Cognitive computing and IBM Watson IoT: unlocking the power of information
Businesses, academic institutions, cities, and other enterprises may have diverse objectives and missions but all have two things in common: Data, lots of it, and the need for a comprehensive—and comprehensible—way to make use of it. Most executives leading these institutions recognize the value of mining this data for the insights it can provide, and many also realize that the data onslaught from sensor-based devices within the Internet of Things will exponentially increase the amount of that data. But the sheer enormity of collecting and assembling this information, along with analyzing and producing meaningful results from it are simply too daunting a task to consider. So most of it sits on servers, workstations, and other computers, and in paper documents of various formats, unused and losing value every minute.
However, cognitive computing, exemplified by IBM® Watson, makes it possible to learn from these data sources to reveal insights that, practically speaking, have never been accessible before. The insights it can provide allow any organization to better support its constituents and clients, operate more efficiently, develop new initiatives and rejuvenate existing ones, and provide products and services more effectively to more people. Watson can ingest and analyze any kind of data, in virtually any format, from any type of source, and correlate it with other sources, such as comprehensive weather data to uncover key insights that can have a big impact on business decisions. Watson’s abilities also continue to grow as it becomes more “knowledgeable” over time. To see how this is possible, it’s important to understand what cognitive computing is.
Cognitive solutions can bring intelligence into things, systems, and processes so they can understand your goals, and then integrate and analyze the relevant data to help you achieve them. This is considerably different than what conventional computers do, as they act on only those instructions they have been programmed to execute without the ability to add much insight into what they have computed. Cognitive solutions build on this core processing functionality with the ability to learn and reason from their interaction with more or less anything that produces information—including people--and provide the information required to make decisions, also by people. So contrary to the media and the film industry, cognitive computing provides people with the information they need to be more effective, rather than making them obsolete.
Too Much Data, So Little Time
There is far too much information being generated today throughout the world to even consider attempting to assimilate or even collect all of it manually. However, the more information available with which to make a decision the better. Proof of this is illustrated by a program underway at Baylor College of Medicine in Houston TX, which shows how Watson can accelerate medical research by making more information available than even a team of people could assemble and digest.
Image courtesy of IBM
Biologists and data scientists at Baylor are using Watson within the institution’s Knowledge Integration Toolkit (KnIT) to unlock patterns and make discoveries with greater precision. For example, gene p53 codes for a protein that regulates the cell cycle and effectively functions as a tumor suppressor. The researchers wanted to determine proteins that modify it in order to make drugs and other treatments more effective. After training Watson to “think” like a human researcher, Watson analyzed 70,000 scientific articles on p53 to determine which proteins turn its activity on or off. This would have taken years to accomplish manually but with Watson’s cognitive capabilities the test was completed in a few weeks.
Another similar example is medical diagnostics, in which radiologists and other diagnosticians base their recommendations on their own experience and available data. Watson can exponentially increase the amount of the data upon which these decisions can be based as it can combine local results with diagnoses taken from thousands or tens of thousands of scientific publications and other sources. This can potentially present alternatives that would otherwise be difficult if not impossible to locate.
Even more data from even more places
Few applications are in greater need of what Watson can provide than those under the broad umbrella of Internet of Things (IoT). In the coming years, data from an extraordinary number of sources will communicate huge amounts of information to and from their hosts every day. It can be in data from sensors, voice from customer calls, video and still images from cameras, written maintenance and service reports from field technicians, and many others.
There are already billions of connected devices in place and this is expected to grow to many tens of billions globally in the coming years, making IoT devices the greatest data generators on the planet by a significant margin. Sensors will be everywhere, from walls to walkways to doors, windows, elevators, pipes, lights, traffic signals, vehicles, industrial equipment -- virtually everywhere something needs to be monitored. Even a single source can generate truly astonishing amounts of information:
- Pratt & Whitney’s Geared Turbo Fan (GTF) engine has an astonishing 5,000 sensors that generate up to 10 GBytes of data per second, so a twin-engine aircraft with a flight time of 12 hours can produce up to 844 TBytes of data. This dwarfs even Facebook, which is estimated to accumulate more than 600 TBytes per day.
- A Formula 1 race car typically has more than 200 sensors communicating over 1000 channels with up to 40 people continuously monitoring the resulting data during a race. On a typical race weekend, between 120 and 150 Gbytes of data has been generated by a single race car.
It doesn’t take much imagination to visualize how much data is currently being generated worldwide and how much more will be produced in the future. Although until recently most of this data was either collected manually or perhaps not at all, new wireless standards and other emerging communications networks dedicated to IoT are allowing these islands of information to be aggregated to form ecosystems whose benefits are just now being realized.
That said, while gathering this information is an astonishing feat, it serves little purpose unless there is a way for this information to be used, a task for which Watson is uniquely suited. Watson can do this whether the information is about a fleet of vehicles, a city’s environmental sensors and connected meters, a globally-installed base of manufacturing equipment, or countless others. The immense amount of information that IoT devices create and communicate obviously makes security essential, especially as even though the connected world of “smart” devices is relatively new, breaches have already occurred. IBM helps developers build security into IoT applications from the earliest design stages, from securing physical devices or gateways to network and transport security, IBM Watson IoT, and every other element of the IoT environment.
In the industrial world, the IBM Watson IoT Platform can continually monitor incoming information in real time and based on what it “knows” and is continually learning, to understand current conditions, what's normal, what’s not, determine trends, and suggest actions. For example, using IBM’s rules-based IoT Real-Time Insight analytics tools, Watson can combine real-time information with stored master records about each system from IBM Maximo Asset Management software including make and model, installation date, and historical information about repairs and maintenance. All of this can be complemented by other data such as weather conditions, which are a major factor affecting the performance and reliability of equipment operated outdoors.
Image courtesy of IBM
Together they provide a wealth of material from which Watson can make decisions based on rules and evidence-based data. A rule can be configured to create a work order using Maximo Asset Management’s work order engine, which can assign a person with the right skills, tools, and spare parts to efficiently complete the job. Maximo can then escalate the order if it is not completed satisfactorily. The result is a closed-loop response to an operational issue that is traceable from the sensor data to the end result.
In surveillance and monitoring applications, Watson’s ability to analyze video and still images makes it possible to identify scenes and patterns that combined with other data can provide a greater understanding of past events and emerging situations. So if security cameras show that a forklift or other type of equipment has entering a restricted area, a low-level alert will be noted. If several days later the forklift’s performance begins to decline, the two incidents can be correlated to determine what the cause might be, such as a collision between the forklift and some other piece of equipment. This would not be readily apparent from the raw video or data from the forklift alone.
While IoT data is most frequently associated with some type of equipment, it also includes text from such diverse sources as transcripts from customer calls, technician maintenance logs, blog comments, and tweets. Watson can handle these too, correlating information and detecting patterns based on phrases such as "my brakes squeal", "my car seems slow to stop," and "the brake pedal feels mushy." When the phrases are analyzed, Watson can identify potential issues with the brakes of a specific make, model, and year of a vehicle.
Watson’s natural language processing capability also allows users to interact with systems and devices using simple spoken language, which when added to text from other sources can be placed into context. In this case, when a technician working on a machine notices an unusual vibration he can ask "What is causing that vibration?" Watson then will automatically link words to meaning and intent, determine what machine the technician is referring to, and correlate recent maintenance records to identify the most likely source of the vibration and then recommend an action to eliminate it.
Watson IoT in action
Dubai International Airport in the United Arab Emirates is one of the fastest-growing airports in the world and as of 2015, was the world busiest for international passenger traffic, the third busiest in total passenger traffic, and the six busiest cargo airport in the world. Traffic has grown from 15 million passengers in 2002 to 78 million in 2015.This has obviously required significant changes to its infrastructure, maintenance processes, and many other operations, with an extremely high level of automation.
To accomplish this feat, Dubai Airports Company is using IBM WebSphere, Maximo Asset Management, and Maximo Mobile Work Manager to standardize its processes, minimize resource requirements, and generate job plans tagged with task, duration, and resource benchmarks for every activity. The company has developed a comprehensive database that matches the type of labor required to skills possessed by employees and contractors. Work assignments are now automated and employees can remotely access asset management and maintenance scheduling processes. The system has already reduced maintenance costs, increased equipment uptime, and enhanced other metrics. And the company believes it will save about $100 million by 2020.
Forecasting the weather arguably generates at least as much if not more data from more sensors than any application, and the information it provides is critical for efficient operation of many industries. Few applications aren’t impacted in some way every year by the weather, from agriculture to land and sea shipping, aviation, and hundreds more. So it’s not surprising that IBM last year acquired The Weather Company’s business-to-business, mobile, and cloud-based Web properties, including WSI, weather.com, and Weather Underground.
The Weather Company’s models analyze data from 3 billion weather forecast reference points, more than 40 million smartphones, and 50,000 airplane flights per day. This provided a comprehensive foundation for IBM’s new Watson IoT unit and combines IBM’s cognitive and Watson analytics capabilities with The Weather Company’s cloud-based data platform (which was recently moved to the IBM Cloud). This combination allows IBM to collect massive amounts of global weather information that can be stored, analyzed, and made available throughout the Watson platform.
Hopefully, by this point it’s obvious that the broad resources available within IBM Watson can deliver huge benefits for any organization, and especially for those in which IoT either is or will become a major component. So an equally obvious question is how to begin exploiting what this platform can offer, which might logically seem difficult.
Fortunately, it’s not, as IBM provides step-by-step instructions, videos, documents, and many other resources to streamline the process. Readers can register for a free trial at www.ibm.com/iot and a unique identifier is provided that is accessible only by the registrant’s devices and applications. After this, all of the required tools are immediately available for getting started. It’s that simple. In addition, Avnet provides videos, documentation, and other resources to help anyone get started with Watson IoT and produce results quickly.
In addition, Avnet provides comprehensive support for IBM Watson IoT Platform including the Avnet MicroZed Industrial IoT Start Kit supported by IBM, Wind River, and Xilinx, that simplifies prototype and development efforts and streamlines the transition to production. It includes all the necessary building blocks for developing a production-ready, IoT-enabled, industrial processing system based on Avnet’s MicroZed system-on-module (SoM) with a Zynq-7000 All Programmable SoC from Xilinx and pluggable sensor solutions from Maxim Integrated and STMicroelectronics.
The kit integrates the IBM Watson IoT Platform agent on top of a custom-configured, certified image of the Wind River Pulsar Linux operating system, now standard on MicroZed and PicoZed system-on-modules. Using a standard MQTT messaging protocol, IBM’s Watson IoT Platform agent enables registered, secure connection to the Watson IoT Platform and additional IBM cloud services and applications.
Avnet IoT University: Getting Started with the IBM Watson IoT Platform
Written By: Kevin Larsen
Today's case for AI at the edge
AI might feel far away, but engineers are innovating everyday. Should you consider edge AI? ...
Specialists vs. end-to-end partners: the pros and cons for developers
Getting a product to market requires more than just the right idea; it also needs the right partner....
Making sense out of sensors
If creators still don’t think that sensors are getting the attention they deserve now, then there re...