Robot Care Systems Takes Robotics to the Cloud Using AWS RoboMaker

RCS was proud of the many advanced features included in the first version of Lea, which was released in 2018. Still, because of Lea's limited onboard compute resources, the company was forced to leave a few wish-list items for a later release.

"The two most important features we weren't able to include in the first generation of Lea were the ability to collect and share movement and behavior data with patients' physicians, and the ability to remotely receive maintenance alerts and diagnostic information about the device," says Lopes. RCS also would have liked to include a voice interface, AI-powered analysis of data from Lea's 72 sensors, and image processing to improve autonomous navigation, but any one of these would have overtaxed the product's onboard compute resources. "We knew cloud computing offered solutions to these sorts of challenges, but we didn’t see a simple, cost-effective path for connecting Lea to the cloud."

That changed when Amazon Web Services (AWS) invited RCS to join testing of AWS RoboMaker, a service that enables robotics developers to easily develop, test, and deploy intelligent robotics applications at scale. RoboMaker does this by providing cloud extensions for the Robot Operating System (ROS), the most widely used open-source robotics software framework and the one Lea is based on.

"Before evaluating RoboMaker, setting up cloud services for Lea seemed too challenging," says Lopes. "It was a revelation seeing how easily cloud connectivity could be accomplished with RoboMaker, just by configuring some scripts. We immediately realized we could use RoboMaker to take the next release of Lea to a higher level."

By using AWS RoboMaker, RCS is making the next generation of LEA even easier—and safer—for customers to use. The next release of Lea will integrate Amazon Lex, a service for building conversational interfaces using voice and text; Amazon Polly, a service that turns text into lifelike speech; and Amazon Rekognition for facial recognition of the users.

"An intuitive voice interface was on our wish list for the first generation of Lea, but it would have been too demanding for the constrained onboard resources of that version, and too costly for us to develop from scratch" says Lopes. "With RoboMaker, we were able to access Amazon Lex and Amazon Polly and test a prototype voice interface for LEA in just a few hours. That's only one example of the many different AWS services we can now make use of with AWS RoboMaker."

The company will also use Amazon SageMaker to further strengthen its ability to identify risk factors for customers. "Because RoboMaker helps us easily access AWS machine-learning services, we will be able to analyze sensor data from LEA to make predictions and send alerts," says Lopes. "By using RoboMaker to stream data to the cloud, we're enabling the next generation of Lea robots to better detect movement and behavior changes that might suggest a heightened fall risk and reduce device speed or alert caregivers. Without the access to cloud services that RoboMaker provides, LEA wouldn't have been able to use its onboard compute resources to analyze the data quickly enough to manage this."

The AWS Cloud will further augment this crucial analytical capability by using data from all deployed Lea robots for machine-learning model training. "By analyzing data from the robots in the cloud, we can increase the collective knowledge of all our robots at once," says Lopes. "Because of the way RoboMaker helps us connect to machine-learning services in the AWS Cloud, scale is no longer a challenge but a benefit."

RoboMaker's fully managed robotics simulation service will also help RCS save time and money. "With RoboMaker simulation, we can evaluate how our robots will behave in complex, variable situations without needing to spend tens of thousands of dollars for on-premises servers and GPUs," says Lopes. "Once our code works as expected in the simulation, the same code base will work in production, and there is a simple process for releasing it. On our RoboMaker pipeline, the time needed for testing and deployment of new features will drop from months to days."

RoboMaker fleet management features will also help RCS save money and time—and eliminate a scalability obstacle. "Currently, we can only support about 100 robots in the field because updates and maintenance require on-site visits by human technicians," says Lopes. "With RoboMaker fleet management, we can set fault tolerances, receive diagnostic alerts, push out bug fixes and new features—all remotely and over the air. RoboMaker fleet management will enable us to reassign one full-time employee, grow our fleet from 100 to 10,000 or even beyond, and distribute Lea internationally instead of being limited to just one country."

RCS is looking forward to unlocking even more value and creativity in its robotics development processes by using AWS RoboMaker. “Robots used to be restricted by the size of their onboard compute resources, but those days are now over,” says Lopes. “All we needed was a way to bridge ROS and the cloud. RoboMaker is that bridge between robotics and cloud services.”