AI can help IoT-connected systems better analyze data, make decisions and change behaviors. Using AI is how IoT can finally deliver on its promise of a sky-high ROI.
The holy grail for IoT has always been better decision-making and data-driven actions. In this quest, IoT has evolved from pure remote control to a cutting-edge technology that incorporates machine learning.
IoT has always been about getting data, and then acting on the data. Once confused with remote control use, which includes both local and internet-enabled control from anywhere, IoT has now started to use machine learning to react to inputs -- inputs that are based on anticipating previously observed effects from connected devices. The result of this is increased convenience that has driven energy savings, as well as increased safety, thanks to the ability to automate actions that previously may not have been done in a timely manner.
An example of increased convenience and safety can be seen in IoT-connected HVAC systems. Machine learning can automatically change thermostat settings based on the utility price tables and the fluctuating cost of energy, thus ensuring maximum cost savings. At the same time, at an enterprise level, machine learning can aid in managing demand response in a timelier manner than phone calls and wall switches. Yes, manual intervention is still a demand response thing. This means an enterprise can more intelligently match the demand for power with the available supply.
As the base levels of IoT implementation improve beyond infrastructure and platforms to vertical integrations, we are starting to see system integrations being fielded rather than individual products. For instance, we are seeing demand response services instead of automated thermostats. Further, with sensors on distribution lines, it is possible to gain visibility into grid power distribution and control large systems with automated or manual data-driven control.
How IoT can change human behavior and efficiency
While IoT is wonderful as an automaton, it's real promise is when it can change human behavior by anticipating our actions and reducing wasted efforts. For this upper echelon of ability, we need increased vertical systems integrations, and a migration of these capabilities closer to the edge where the activity is taking place. If done properly, this can lead to more effective use of human effort, as well as greater efficiency and throughput from machines, ultimately resulting in systems that better serve us.
One example of integration is from a startup called Future Acres. This company uses AI, autonomous driving, safety, environmental and status sensors to improve crop delivery for pickers, removing an otherwise wasteful, tiring and nonproductive part of a critical job. The company's AI-powered farm equipment automates the toughest tasks to enable higher yield, profitability and growth.
What to do with all that data
IoT by its nature can create a tremendous amount of data. Sending this data to the cloud allows us to clean it, format it and use it to create models for higher value uses back at the point of origin. With initial data from a particular use case, machine learning can be used at the edge to send only the data of interest to the back end and reduce bandwidth needs.
However, by doing this too early, you risk failing to store data that possesses hidden and highly valuable information. Once we have created models in the cloud, use of this intelligence via a digital twin may not have the low latency required for true real-time systems.
That's why we are now seeing the rise of infrastructure, such as fog networks, and pressure for edge-based AI processing. This will drive the need for a hierarchy of computation, as well as a need for a hierarchical infrastructure for IoT devices.
About the author
Mark Wright is technology evangelist at GSI Technology Inc. He comes from a career in electronics systems spanning high-reliability designs, and new technology applications. He began his career designing board-level product for Computing Devices Corporation, now GD Canada. He has driven product management for new technologies such as ternary content-addressable memory and IoT at companies such as Integrated DNA Technologies, Ayla Networks and Samsung Electronics. Prior to joining GSI Technology, Wright consulted to early stage startups on technology and business development issues.