OT (Operational Technology) systems consist of hardware and software that directly monitors and controls physical devices, processes and events in a wide variety of commercial and home applications. The OT industry is a prime candidate for disruption by artificial intelligence (AI) and machine language (ML) technology. OT systems require AI/ML technology to tackle multi-dimensional problems while meeting high-speed, continuous real-time requirements—in contrast to the batch-oriented “statistical” AI systems being developed for IT systems. At the same time, OT systems have much smaller power budgets compared to the multi-megawatt power systems in Cloud-based data centers. These diverse requirements will mean that AI/ML solutions for OT systems will differ markedly from AI/ML solutions already developed for Cloud-based AI, and those solutions will also vary by OT market sub-segment.
Table 1 lists the fastest growing system market sub-segments in the Consumer, Infrastructure and Smart Factory industries, as estimated by IDC. The table also highlights the CAGR revenue forecasts of sub-segment markets from IDC's Embedded Market Model.
|Diagnostics & Monitoring||25%||PHV/EV Industrial Infrastructure||37%||Time-Sensitive Networking||116%|
|Drones||23%||Digital Signage||25%||Industrial Wearable||59%|
|Healthcare Gateway||17%||Remote Tracking||18%||Industrial Gateway||13%|
|Wearable||14%||Commercial Motor Vehicle Telematics||13%||Industrial Automation||6%|
|Smart Home||5%||Video Surveillance||10%||Functional Safety||6%|
Source: IDC Embedded Market Model, 2018
Many opportunities exist for the addition of AI/ML technology into each of these OT market sub-segments. Here are some examples:
The infinitely complex and unstructured operational environment that automobiles experience requires that the vehicle sense its environment and respond accordingly. AI/ML is already being employed to help make these decisions in upcoming vehicles coming to market. Already, deep-learning algorithms are required to allow a vehicle to operate autonomously, and new embedded automotive AI applications for the automobile are emerging for applications as diverse as preventative maintenance, software security and optimized engine operation.
Industrial Automation / Smart Factory
One of the earliest applications of embedded AI in the factory is for enterprise-grade cameras used for defect monitoring. Computer vision and the necessary ML and inference software are already used to improve quality control accuracy and to recognize more serious operational issues. Moving intelligence to the endpoint enables real-time defect recognition while reducing cost and energy consumption.
Another important emerging application for embedded AI in the factory is predictive analytics. Predictive analytics utilizes data from vibration, audio and other sensors in order to enable companies to detect the earliest possible signs of potential equipment failure. Predictive analytics can easily save millions of dollars per failure avoided. The same algorithms can be used to make operational improvements that reduce cost and improve efficiency over time in a factory.
Service Robotics / Commercial Robotics
Commercial robots are already used in many industries. These robots will increasingly leverage embedded AI. For example, warehouse robots that move inventory around for storage or delivery can make use of embedded AI to help them navigate faster, coordinate better with other robots moving around them, and find more efficient paths to get work done more quickly. IDC predicts that this shift will be especially true as the commercial robotics market shifts from robot product sales to a service model, sometimes referred to "robotics-as-a-service."
The need to reduce cost, to lower the carbon footprint of buildings by optimizing the use of energy, and to address occupant lighting and comfort requirements is driving investment in building automation and the redesign of commercial and retail buildings. Today, building automation systems (BAS) increasingly use cloud-based facility management. These management algorithms rely on a growing volume of sensor data to manage and control the building environment.
In the future, BAS will continue to introduce more intelligence and add features to lighting systems, ventilation systems, electrical systems to improve occupant comfort, increase automation, unify resources, optimize services, and reduce costs across a site or multiple sites without human interaction. As building automation continues to evolve, IDC expects to see more demand for flexible architectures that can be optimized for real-time workloads in order to improve efficiency.
The intersection of consumer appliances and IoT offers the greatest growth potential in the smart home. Vision and voice processing are the major technology growth drivers here. The diverse home applications for this technology include home security cameras, interactive smart speakers with voice recognition, and immersive gaming using VR/AR. In smart home applications, home IP cameras will be used to recognize and authorize entry access for family members, and to notify homeowners when deliveries are made. Home automation (similar to building automation) continues to be an emerging opportunity but IDC predicts that this technology will be adopted more quickly in the home given the appeal and convenience it offers to consumers.
The number of older adults will continue to grow exponentially and will increasingly drive healthcare services directly into our homes. AI will be the only way to care for all of these people, especially with the unsustainable growth in healthcare costs. Wearable applications will augment the growing demand for remote healthcare and assisted living. Low-power computing intelligence, sensors and AI will create new markets for infusing intelligence and augmenting healthcare devices of all types including bringing devices closer to our bodies.
As a result, health services will be able to provide a higher quality level and greater flexibility to each patient and will allow doctors to improve their level of personal care. AI will also help find patterns in patient data, automate record process, digitize patient records, and enable higher resolution images for early detection and tracking of chronic states of patients.
While most of today's smart meters simply report energy usage, the addition of embedded AI will enable smarter meters to forecast electrical loads for utilities. Incorporating weather information with microgrid generation and smart meters will help utilities anticipate spikes or dips in power and enable utilities to better stabilize the larger power grid. The shift from basic smart meters to embedded AI smart meters will allow utilities and countries to shift from large energy plants to more diverse and environmentally friendly energy sources.