Cloud Robotics: Distributed Robot Intelligence


Just a decade ago, the concept of robots communicating with a central computer “brain” was more science fiction than science fact. This vision was often portrayed in dystopian views of the future; for example, the sentient computers in the world of “The Matrix” and the implacable Borg in the “Star Trek” universe.

But researcher groups, including some at Google, are now looking seriously at the idea of robots that tap the processing power and data made available by cloud computing, seeing it as a way to create “lighter, cheaper, and smarter” robots.

The fruits of research into cloud robotics are just beginning to emerge from the laboratory, and some early examples of robotic systems that use the cloud approach seem more like toys than serious working robots. But these systems may in fact be the precursors of a new generation of robotic devices that will have broad commercial and consumer impacts.


  • Distributing the “brains” for robots to the cloud will reduce onboard processing requirements, and thus the cost of robotic systems. It should also increase functionality as general-purpose robots will easily upgrade skills.
  • This could expand robotics beyond today’s primary area of use — manufacturing — to people’s homes and offices, to the service sector, to schools, etc.
  • Smartphones and tablets may be a key tool for consumers to communicate with and control cloud-connected robots.


Robots have conventionally been limited by their onboard hardware and software, and are usually designed to complete clearly specified tasks in structured environments. Cloud robotics could help the field progress past these traditional boundaries.

Cloud robotics uses the Internet to augment the robot’s capabilities by off-loading computation and providing services on demand. It uses wireless communications to tap distributed processing power and data, in effect moving the robot’s “brain” to the cloud. Taking this approach can provide several benefits:

  • Reduced onboard processing and power. By tapping into computing power in the cloud, designers can sidestep a classic challenge: packing enough processing power — and the batteries to support it — onto the actual body of the robot. In other words, moving the robot’s “brains” to the cloud reduces the need for large, power-hungry onboard computers, allowing robots to be smaller, lighter, less energy-intensive, and less expensive. For example, robotic systems might hand off computing-intensive tasks like image processing and voice recognition to the cloud.
  • Reduced hardware complexity, costs. When designed to leverage the displays and onboard sensors of personal devices (e.g., iPad, Android phones) — as some early cloud-enabled robots do — cloud robots can reduce the cost of hardware. As one robotics industry executive noted, “The built-in sensors and touch screens found in smartphones and tablets are ideal for a simple, user-friendly interface. Imagine a manufacturing worker, not long from now, lowering a robotic arm simply by lowering his or her phone.”
  • Robots that “learn” on the fly. If the vision of cloud robotics is fulfilled, general-purpose robots could use cloud-based services to expand their capabilities far beyond what they are initially programmed to do. As they encountered new challenges, they would simply tap into a “library of skills or behaviors” — relieving them of the need for vast amounts of onboard intelligence or specialized programming. For example, a robot could access an existing cloud service like Google ‘Goggles’ to recognize objects rather than implementing object recognition on each robotic platform. Google sees cloud-connected robots tapping into other of its software-as-a-service features, including mapping, navigation, voice recognition, text-to-speech, and translation.


The successful application of cloud-based approaches to robotics could lead to a variety of potential outcomes.

  • An Internet of (robotic) things. The emergence of the Internet of things will see the proliferation of connected devices. Some estimate that there will be 50 billion devices connected to the Internet by 2020. If the vision for cloud robotics is fulfilled, many of these connected devices will be mobile cloud robots, benefiting from their connection to each other — and to other smart devices, from cameras and remote sensors to home appliances.
  • An app-store model for robots. Thanks to the app-store model, smartphones are versatile, multi-utility devices — and consumers increasingly rely on them to fulfill a wider variety of functions in their lives. While the average number of apps per smartphone did not change significantly from the end of 2012 (26.5) to the end of 2014 (26.7), the time people spent on their apps changed dramatically, increasing from 23 hours a month at the end of 2012 to 37.5 hours at the end of 2014. The emerging cloud-based approach to robotics could take personal service robots down a similar path and lead the kind of consumer enthusiasm and adoption currently seen in the smartphone market. In this future, service robots would no longer be characterized by specialized designs (e.g., Roomba cleaning robots, telepresence robots, robotic toys). Instead, standardized general-purpose robots would download new skills in real-time as users wanted them to do new things. While this would not likely have widespread impact until the 2020–2030 time frame, it is interesting to see, as noted above, that early groups are already pursuing the app store model as a way to extend robots’ capabilities.


  1. By providing a platform for development, and by reducing or eliminating the costs of packaging, marketing, and distribution, the cloud-based approach to robotics will lower the barriers to entry for developers and will encourage both creativity and competition among application developers — much as it has done in the smartphone space. A rush may ensue to develop and market specialized apps for home and commercial robotic systems.
  2. There will be opportunities for brands to take part in this rush for app development. For example, toy and game companies could extend beyond iPad apps to apps that turn general-purpose robots into playmates. Chemical or home cleaning companies could target competitors like Roomba through software and peripheral devices for generic home robots.
  3. Niche developers will find opportunities to focus on specific aspects of robot performance, rather than having to deal with a wide variety of technical challenges. A long tail of services and functionality should follow. Consider that there were more than 1.5 million apps in the Apple App Store alone in July 2015, and more are published daily.