Coupled with the rise of this next generation of computing has been the expansion of cyber-physical systems. With the Internet of Things, devices, objects, and all type of technologies are becoming instrumented and networked to algorithms running in the cloud. With this next generation, computers are starting to come out of the world of well-structured data into the everyday world of unstructured environments, where they use large amounts of data to create the context within which they can interpret new things and interact with the world in a fluid fashion. This next generation of robots are safe for and can interact with humans in a more fluid fashion, what is called human-centered robotics. They have a sense of touch and are very precise. Amazon’s warehouses that may have up to ten thousand robots assisting people in bringing them the packages is one example. The same evolution in computing that has developed into the platform model is also coming to robots, where multi-purpose physical robotic capabilities can be delivered via a cloud platform that enables developers to bundle them into new processes and applications.
Again it can’t be over emphasized how important the rise of cloud computing is to this whole equation. Fei-Fei Li of Google Cloud presents it well when she notes1 “It took me a while but I start to realize that cloud is the biggest computing platform humanity has ever created, and what is computing today, computing is to make your data speak intelligently and to act intelligently to solve your problem or your customers problem, so really this marriage between AI and cloud is like this perfect vehicle to democratize AI.” The combination of smart systems, cloud platforms, and cyber-physical systems will revolutionize our technology landscape in the coming decades. With the rapid commodification of smart systems, connectivity to the cloud, and sensing devices more and more of our technologies will become cyber-physical, from shopping trolleys to shoes, cars, to whole houses. But also smart platforms will be plugged into whole infrastructure systems like the power grid, internet routing, city transport systems; taking in massive amounts of data, learning from it in order to optimize the system.
Much media attention and public imagination is currently focused on robotics and individual cyber-physical systems, although a little robots cleaning our house or delivering a pizza might be the most apparent manifestation of this change the real innovation will be delivering these machine learning algorithms as a service to IoT platforms that network whole infrastructure systems, whether that is the cloud analytics connected to the smart grid, transport networks, or connecting an enterprise’s whole supply chain up, or the smart city itself. Take for example the mining industry that is currently going through a massive wave of automation as mining companies are rolling out autonomous trucks, drills, and trains. From a control center in Perth, Rio Tinto employees operate autonomous mining equipment in Australia’s remote but mineral-rich Pilbara region.2 73 trucks – each the size of a small two-story house – find their way around using precision GPS and look out for obstacles using radar and laser sensors, and work alongside robotic rock drilling rigs. From their single operation center, they integrate information from all their mides, ports and rail systems and visualization technology gives their personnel a 3D display of all their operations. As the company says “these technologies take us ever closer to whole mine automation.”
Cloud analytics is a service model in which elements of the data analytics process are provided through a public or private cloud. Cloud analytics applications and services are typically offered under a subscription-based or utility (pay-per-use) pricing model. Google, Microsoft, and Amazon also expect to increase profitability and enhance their cloud services through machine learning. Their strategy is to allow other companies – which are unable to develop ML solutions at that level – to access their cloud-based ML services through APIs. One example of this is an app recently developed by Microsoft for the facial recognition of Uber drivers. Companies like Uber already have a machine learning platform – or at least a group responsible for building and scaling their machine learning technology – and increasingly these new technologies will be used to analyze the data coming from their car sharing platform and optimize where cars go, what route they take, how much they charge, essentially automate the most basic management activities of their platform.3
The platform model will be important in developing smart solutions in that it will enable different smart capabilities to be offered as modular utility functions that can then be plugged into and bundled together by enterprises according to their specific needs. Instead of having just one general purpose system, a platform model allows developers to draw upon specific capabilities and integrate them into their solutions, such as machine learning to recognize a face, or voice recognition software, or advanced analytics for specific domains. Equally the platform, plug and play model will work to commoditize smart systems making them available as a service to almost any technology developer. APIs and developer toolkits are already offered by IBM for their system Watson that can be plugged into a wide variety of applications from health diagnostics to analyzing data coming from transport systems. In such a way smart capabilities will flow to almost all types of technologies in the coming decades.
This cloud-based platform model to smart systems will mean that through an internet connection even the smallest of computers, like a mobile phone, can operate like the most powerful computers in the world, by simply sending the inputs to the cloud where it is processed and then output the information that is returned. This is quite an extraordinary phenomena in that it means the most powerful computer operations and algorithms can be accessed anywhere there is internet connection on the planet. This means that the most advanced technologies of our age can be accessed and used virtually anywhere on the planet through just a mobile phone and internet connection. Whereas previously we put computing devices into the hands of people, now we are putting supercomputers in their hands. The primary beneficial function of analytical systems will be in the management and optimization of large complex networks, they will be connected into whole transport networks, power grids, cities and possibly even whole urban networks analyzing that data to make predictions, optimizations and adaptations. API’s will make high-end machine learning capabilities available to all forms of devices and physical systems. As one commentator noted, “APIs are not a dime a dozen, they are a dime a million.”
One aspect of the platform model is that it can harness fleet learning. Because any component is operating within a network when one robot learns something then all can have access to that new information. That kind of network effect means that the system could improve in capacities at an exponential rate. The system will learn over time due to network effect and big data machine learning. Networked components can help each other create a wisdom of crowds effect. The CEO of Tesla explains fleet learning within their network.4 “The whole Tesla fleet operates as a network. When one car learns something, they all learn it” he goes on to explain how each driver using the autopilot system essentially becomes an “expert trainer for how the autopilot should work.” The company’s autopilot service is constantly learning and improving through machine learning algorithms.
Because all of Tesla’s cars have an always-connected wireless connection, data from driving and using autopilot is collected, sent to the cloud, and analyzed with software. For autopilot, Tesla takes the data from cars using the new automated steering or lane change system and uses it to train its algorithms. Tesla then takes these algorithms, tests them out and incorporates them into their upcoming software. In this way we can see how cloud platforms, machine learning and the internet of things can work in a synergistic way.
1. YouTube. (2018). Past, Present and Future of AI / Machine Learning (Google I/O ’17). [online] Available at: https://www.youtube.com/watch?v=0ueamFGdOpA [Accessed 13 Feb. 2018].
2. YouTube. (2018). Globalfuturist.org: Rio Tinto Mine of the Future – People and Technology working together. [online] Available at: https://www.youtube.com/watch?v=nWAQDa1XukU [Accessed 13 Feb. 2018].
3. Uber Engineering Blog. (2017). Meet Michelangelo: Uber’s Machine Learning Platform. [online] Available at: https://eng.uber.com/michelangelo/ [Accessed 13 Feb. 2018].
4. Fortune. (2018). How Tesla is ushering in the age of the learning car. [online] Available at: http://fortune.com/2015/10/16/how-tesla-autopilot-learns/ [Accessed 13 Feb. 2018].