Cloud Robotics — A 10,000 Foot View
Industrial robotics and traditional industrial automation differ in two important ways — reprogrammability and (increasingly) autonomy. Reprogrammability distinguished robots from single-function/bespoke machines in industrial automation for many years. But autonomous/semi-autonomous operations is the promise of modern-age robots and autonomous operations are tied to advances in Artificial Intelligence (AI)/Machine Learning (ML). AI/ML is making its way into our physical world, out from the trenches of the digital world where it was once incubated through the ever-advancing search engines, recommendation engines, etc. A lot of these AI/ML systems have been commoditized by the major cloud providers and they will play an increasingly important role in Cloud Robotics.
Cloud Robotics, as a term, is not well defined. It is however interesting to see what the market has to offer us today. The big cloud providers, like what they have done with other software verticals, want to seize the software opportunity in robotics. The robotics industry, in general, hasn’t caught up to modern software practices. So the cloud providers are trying to advocate and support the use of cloud-native tools and best practices for robotics software. The cloud providers also seem to be the best positioned to serve the cause, especially when you notice that most modern software stacks are based on open source software and the big cloud providers run the most well-oiled managed services around them.
The adoption of cloud-native software design would itself be a big part of the cloud robotics narrative and would not be unlike the modernization of software development in other industries. This shift in robotics software design will, in turn, pave the way for modern robots to interact with the cloud providers’ gamut of complementary cloud services, especially AI/ML services that will power robot autonomy. Again, the big cloud providers seem to be the best positioned to serve the need with heavy investments in multiple AI/ML verticals, offering first-class services critical for Robotics like Simultaneous Localization and Mapping (SLAM), Spatial Intelligence, Object Intelligence, etc. and services enhancing the overall capability of Robotics like Voice Intelligence, etc.
A cloud robotics stack includes various features like Simulation, Fleet Management, Telemetry, Teleoperation, Telepresence, Data Ingestion and Data Analytics. Except for robot simulation tools, most of these features are also found in Internet of Things (IoT) stacks, which power solutions like Amazon Key.
Cloud Robotics platforms, like most other modern software platforms, also feature app stores/marketplaces which vary widely in their purpose, from the simple search and discovery catalogs to more integrated ecosystems where listings are bundled with platform compatibility certifications, flexible pricing and licensing models, integrated billing, etc.
It is helpful to note however that cloud robotics platforms do not help you decide how to distribute/partition a task between the robot(s) and the cloud and this responsibility lies with the customers.
Internet-connected robots are a basic requirement of the newly popular Robotics-as-a-Service (RaaS) business models. Modern industrial robots promise flexible production lines (enabling high-mix, low-volume production) and 24x7 lights-out operations. But industrial robots are still expensive and high upfront capital investments are difficult to justify when the return on investment (ROI) is hard to measure. RaaS is breaking the barriers to entry. Many of the features available in a cloud robotics stack are of great value to RaaS providers.
The big cloud providers, as part of their cloud robotics efforts, are also making investments in ROS 2 — the second and the latest revision of Robot Operating System (ROS), which is the most popular robotics middleware framework in use today. ROS 1 was designed by a single company (Willow Garage) primarily for academic research (in the pre-cloud era) whereas ROS 2 is an industry-led project. ROS 2 mostly is a catch-up effort to the advances in distributed systems and modern cloud-native practices. Once stable, ROS 2 would also be the preferred choice for cloud robotics stacks.
Looking beyond cloud robotics, investments and advances in technologies like Edge Computing and 5G should push modern robots further and make them even smarter, faster and cheaper.
Open Factory provides Cloud Robotics consultancy at various forums. Drop us a message on our website if you need help with your robotics strategy.
