Uploaded December 2019 | Updated September 2026, 5 days ago
Data collected by vacuum cleaning robot sensors is highly privacy-sensitive, as it includes details and metadata about consumers’ habits, how they live, when they work or invite friends, and more. Connected vacuum robots are not as low-budget as other IoT devices and vendors indeed invest into their security. This makes vacuum cleaning robot ecosystems interesting for further analysis to understand their security mechanisms and derive takeaways.
In this talk we discuss the security of the well-protected Neato and Vorwerk ecosystems. Their robots run the proprietary QNX operating system, are locally protected with secure boot, and use various mechanisms that ensure authentication and encryption in the cloud communication. Nonetheless, we were able to bypass substantial security components and even gain unauthenticated privileged remote execution on arbitrary robots. We present how we dissected ecosystem components including a selection of vacuum robot firmwares and their cloud interactions.
Data collected by vacuum cleaning robot sensors is highly privacy-sensitive, as it includes details and metadata about consumers’ habits, how they live, when they work or invite friends, and more. Connected vacuum robots are not as low-budget as other IoT devices and vendors indeed invest into their security. This makes vacuum cleaning robot ecosystems interesting for further analysis to understand their security mechanisms and derive takeaways.
In this talk we discuss the security of the well-protected Neato and Vorwerk ecosystems. Their robots run the proprietary QNX operating system, are locally protected with secure boot, and use various mechanisms that ensure authentication and encryption in the cloud communication. Nonetheless, we were able to bypass substantial security components and even gain unauthenticated privileged remote execution on arbitrary robots. We present how we dissected ecosystem components including a selection of vacuum robot firmwares and their cloud interactions.
