Read any major business news site, and you’re probably going to notice a familiar refrain: Automation. We’re living through another business revolution driven by increased automation, and this time it’s all about machine learning and artificial intelligence, and the robots that use both.
For many businesses, wanting to get in on the automation transformation is as far as they’ll get because resources aren’t available for actual execution. Developing a custom-tailored robotics product takes a wide variety of skills, many of which exist as separate disciplines.
Amazon Web Services (AWS) has announced a new product that can take the hesitation of out of automation, and it’s called AWS RoboMaker. RoboMaker, in Amazon’s words, is “an integrated set of software and services for customers to develop, test, and deploy intelligent robotics applications at scale.”
If your business has been on the fence about the cost-to-benefit ratio of adopting robotic technology, you may have a new solution in AWS RoboMaker. Read on to find out all you need to know about this cloud robotics platform.
What is AWS RoboMaker?
Amazon defines a robot as a machine that can sense, compute, and take action, which is exactly what AWS RoboMaker is designed to help its users create.
RoboMaker, like many other AWS products, takes the legwork out of IT infrastructure. In this case, it isn’t cloud storage or big data processing–it is the development, testing, deploying, and fleet management aspects of business robotics.
Roger Barga, GM of AWS RoboMaker, said the impetus behind RoboMaker’s design is comments AWS has received from users. “They spend a lot of time setting up infrastructure and cobbling together software for different stages of the robotics development cycle, repeating work others have done before, leaving less time for innovation.” To that end, RoboMaker contains all the backbone of robotics development, so all a RoboMaker user has to do is provide the hardware.
SEE: Amazon Web Services: An insider’s guide (free PDF) (TechRepublic)
RoboMaker is made up of four components: Cloud extensions for the Robot Operating System (ROS), a development environment, simulation services, and an over-the-air (OTA) fleet management system.
The first component, cloud extensions for ROS, is designed to make connecting the open-source Robot Operating System to AWS services as simple as connecting an API. Extensions exist to connect ROS-powered robots to:
- Amazon Kinesis for extracting data from real-time video streams;
- Amazon Rekognition for analysis of objects in videos and images;
- Amazon Lex for speech recognition and natural language understanding;
- Amazon Polly for deep-learning powered speech synthesis; and
- Amazon CloudWatch to monitor and manage robotics software and hardware.
The second component, the RoboMaker development environment, places everything needed to build robotics software onto the AWS integrated development environment, Cloud9. The development environment includes:
- Pre-installed and pre-configured versions of ROS for building robot software based on certain basic frameworks;
- A ROS build tool for bundling dependencies into ROS code specialized for different kinds of hardware;
- Sample applications that are bundled with simulation code for immediate testing of basic models; and
- A browser-based editor with auto-completion and suggestions to speed up coding and help developers avoid errors.
The simulation component of AWS RoboMaker is a software environment designed to simulate real-world scenarios to make testing robot hardware more affordable. It can simulate a variety of physical environments, such as retail stores, racetracks, and warehouses, and it supports parallel simulations; plus, like other AWS, products it scales computing infrastructure use based on need.
AWS RoboMaker simulation software includes:
- The ability to gather simulated machine learning data for training models before hardware rollout;
- Integration with the Gazebo simulation engine, Open Dynamics Engine physics simulation, and the OGRE rendering engine;
- Command-line tool support for Gazebo, rviz (visualization engine for ROS), and rqt (QT-based framework for ROS GUIs); and
- Amazon CloudWatch and Amazon S3 integration for job monitoring and logging of simulation data like collision, velocity, simulated hardware battery levels, and other metrics.
Fleet management, the final piece of the AWS RoboMaker pie, is a pre-built OTA hardware management tool that includes:
- Robot registration, security, fault tolerance monitoring, support for multiple fleets, and other features necessary for maintaining large groups of hardware;
- OTA software deployment for OS updates, new applications, patches, and features; and
- AWS IoT Greengrass integration that allows fleet managers to use Greengrass features including local Lambda functions, machine learning inference, local messaging, and hardware security to both x86 and ARM-based hardware.
All four components combine to make AWS RoboMaker a one-stop product for organizations aspiring to a higher level of automation. While it remains to be seen how well RoboMaker will work in the long run, it makes a great place for companies in the trial phase of robot automation to begin their experiments.
- Five most important cloud announcements at AWS re:Invent 2018 (TechRepublic)
- AWS launches RoboMaker dev service for building intelligent robotics apps (ZDNet)
- AWS IoT suite now includes SiteWise for industrial data collection (TechRepublic)
- Amazon’s free training: Internal machine-learning courses are now open to all (ZDNet)
Who is affected by AWS RoboMaker?
As Amazon said in its RoboMaker press release, “developing, testing, and deploying intelligent robotics applications is difficult, time consuming, and demands a diverse set of hard-to-acquire skills.” The need for many of those skills has been rolled up into RoboMaker, which is great news for businesses that want to save time and money on robot-based automation.
The effects, and benefits, to companies that use RoboMaker is obvious: Money will be saved on hardware deployment costs, building basic ROS models and simulation environments, rolling out a fleet management tool, and finding the right dependencies to make hardware and software work together. Companies will also potentially save months of development and deployment time–everything an organization needs to roll out a fleet of robots (save the actual robots) is part of RoboMaker.
There’s no reason for organizations not to at least give RoboMaker a trial run–the time and money saved alone can have a huge effect on a company’s bottom line and its ability to deploy a fleet of machine learning-powered, semi-autonomous, cloud-powered robots before its competitors who opt for the longer, in-house development route.
Data scientists, developers, and infrastructure engineers may not be so thrilled with RoboMaker–it’s essentially another layer of automation that gets rid of their roles in setting up the tools that RoboMaker provides. That’s not to say there may not be a role for those professionals in an organization that uses RoboMaker, but as with any automation tool (cloud services included), jobs are either lost or transform quite a bit.
Developers may be the least negatively affected by RoboMaker; instead of spending time programming the backbone of a RoboMaker-like suite of tools, they can get right to work on developing the robots themselves.
Hardware engineers who specialize in robotics may benefit from RoboMaker because, instead of spending time in a server room setting up robot support infrastructure, they can focus on the robotics portion of their job.
Data scientists will still be needed as well, and those that want to work with robotics should become familiar with the data collection tools included in RoboMaker, like Kinesis, Rekognition, and CloudWatch.
SEE: More must-see cloud computing coverage (TechRepublic Flipboard magazine)
It remains to be seen what larger effect RoboMaker will have on the tech industry. Though, if general trends in cloud computing are any indicator, it’s likely that RoboMaker will be a smash success that could change the face of robotics, moving the tech world solidly into a robotics-as-a-service (RaaS) world.
- AWS re:Invent 2018: A guide for tech and business pros (free PDF) (TechRepublic)
- Top 5: Reasons not to be scared by automation (TechRepublic)
- Robots and the NHS: How automation will change surgery and patient care (ZDNet)
- 3 ways robots can support human workers (TechRepublic)
- Automation will become central to business strategy and operations (ZDNet)
What are examples of AWS RoboMaker in action?
Amazon provided several examples of real-life organizations using RoboMaker prior to its public release; all of the examples show different applications of the platform.
Tool manufacturer Stanley Black & Decker has used RoboMaker to train autonomous ground vehicles and drones to aid in the construction of buildings. It took advantage of 3D simulation to train its 3D site model creation algorithms, and is using fleet management tools to make drones, ground vehicles, and IoT sensors work together to reduce construction costs.
Robot Care Systems (RCS), which provides robot assistants called Lea for elderly and disabled individuals, has used RoboMaker’s ROS cloud extensions to improve Lea’s ability to collect data and provide voice interaction services. By moving much of the computing work to the cloud, RCS has been able to make Lea do more without needing to add more on-board computing power.
FIRST, a robotics educational organization that sponsors robotics competitions and other robotics-based STEM educational programs, is utilizing RoboMaker as an educational tool to make programming, testing, and deploying advanced robotics tools accessible to students.
- 5 ways AWS RoboMaker is transforming the robot programming game (TechRepublic)
- Cleaning machine: SoftBank brings AI to a dirty job (ZDNet)
- How to Automate the Enterprise (ZDNet Special Feature)
- 25,000 robots and counting: Welcome to the fastest-growing segment of industrial automation (ZDNet)
Why should I choose AWS RoboMaker?
If you’re unsure of your organization’s suitability for AWS RoboMaker, there are a few reasons why you should consider it, at least on an experimental basis.
First off, RoboMaker pricing is similar to other AWS services in that you only pay for the computing power you use; this extends to individual tools including Rekognition, Polly, CloudWatch, Lex, and Kinesis–if you aren’t using them as part of your development process, you won’t have to pay for them.
Development environment and fleet management use AWS Cloud9 and Greengrass, and rates are based on the use of those platforms. Simulation pricing is based on individual simulation units (1 SU = 1 vCPU and 2 GB of memory per hour), which cost $0.40 per hour.
If you’re unsure of the financial burden of using RoboMaker, check out its pricing page, which has additional tools to calculate cost.
SEE: Hiring kit: Robotics engineer (Tech Pro Research)
A second, and perhaps the most important, reason to give AWS RoboMaker a try is because it’s the first platform of its kind in existence. The closest second is Google’s as-yet-unreleased Cloud Robotics platform, which Google says will be available to developers sometime in 2019. Until its release, Amazon has a monopoly on cloud-based RaaS software, so if you’re interested in seeing what the future of intelligent robotics looks like, you don’t have anywhere else to go.
- How to automate your business: 3 critical steps (TechRepublic)
- Looking to compare all different types of automation? Now you can (ZDNet)
- How to measure ROI on your company’s automation projects (TechRepublic)
- Could automation and AI be making work more human? (ZDNet)
How can my business start using AWS RoboMaker?
As with other AWS tools, individuals and organizations with an AWS account can get started using RoboMaker right away, even those with free tier accounts.
Those who want to experiment with RoboMaker should be aware that charges may be incurred based on usage.