iRobot Case Study

iRobot, a leading global consumer robot company, designs and builds robots that empower people to do more both inside and outside the home. iRobot created the home-cleaning robot category with the introduction of its Roomba Vacuuming Robot in 2002. Today, iRobot is a global enterprise that has sold more than 20 million robots worldwide. iRobot's product line, including the Roomba and the Braava family of mopping robots, features proprietary technologies and advanced concepts in cleaning, mapping, and navigation. iRobot engineers are building an ecosystem of robots and data to enable the smart home.

The first Amazon Prime Day was a good one for iRobot. On that July day in 2015, the company sold 14,000 of its Roomba robotic vacuums. This one-day sales number underlined the reputation for innovation and value iRobot had built among its customers. It also underlined a challenge iRobot would face just a few months later, in September, when the company would release its first internet-connected Roomba vacuums.

“Roomba vacuums are popular Prime Day purchases and holiday gifts, and people want to try them out as soon as possible,” says Ben Kehoe, a cloud robotics research scientist at iRobot. “So there are certain days, particularly after big sales events like Prime Day, when a large number of customers send their newly purchased Roomba vacuums on their first cleaning missions.”

In July 2015, spikes in new Roomba use did not directly impact the company’s day-to-day operations. But starting that September, large numbers of people trying out the new connected Roomba vacuums would result in large volumes of traffic through the iRobot HOME App, the mobile app customers would be using to set up and control their connected robots. In other words, iRobot, which had so far operated primarily as a hardware vendor, was about to bet its central line of business on its ability to run a high-availability, customer-facing cloud application and an Internet of Things (IoT) backend platform.

It was a gamble that paid off. Today, iRobot reports that connected Roomba vacuums operate in more than 60 countries, with total sales of connected robots projected to reach more than 2 million by the end of 2017.

iRobot isn’t stopping there. Not only is the company sending its connected Roomba vacuums to save us from the drudgery of vacuuming; the company believes these robots can one day help solve a problem that has so far prevented the IoT industry from delivering on the true promise of smart-home technologies: a house that simply knows what to do and reacts accordingly.

First Priority: Customer Focus
As always, however, iRobot remains focused on its central value proposition: leading-edge robots that relieve customers from menial chores and give them time for what’s more important. That customer focus is why iRobot started offering connected Roomba vacuums. “Customers are demanding easier ways to interact with a growing number of products and technologies throughout the home,” says Kehoe. “Cloud connectivity provides Roomba customers with even more convenience and control, so they can use their phones to manage their Roomba, wherever and whenever it’s convenient.”

In 2013, as it planned the platform that would be the foundation of this new offering, iRobot decided to build it in the public cloud. For the initial launch, iRobot worked with a vendor that provided a turnkey solution for launching the first cloud-connected Roomba in September 2015. However, as iRobot began to build out its family of connected Roomba vacuums—and as the sheer number of connected customers and services quickly multiplied—iRobot recognized it needed a solution that could scale more quickly and allowed for more direct control.

“Following the launch of our first connected product, it became increasingly clear that we couldn’t achieve the scale and extensibility we needed with the turnkey solution we were using,” says Kehoe.

iRobot decided to move its mission-critical platform to the Amazon Web Services (AWS) Cloud. “The AWS Cloud offered an essential combination of scalability, global availability, and breadth of services,” says Kehoe. “What really grabbed our attention is that AWS offers powerful tools and integration capabilities that enable us to use a serverless architecture that saves us the headaches of learning to scale.”

To run the web applications that connect to the new Wi-Fi-connected Roomba vacuums, iRobot is using about 25 AWS services. At the core of the iRobot platform are AWS Lambda and the AWS IoT platform. AWS IoT, which can process trillions of messages between billions of devices and AWS or other endpoints, provides the connectivity layer between the robots and the iRobot cloud platform. AWS Lambda runs code in response to events to provide function-based compute services for the serverless backend that powers the iRobot cloud application. The solution also uses Amazon Kinesis, which ingests real-time data streams from robots, and the Amazon API Gateway, which can publish APIs and secure them at any scale.

By using a serverless architecture based on AWS IoT and AWS Lambda, iRobot is able to keep the cost of the cloud platform low, avoid the need for subscription services, and manage the solution with fewer than 10 people. “Running on an AWS serverless architecture lets us focus on code and customers rather than operations,” says Kehoe. “The AWS serverless architecture and the ease of use of the AWS services inside it free up developer time to produce business value.”

Gartner, Inc., forecasts that there will be 20.4 billion “connected things” in use worldwide by 2020. In other words, the number of homes containing connected refrigerators, lights, HVAC systems, and security alarms is only going to grow. But iRobot argues that the value of all this connectedness will remain out of reach without a crucial category of data: up-to-date mapping information about the houses where the devices reside.

“To achieve a seamless smart-home experience in which the home, and the smart devices within it, responds to our daily needs autonomously, two things must happen,” says Kehoe. “First, the burden of programming devices must be removed from the consumer. And second, the home needs to understand itself: what the layout of the home is, the location and purposes of each room, and where the home’s various smart devices are located.”

According to Kehoe, connected robots like the Roomba hold the key. iRobot’s Roomba 900 Series vacuums, for example, use visual simultaneous localization and mapping (vSLAM) technology to navigate an entire level of a home. As the Roomba vacuums, it builds a map of the home, and it is this map that could provide the foundational information needed to help a smart home understand itself and enable truly seamless, autonomous home automation. While today’s Roomba is creating maps to clean more effectively, future generations of Roomba vacuums could combine these maps with additional sensors and supportive cloud services to manage the myriad smart devices within a home.

As straightforward as this vision is to state, it’s quite challenging to achieve, but Kehoe says it’s possible in the AWS Cloud. “The information pathways for something like this are incredibly complicated, security for customer information is crucial, and execution won’t ultimately be possible without powerful artificial-intelligence and data-analysis tools. That’s why we’re building all this in the AWS Cloud. Amazon Web Services is the key to making this vision a reality.”