AWS RoboMaker shuts down after failing to gain traction
Amazon Web Services, or AWS, has officially discontinued RoboMaker, its cloud-based robotics simulation platform. This marks the end of a service that seemed to be misaligned from the start.
RoboMaker offered cloud simulation at scale through the open-source Gazebo physics engine. The system made it possible to spin up thousands of randomized environments and generate pass/fail metrics across them.
“When AWS decides to retire a service or feature, it is typically because its capabilities are better addressed by newer AWS solutions or offerings from our AWS Partner Network partners that better meet customer needs,” an AWS spokesperson told The Robot Report. “In making such decisions, our priority is to provide customers with guidance on available alternatives—whether they are solutions or partner offerings—along with how to migrate their workloads seamlessly, ensuring minimal interruption to their operations.”
RoboMaker users have been encouraged to pivot to AWS Batch. The company told The Robot Report Batch stands out as RoboMaker’s alternative with its multi-container support, allowing multiple containers to run in a single job. AWS wrote a blog for those looking to transition off of RoboMaker.
“This eliminates the need for monolithic containers and enables separate simulation components, making it ideal for autonomous systems testing,” AWS said via email. “Batch provides better cost control by charging only for compute resources and supporting Spot instances.”
“Unlike RoboMaker’s limitations, Batch handles any containerized workload and integrates smoothly with other AWS services,” it added. “Its flexibility in supporting various compute environments and ability to scale from small to large simulations makes it a more versatile solution for modern robotics development.”
AWS RoboMaker tied to iRobot
AWS launched RoboMaker in 2018. iRobot, developer of the Roomba robot vacuum, was at the time one of its largest robotics customers. iRobot expressed interest in a scalable simulation service. It thought cloud-based simulation would be useful for developing robots that operate in diverse environments, like homes with different layouts, flooring, and lighting.
The Robot Report spoke with multiple sources, who wished to remain anonymous, about RoboMaker. They said the product was misaligned with market need and obviously didn’t gain enough traction. One source who previously worked for RoboMaker said the product was “spun up” essentially for iRobot.
“It worked well for iRobot,” the source said. “But there wasn’t much due diligence to see if it was useful for anyone else in the market.”
For a company like iRobot, the ability to quickly simulate in different environments was a valuable capability. But most robotics companies didn’t need simulations at that scale.
“Most companies don’t need to simulate thousands of different environments,” said one source. “They just need a few.”
Amazon Robotics didn’t use RoboMaker internally
The source said the mismatch became clear over time. The source said AWS underestimated how fragmented the robotics industry is, and assumed that it could find “nine other iRobots” to scale RoboMaker adoption. If RoboMaker had been a startup, this source said, it likely would have failed fast. But inside Amazon, jobs and inertia kept the project alive longer than the market justified.
Amazon Robotics, the largest robotics developer in the world, having deployed more than 1 million robots, never adopted the service internally, according to multiple sources.
“The target market for RoboMaker was large robotics companies that wanted to do massive simulation projects,” one source said. “I don’t think most of them found a lot of value in it. Once your product mostly works, are you really going to spend six figures on Amazon to find one or two edge cases?”
The shuttering of RoboMaker underscores a familiar lesson in robotics: what works for one company doesn’t always scale. For AWS, which prides itself on building tools that scale universally, RoboMaker’s discontinuation is a reminder that not every experiment pays off.
