service hopes to better monitor the spread of wildfi res. Drones might soon be able to effectively deliver critical supplies in disaster and searchand- rescue situations. In short, the technology can save money, provide superior data and keep people out of harm’s way. Nevertheless, civilian agencies’ adoption of UAS is not a straight path forward, and regulatory hurdles and institutional caution are slowing the technology’s adoption.

Regulatory restrictions: One major impediment to faster drone adoption is the Federal Aviation Administration. Charged with regulating the nation’s airspace, the agency is naturally reticent to open the drone fl oodgates. And the FAA’s reach extends further than many think. In a recent myth-busting release, the FAA reaffi rmed that its jurisdiction starts at ground level, not at 400 feet as commonly stated, and NASA scientists confi rmed that FAA rules follow the space agency to parts of the Atlantic Ocean and the Arctic. In 2012, Congress tasked the FAA with developing a plan for safely integrating UAS into the national airspace by Sept. 30, 2015, and until that plan is completed, civilian agencies must obtain special FAA clearances to use drones. Public operators, including civilian agencies and numerous universities, held 613 active clearances, called certifi cates of authorization, as of April 8. Besides the trouble of obtaining authorization from the FAA, agencies must also follow rules that, although well-intentioned, can neutralize the benefi ts of UAS. For many smaller drone models, like those deployed by USGS, FAA rules require operators to maintain line-of-sight contact with the vehicles, which Sloan said limits the drones’ utility. The requirement also keeps the Forest Service from sending drones into the smoke of wildfi res, thereby undermining one of the biggest potential benefi ts of UAS: gaining a vantage point unattainable by human pilots.

Plunging in unprepared: The FAA is not the only source of UAS hiccups; other agencies have made some mistakes along the way. In 2007, for example, the Forest Service spent $100,000 on a pair of SkySeer drones that it planned to use to spot illegal marijuana-growing operations on federal land. Unfortunately, the agency lacked trained operators and FAA approval. Jeff Ruch, executive director of Public Employees for Environmental Responsibility, said the Forest Service’s poorly planned purchase is evidence of a “boys with toys” attitude toward emerging technology. “There was no planning,” said Ruch, whose organization publicized the issue. “They saw the Border Patrol’s use of drones and said, ‘Oh, that’s neat.’” The Forest Service’s drones have now been slated for wildfi re tracking, he added, but “it’s not clear if that transition will take.”

Drones on a budget: Regulatory restrictions have all but forced civilian agencies to be followers in the realm of UAS development, but there’s a substantial benefi t to letting the military and private industry take the lead: Interested agencies can pick up drones for free. The Interior Department’s USGS owns a fl eet, valued at $15 million, of 20 T-Hawks (20-pound drones made by Honeywell) and 15 tiny hand-launched, remote-control Ravens made by AeroVironment. Although USGS has spent around $1 million on UAS operator training and sensor systems, it paid nothing for the drones themselves. “Our Ravens are from 2005,” said Mike Hutt, UAS project manager at USGS. “The military has moved three generations past those initial Raven models, so they’re surplussing the old ones to us.” That military/civilian cooperation has been a boon to USGS. The free Ravens “really helped us cut our teeth on what we can and can’t do with drones,” he added. NASA is another agency that is beating swords into plowshares. The agency’s Airborne Science Program has been dabbling in UAS since the early 1990s and currently uses such varied drones as the 25,000-pound Global Hawk, the customized- for-science Predator variant Ikhana and the small, maneuverable Dragon Eye. NASA’s fleet of Dragon Eyes was acquired for free from the Marine Corps. “We take whatever we can get,” said Bruce Tagg, manager of the Airborne Science Program. “Our scientists are very entrepreneurial; they have their eyes on just about everything.” NASA obtained the Dragon Eyes through the Rube Goldberg process that is interagency procurement: A NASA scientist heard the Marine Corps was getting rid of the drones and asked about having them sent to NASA. The drones went fi rst to the Interior Department, then to the General Services Administration and, fi nally, as a result of the scientist’s persistence, to NASA.

Weighing costs and benefits: Although UAS can bring many benefi ts, saving money is not always one of them. “There’s a misconception that these UAVs are so much cheaper than manned aircraft,” said Matt Fladeland, NASA’s UAS manager. “For [small drones such as] Dragon Eyes, that might be true, but for bigger systems like the Global Hawks, there’s not much difference [in cost] between running them and running a twin-engine B200.” Between the costs of fuel, trained operators and support systems, fl ying a large UAS can be just as expensive as a manned fl ight. Tagg said the real benefi t of large drones is not that they save money but that the unmanned craft can stay aloft for 24 hours in situations where a manned aircraft would last half as long. When monitoring a developing hurricane, for example, the extra airtime can be hugely benefi cial, he added. Small drones bring more direct savings. “In smaller areas — 10 kilometers by 10 kilometers — UAS are very good for surveying and bring us a substantial cost savings,” said Hutt, who estimated a 10-to-1 savings over traditional manned fl ights. Drones also enable agencies to save in other ways. For instance, instead of relying on satellite imagery, USGS can get better photos for less money by strapping a GoPro camera to a low-fl ying drone. USGS uses data-processing software to make sense of the images collected by drones and gain a sophisticated sense of topography, vegetation cover and more. “We’re fi nding that $1,000 cameras are giving us data that we used to rely on $400,000 mapping tools to get,” Hutt said.

The path forward: The peaceful potential of UAS seems indisputable. “UAS will assist public safety agencies in responding to natural disasters, locating missing persons or helping to fi ght wildfi res,” said Melanie Hinton, senior communications manager at the Association for Unmanned Vehicle Systems International. “In addition, UAS will help farmers care for their crops, [help] to identify diseases, and more precisely and safely spray pesticides.” The organization estimates that the fi rst decade of widespread UAS adoption could produce an $82 billion economic boost in the United States. The military likes to say drones are used for dull, dangerous and dirty missions, Hutt said, “but we’re focused on better Earth science applications, greater safety and savings.” He said he expects FAA regulations, especially the requirement that operators maintain line-of-sight contact with drones, will ease as better radar systems and transponders are developed to keep drones out of the way of other aircraft and one another. Industry insiders are developing new tools, but they say the existing technology is exciting in its own right. “Lost-link procedures are pretty standard now, and fl ight planning is getting better,” said Hutt, adding that although they are not fully autonomous, many drones have sophisticated programming to handle emergency landings and extended fl ights on their own.

Roei Ganzarski, CEO of software developer BoldIQ, is particularly bullish on drones. “The civilian market, once it’s opened up, will be a lot bigger than the military market,” he said. Real-time optimization of data is BoldIQ’s stock-in-trade, and Ganzarski said software can make sense of drone data nearly instantaneously. Noting that concerns about the prevalence of drones are similar to public fears surrounding the advent of commercial aviation, he added, “There’s a view that [drones] will be flying around like mosquitoes, en masse, crashing into each other.” But with modern programming, drone fleets can be integrated with one another and the surrounding environment, and dynamic optimization will allow drones to react quickly and competently to changes in the environment, Ganzarski said. “The tech barriers [to UAS integration] don’t exist,” he said. “The barrier is the fear of the unknown.”

Click to see story online

BoldIQ and Drones In Action: 5 Non-Military Uses – by Elena Malykhina in Information Week

The Federal Aviation Administration (FAA) estimates that as many as 7,500 commercial drones — ranging in size from the large wingspan of a Boeing 737 to a small radio-controlled model airplane — will be hovering in the US airspace by 2018. Beyond the military, there are numerous potential uses for drones, or unmanned aircraft systems (UAS), such as law enforcement, storm tracking, search and rescue, and aerial surveying. But managing drones domestically comes with its own challenges, which still need to be addressed by the US government and the private companies involved.

The FAA in December set up six sites to test drone operations around the country. The congressionally mandated sites are tasked with conducting research into the certification and operational requirements for safely integrating commercial drones into the national airspace. The six sites include the University of Alaska, the state of Nevada, the North Dakota Department of Commerce, Texas A&M University/Corpus Christi, Virginia Tech, and Griffiss International Airport in Rome, New York.

The FAA’s move to set up drone test locations follows the release of a roadmap in November, addressing current and future policies, regulations, and procedures that will be required as drones continue to become more mainstream. “We have made great progress in accommodating public UAS operations, but challenges remain for the safe long-term integration of both public and civil UAS in the national airspace system,” FAA administrator Michael Huerta said in the document’s introductory letter.

Safety tops the list, especially when it comes to the logistical challenges of managing drones. “Buildings, antennas, manned airplanes, and other drones can make it a chaotic place, and safety needs to be the number-one focus of those managing drone implementation,” said Roei Ganzarski, CEO at BoldIQ, in an interview with InformationWeek Government. BoldIQ, a provider of optimization software, recently completed analysis of Silent Guardian, a solar-electric drone to highlight the benefits of using hybrid technology.

Companies managing drones need to consider logistical planning involving individual drone operations, coordinated drone fleet management, and incorporating drones into a “manned airspace,” all while processing enormous amounts of real-time data, according to Ganzarski. “When assessing a fleet of drones operating autonomously or even semi-autonomously, it becomes impossible for the human brain to process and manage the data to keep the entire system operating smoothly. It requires sophisticated real-time dynamic optimization software,” he said.

Beyond logistics, another issue is the security of the drones themselves, and the cargo they may be carrying. It’s vital that systems are in place to protect these expensive technologies while in flight and on the ground. Privacy is also a major concern for the public. Organizations need to make sure that UAS equipped with cameras do not violate privacy laws, said Ganzarski.

At the moment, almost all commercial drones are banned by the FAA. But that should change in 2015, when the agency expects to release its guidelines for safely operating drones. In the meantime, government agencies, a number of universities, and a handful of private companies are putting robotic aircraft to good use — and in some cases challenging the FAA’s authority.

A judge agreed March 6 the FAA had overreached fining businessman Raphael Pirker, who used a model aircraft to take aerial videos for an advertisement. The judge said the FAA lacked authority to apply regulations for aircraft to model aircraft. That may open the skies to a lot more privately controlled drones.

Click to see story online