Teresa Escrig

News and oppinion about Cognitive AI & Robotics

Cloud Robotics: benefits to adopt, drawbacks to solve


For us humans, with our non-upgradeable, offline meat brains, the possibility of acquiring new skills by connecting our heads to a computer network is still science fiction. It is a reality for robots.

Cloud Robotics can allow the robot to access vast amounts of processing power, data and offload compute-intensive tasks like image processing and voice recognition and even download new skills instantly, Matrix-style.

There is an excellent post at ieee spectrum about Cloud Robotics that I absolute recommend to read for those who want to know what is next in the Robotics world.

Here are the benefits I see by using Cloud-enable robots:

  • Provides a shared knowledge and skill/behaviour database, therefore we can make the robots smarter:
    • With common ontologies giving names and meaning of all kinds of things and concepts, the robots will be able to interchange knowledge and how to name and use things, or how to do things.
  • Offloads heavy computing tasks to the cloud:
    • Cheaper, lighter and easy to maintain hardware.
    • Longer battery life.
    • Invisible software and hardware upgrades.

Any old robot full of dust, stored in the closets of the research labs of the world can be reused again by using the hardware and software infrastructure of the cloud.

There is also some drawbacks that we need to consider when we want to provide a solution to our clients which integrates Cloud Robotics:

  •  It is necessary to careful define an equilibrium between software in the cloud and on-board depending on the application for several reasons:
    • The robot can become suddenly stupid if it looses internet connection.
    • It can be hacked.

I am starting to define a big research project on Cloud Robotics at University of Washington with the potential collaboration of Cognitive Robots. If you belong to a Robotics company or a research group and are interested in collaborating with us, please, send an email to mtescrig@uw.edu.

Your comments about benefits / drawbacks / examples of Cloud Robotics / groups that are working in the area / etc. are more than welcome.

22 Responses to 'Cloud Robotics: benefits to adopt, drawbacks to solve'

Subscribe to comments with RSS or TrackBack to 'Cloud Robotics: benefits to adopt, drawbacks to solve'.

  1. Dear Teresa,

    You probably understant Spanish and Catalan language but let’s use the language of your excelent blog despite my english is horrible.

    I present myself, I’m the Founder, Owner and President of RO-BOTICA GLOBAL S.L., founded in year 2006, a company specialized in distributing and selling which I consideer best commercial educational (our strategic focus) and personal robots around the world. We have a face shop in Barcelona, I think is the unique in Europe.

    I have checked how research, educational and personal robots evolve in these 6 years as my main professional value which is selecting the robotic product to distribute (and of course make my company grow each year and sell more with coherence and hard work).
    I’m privileged observing the new tendences of the robotic market and it’s technology. I have a large professional past developing in the Cloud that makes to me have the ideas clear with Internet technologies and Cloud Robotics.

    I encourage you working hard in your new project, this emerging technology and new concept which I call CLOUD ROBOTICS.

    I autodefine myself as conceptual definitor of the new concept Cloud Robotics, because I consideer that now the required technology is mature and growing exponentially (mobile-smartphones technology and cloud computing technology over Internet).

    I have been more than two years thinking in this new concept and how make real and how could be possible create this new high level layer of applications specialized only for robots, in a time of 5 years or less, that in my opinion will bring cheap and popular cloud-robots, using “on demand” thousands of specialized tools (which I call “AI bricks”, or expert tools), like this helicopter pilot program of the video of the excelent movie Matrix. Yes, this is not science fiction. The objective is every robot can do and complete the mission ordered by his owner human, and if the mission fails, learn from the own experience of the robot in repositories of data hosted in Cloud.

    I think that thousands (milions?) of AI bricks (Tools) on Cloud giving concurrent service to millons of cheap robots can configure a strong and generalist intelligence from the Turing Test point of view that will cause to each cloud-robots with a Wifi or 3G, 4G connection to the cloud a strong and generalist intelligence indistinguishable from human one, or higher than human ones (maybe the resultant intelligence will be suprahuman).

    The challenge is critically big, envolving thousands (milions?) of transdisplinary persons, jobs, startups, etrepeneurs, the open source community, companies of all sizes including Google, Apple or Microsoft, as WWW or Web 2.0 (the high level layer application thinked only for humans) did from the second half of 90’s years until now.

    In my opinion is required to create a new high level layer of applications on the cloud thinked only for machines and cloud-robots over basic and low level TCP/IP protocols of Internet, wich were invented in 60’s years for military and defense purposes for having a descentralized network of communications invulnerable to a atomic attack from URSS, during cold war.

    I encourage you read a part of my reflexions and thoughts about CLOUD ROBOTICS, and what is required to do to create it, and the bases of this new concept in the RO-BOTICA’s corporative blog in tens of posts writen by myself during last 2 years:


    The social changes envolved for human civilization are enormous when a machine or robot can do near everything a human does in way indistinguishable from humans. Changes in education will be critical.

    I offer to you all my help and experience in your new project and this new discipline of big future, and let’s create the colosal project called CLOUD ROBOTICS as soon as possible with the collaboration of everybody, and with standard protocols and rules for making it independent of the Brand, OS, and technology of the robot (client side) or cloud server-side, adding everybody to the required definition and developing.

    Best regards,
    Toni Ferraté
    Owner & President

    Toni Ferraté

    9 Aug 12 at 11:00 pm

  2. Hi Toni, yes I speak Spanish and Catalan as well. Lets continue with English in here, just to reach a broader audience. We can talk in Catalan through skype.

    I am glad to see Spanish efforts towards that area.

    I totally agree is the future of Robotics.

    Lets collaborate.

    Kind regards,

    Teresa Escrig

    10 Aug 12 at 12:35 am

  3. Please excuse the self-promotion, but I have an opportunity to receive funding in this area, and if you have a few moments you can help.

    Mozilla Ignite is a competition for ideas on how to use very fast internet connections. A few months ago I submitted an idea along the lines of this thread called A Mind in the Clouds.

    You can help by reading over the short proposal, signing up and then voting for the idea.


    Thanks for your time!
    Ian Danforth

    Ian Danforth

    10 Aug 12 at 4:55 pm

  4. Hi Ian, I voted for your idea and I promoted it in twitter. It is great!

    Good luck!


    Teresa Escrig

    10 Aug 12 at 7:01 pm

  5. Teresa – I’ve been reading your posts avidly over the past few months. Its great to see such a passionate advocate for all things robotic and I’d welcome the opportunity to speak further about cloud robotics offline. I’ll be sure to connect with you to start that dialogue.

    For some back ground, my company is Bossa Nova Robotics, a spin off from Carnegie Mellon and we have to date deployed tens of thousands of cloud connected toy robots.

    Toys is a market that enabled us to get from proof of concept to deployment and scale very quickly and we did so by leveraging the power of the cloud. We’ve been using Robot Cloud Services in commercial deployment for over a year now and have seen the immediate benefits not just cost savings and improved performance but the ability to personalise the user experience at the individual level. Its this last point that for me is the magic of cloud robotics that enables the delivery of social robotics.

    The other beauty of cloud robotics is its modularlity. It is quickly scalable and product agnostic so what we built for toys is now being tested for much more complex robots with the same goal – a personalized user experience.

    I’d be interested in speaking further and understanding how we may be able to help with your University of Washington research.


    Martin Hitch

    10 Aug 12 at 9:29 pm

  6. Hi Martin,

    Thanks for reaching me here. Please, contact me offline. I will love to know all you have done about Cloud Robotics, and see how we can collaborate in the project.


    Teresa Escrig

    10 Aug 12 at 10:17 pm

  7. Hello Ian,

    I voted for your excellent idea and I liked it in Twitter.

    Once login I registered my idea with some similitudes with yours, but not exactly: what I’m proposing is a new high level layer of aplication full of “AI bricks” to integrate, and thinked only for cheap robots and machines and different to human layer app WWW and Web 2.0 with the first objective to passing Turing test, and then a supra-human intelligence for resolve the hard problems of human civilization as unfair and clima change that will autodestroy us and Earth.


    Toni Ferraté

    10 Aug 12 at 10:56 pm

  8. Thank you Diego García Sánchez, from AISoy Robotics (http://www.aisoy.es/) for your interest in participating in the project. We will be in touch.


    Teresa Escrig

    10 Aug 12 at 11:39 pm

  9. The Robo-Earth project, a research project funded by the EU and being conducted by a partnership of universities in Europe, is currently tackling this challenge, I would highly recommend checking it out!


    12 Aug 12 at 9:34 am

  10. Mike

    12 Aug 12 at 9:35 am

  11. Great Article. We started RobotTwinkee.com last year with the hopes of helping Robot Developers and Robot Consumers procure what they need to either build, buy, or teach about Robots. We also have been discussing building a cloud infrastructure running Ubuntu for our customers. We have been an IBM SmartCloud provider since last year but it has been very difficult to get Ubuntu up on their cloud. Fortunately we are now a Cloud Provider of Amazon Web Services where we can offer Cloud Servers running Ubuntu. Let me know if you need your own Amazon Cloud or if you want me to set you up with a Cloud Server running Ubuntu.

    Eric Gray

    Eric Gray

    12 Aug 12 at 10:19 pm

  12. It’s an excellent idea but I just feel first we need to realize humanoid robots completely..

    Sai Pitta

    13 Aug 12 at 12:11 am

  13. Thanks Mike, I already check out the Robot-Earth project and seems very impressive. It is a very important reference for our project. Thanks again.

    Teresa Escrig

    14 Aug 12 at 12:34 am

  14. Hi Eric, yes we are using Ubuntu in our Cognitive Brain for Service Robotics at C-Robots. I will take your offer into account. Thanks.

    Teresa Escrig

    14 Aug 12 at 12:36 am

  15. @Sai, the robots don’t have to look like us to perform repetitive tasks. In fact, it is even better that adopt the best shape for the task at hand.

    Teresa Escrig

    14 Aug 12 at 12:38 am

  16. Yes teresa you are right. I think I am comparing robots with fiction. Yes robots should be more about completing a purpose rather than being like humans.

    Sai Pitta

    14 Aug 12 at 12:52 am

  17. It’s a really good idea..Combining cloud storage and robotics is the thing for the future. Actually I had an idea of combining cloud computing and robotics to make a group of robots to work on certain tasks. you know for example lets we want to a 5 robots to play soccer we can just use cloud computing to combine the minds of 5 robots into a single one and thus creating an effective gameplay.

    Sai Pitta

    14 Aug 12 at 12:55 am

  18. @Say, I don’t agree with you.

    I think the best and productive option is devolpe thousands (milions?) of AI bricks or specialized tools on high level layer of application on Cloud, which used bricks will survive by Natural Selection and human feedback, to construct a generalist intelligenge like human (or beyond human) which could be used by cheap and popular cloud-robots and machines with a client-side brain not more complex than a simple smartphone of last generation.
    The social changes will be very important ans these new tools (cloud-robots) could help milions of humans and human civilization in avoideng self destruction, climate change, making company to el and alone people, make people more happy, avoiding humans certain tasks,artificial vision for pattern recognition, etc.

    The main reason is that required technology for Cloud Robotics (mobility an cloud computing is now mature and growing exponentially), but a humanoid robot is really a complex and expensive project and we are really far to develope a mechanichal humanoid machine near indistinguishable from humans, including in Research phase, and a humanoid robot without a hard and generalist intelligence indistinguishable from human one has low value for common people.
    Anyway we can make more steps and concentrate in humanoid mechanical parts and human mechanical behaviours like hands, arms and grippers to manipulate things, improve walking and running dynamic alghoritms , improving matherial, sensors and actuators (artificial muscles), security for not makig damage to humans (floppy robots), etc.

    Toni Ferraté

    14 Aug 12 at 7:54 pm

  19. @Mike, the 2 main conceptual differences between Robo-Earth and what I call Cloud Robotics are:

    1) Robo-Earth works over WWW, the high lever layer application over TCP/IP thinked only for humans until now.

    I propose a new high level layer application over TCP/IP differet to human WWW one thinked only for AI cheap and popular robots and machines.

    2) The main puropose of what I call Cloud Robotics is reaching and creating a strong and generalist AI passing Turing Test, or even AI beyond human intelligence, for example, using RTS (Robot Tools Servers) to redirect a cloud robot to optimal “AI Brick/s” for complete “on demand” a cloud-robot mission, as human WWW layer uses DNS where have to connect all human machines in low level protocols for resolve the problem of bad memory that we have humans remembering IP numbers, but not alphanumeric names, words and domains.

    Cloud Robotics is related only to communication standard protocols and ways to structurate the information for stablish communication between RTS/AI Bricks and cloud robots, for example XML and some rules for communicate “on demand” a mission with atached URL media or sensor captured by sensors information, but not says nothing about the OS, hardaware technology, brand, etc, to use, like ROS from Willow Garage, Linux, etc… for adding everybody to this colosal project. All companies, people, open community, entrepeneus, startups, or key players as Google, Apple or Microsoft could design and integrate their AI bricks on Cloud for providing service to thousands (milions?) of concurrent cloud-robots and machines with independence of it’s technology. Is the same spirit like WWW.

    The main purposes of Robo-Earth are explained in the link above.

    Toni Ferraté

    14 Aug 12 at 8:27 pm

  20. Thank you guys for your comments!

    Teresa Escrig

    15 Aug 12 at 12:07 am

  21. Thank you Eduardo Castello at Ishiguro Lab – https://sites.google.com/a/irl.sys.es.osaka-u.ac.jp/top-en/ at Osaka University for your interest in this project.

    Teresa Escrig

    15 Aug 12 at 6:29 pm

  22. Hi Teresa

    Very nice article telling how to shift processing power intensive tasks from onboard to cloud and robots getting knowledge from internet. I have one question(may be a silly one as my knowledge in this field is limited) how to stop these connected robots from getting bad knowledge(crimes and other destructive knowledge) which is readily available on internet these days. As per my knowledge there are limitations for blocking rouge contents on internet.


    24 Sep 12 at 11:13 am

Leave a Reply