Image from Youngester
Future development of robotic intelligence may come in the form of Robotics as a Service (RaaS) - an integral component in creating more flexible, multipurpose robots.
Software as a Service (Saas) and Big Data companies have been at the forefront of delivering products previously unavailable to most businesses due to those products’ costs and complexities when instituted at a single corporate entity. These products allow businesses of any size to focus exclusively on their core competency, outsourcing or subcontracting all but the most critical endeavors to a SaaS provider that utilizes best practices for that specific niche. While the popularization of SaaS solutions has been transformative for knowledge workers and their ability to process abstract, SaaS technology will be nothing less than revolutionary when applied to the field of robotics.
This post will provide a brief look at robots before the popularization of Big Data and SaaS technology, examine several companies currently implementing robots with cloud capabilities, and finally, spotlight several companies that are leaders in the emerging area of Robotics as a Service (RaaS), where the physical robot is but one component in automating a physical task.
“Dumb” robots constitute the first phase of task automation, particularly those that are consumer facing. Perhaps most prominent amongst consumers is the Roomba (made by iRobot and introduced in 2002), a robot that vacuums a consumer’s floors automatically. While the Roomba may accomplish its task given a standard room without obstructions, there is little capability for the Roomba to understand new data models unless they are implemented via a firmware update. The Roomba is “dumb” in that it is unable to learn over time via its own execution of tasks, but more importantly, it is incapable of incorporating the knowledge of other Roomba robots into its own information model.
“Smart” robots introduced the second and current phase of task automation, where robots learn both from their own activities as well as those of their worldwide counterparts. Google’s self-driving cars illuminate how dramatically robot performance can improve given both self-learning and other-learning capabilities. Started in 2009, the Google autonomous car project has driven nearly 700,000 miles autonomously and its cars have yet to be at fault in an accident, an impressive feat. Even more impressive is that every mile driven, by every car, improves the capabilities of every other autonomous car Google creates.
The Datafox platform provides tools for surfacing companies in this sector....
- Momentum Machines is crafting a robot that produces customized burgers with minimal human intervention.
- Foldimate has created a machine that automatically folds clothes in record time.
- SterraClimb has manufactured a dolly that self-balances and climbs staircases automatically.
While the individual machine’s results are notable, the comparison to human labor becomes more striking at scale. Assume that Momentum Machines has 10,000 operational burger makers working at near capacity, 360 burgers/hour, 10 hours/day for 300 days of the year. In total, Momentum Machines will have made 10,800,000,000 burgers, or 1/5th of the total American’s typically consume in 1 year. With a sample size in the billions, Momentum Machines will likely be able to produce a better burger with its algorithm than any fast food worker ever could. It is this abstraction of physical information (road maps, burger making, laundry folding) away from the individual automaton and into a broadly accessible body of knowledge that will allow robotic algorithms to literally perfect themselves over time and is also absolutely critical to the next phase of automation.
Robotics as a Service (RaaS) – Smarter Robots
“Smarter” robots are those that incorporate data from their individual experiences, those of their similarly programmed brethren, and most importantly third party artificial intelligence providers. Similarly to how one interacts with an API or SaaS provider, one should not necessarily need to understand how a box picking, car driving, or steel welding algorithm for a robot works. The knowledge, and therefore the algorithm, can be abstracted away by a third party that is constantly improving and optimizing the algorithm for machines with certain sensors and capabilities. Robot manufacturers and the accompanying software, need not be created by the same company. For example, while one would assume Google may be reluctant to allow other car manufacturers access to the invaluable autonomous car data it has recorded, it may yet realize that the role of autonomous car software provider is strategically more valuable than manufacturing the cars themselves.
Several companies found on the Datafox platform could be categorized as Robotics as a Service providers, as they seek to provide the intelligence necessary for a broad array of robotic tasks. Neurala is creating a neural network that robots can apply for an enormous number of different tasks. Airware is producing a Drone as a Service platform on which developers and manufacturers can use their software for common drone programming challenges and customize when necessary. Universal Robotics is building software that is configurable to the physical task at hand, such as stacking pallets or moving boxes.
This division of autonomous labor will become more relevant as robots become more malleable to their environments and the tasks they are assigned to accomplish. Similar to how smartphones can now generally be assumed to possess certain sensors (GPS, Accelerometer, Compass, etc), were robotics to travel down a similar path, third party robotic intelligence would become even more integral to creating flexible, multipurpose robots with expandable, and improvable, capabilities.
For extensive data on all of the companies mentioned in this post, as well as a summary of the Robotics as a Service sector as a whole, explore our RaaS DataFox watchlist:
Want more exclusive insights? Signup for the FREE weekly DataFox Digest!
Investment Advice Disclaimer
The information contained in this post, while intended to interest investors, does not constitute investing advice or services. Future trends and predictions contained herein are derived from industry news and public information. These “forward looking-statements” are speculative, subject to risks and uncertainties, and are not guaranteed. Actual results may vary. Potential investing decisions should be made by consulting a professional legal, tax, investing or accounting professional, where risks and suitability can be throughly evaluated. DataFox makes no obligation to update or revise any of the statements contained herein.