One of the IoT's greatest attributes is its ability to integrate and work with other emerging technologies, as I wrote recently regarding the synergies between the IoT and blockchain.
Perhaps nowhere are these synergies as great as with the IoT and Artificial Intelligence (AI). One very down-to-earth (OK, down-to-living-room-carpet) example will illustrate.
My Roomba was one of the first smart-home devices. It’s dogged, and eventually cleans a room by bouncing around and following a pretty random path. While it has some “intelligence,” I always suspect some areas may get missed.
Fast-forward to iRobot’s newest versions, the 900 series. They added a whole range of AI functions including “Vision Simultaneous Localization and Mapping,” plus better sensors to gather more real-time data. The resulting machine learning leads to new customer services, including Amazon AlexaTM commands (“Alexa, ask Roomba to begin cleaning,”) and producing a “Clean MapTM” of an entire floor that even shows (on the app that controls the Roomba) information such as where the Roomba encountered more dirt.
The new Roomba beautifully illustrates the closed-loop relationship between the IoT’s ability to gather data and AI’s ability to process it. As perhaps the most comprehensive report on the combination, PWC’s Leveraging the Upcoming Disruptions From AI and the IoT, emphasized, each provides a vital component for the other: AI makes the vast quantities of data harvested from IoT devices valuable, while the IoT is the best source of the real-time data AI needs to digest to learn and progress:
“Data is only useful if it is actionable. And to make data actionable, it needs to be supplemented with context and creativity. It is about ‘connected intelligence’—which is where AI and smart machines come into the equation. AI impacts IoT solutions in two key dimensions—firstly in enabling real-time responses, for example via a remote video camera reading license plates or analysing faces; and secondly in post-event processing, such as seeking out patterns in data over time and running predictive analytics. Through these capabilities, AI supports enhanced IoT applications by enabling:
“The interdependence between IoT and AI also works the other way. IoT’s capacity to enable real-time feedback is critical to adaptive learning systems, since other technologies do not really enable this advanced type of AI/analytics. So they both need each other.
PWC predicts that the combination will lead to truly smart machines: “The ongoing advance of AI is also having a further impact: it’s causing AI to converge with IoT, to the extent that it’s rapidly becoming indispensable to IoT solutions. The core components of IoT—connectivity, sensor data and robotics—will ultimately lead to a requirement for almost all ‘dumb’ devices to become intelligent. In other words, the IoT needs smart machines. Hence the need for AI.”
The smart machines will allow the kind of unprecedented precision based on M2M communication and self-regulation that I’ve written about previously — just think of how my Roomba bumps around vs. the systematic path of the new ones.
An example from the energy field illustrates how important that precision is. The Nest thermostat has advanced AI capabilities (it learns from occupants’ movements and self-regulates accordingly), and, if the owner opts in, provides anonymized data to the utility on use patterns that can be crucial in planning load management. Nest has worked with Southern California Edison for several years on its "Rush Hour Rewards” program, with a goal of getting 50,000 customers to agree that the thermostats would automatically be adjusted remotely to cut the homes’ consumption levels during peak demand. That could save the utility about 50 megawatts of capacity and help avoid brownouts.
Just as companies that are hanging back on the IoT are still paving the way for adopting it by acquiring sophisticated data analysis tools and hiring data scientists to interpret the growing amount of data they are already generating, investing now in cloud-based AI capability will allow an easier transition to the IoT and the even greater quantities of data it will generate.
More advanced sensors will also play a role in the IoT-AI synthesis: I recently wrote about exciting DARPA research on sensors for life-or-death situational awareness in battlefield conditions, where computing power is limited and soldiers can be killed if they have to replace sensor batteries. These new sensors — which must work 24/7 to protect the troops — have: “persistent, event-driven sensing capabilities in which the sensor can remain dormant, with near-zero power consumption, until awakened by an external trigger or stimulus. Examples of relevant stimuli are acoustic signatures of particular vehicle types or radio signatures of specific communications protocols.” That means they do their jobs, but also conserve critical battery power. I suspect the battlefield sensors will quickly be commercialized, reducing both sensors’ power needs and the need for the AI system to process a lot of “background noise” data that has little or no value (the data processing, BTW, will increasingly be done “at the edge,” where it’s collected, to avoid overloading the system). Both IoT and AI will benefit.
There’s a management aspect to this progress as well. According to Future Text’s Ajit Jaokar, interpreting all this new data may require a new type of engineer to deal with it, “who will combine learnings from Electronics (IoT) with Machine learning, AI, Robotics, Cloud and Data management.”
Bottom line: you couldn’t ask for a better combination of emerging technologies: the IoT will yield unprecedented amounts of real-time data, while Artificial Intelligence will help the IoT devices learn from the data and really become smart. Let’s go!
Image courtesy of iRobot.