DJI's new robovac can climb obstacles, vacuum quietly, and claims much improved privacy
DJI announced the Romo 2 series (P2 and A2 models), featuring improved navigation with 2mm obstacle detection and the ability to identify transparent objects like glass and mirrors The robot can now climb steps up to 4cm high using two legs on its base, with suction power reaching 36,000Pa on carpet A new Local Data Mode severs internet connectivity with the DJI Home app, addressing serious security vulnerabilities from the first-generation Romo that exposed floorplans, camera feeds, and allowed
Analysis
TL;DR
- DJI announced the Romo 2 series (P2 and A2 models), featuring improved navigation with 2mm obstacle detection and the ability to identify transparent objects like glass and mirrors
- The robot can now climb steps up to 4cm high using two legs on its base, with suction power reaching 36,000Pa on carpet
- A new Local Data Mode severs internet connectivity with the DJI Home app, addressing serious security vulnerabilities from the first-generation Romo that exposed floorplans, camera feeds, and allowed remote control
- Romo 2 incorporates "human-like reasoning and concept transfer" to learn from past obstacles and adapt to new objects encountered in the home
- US launch is unlikely due to ongoing restrictions and an FCC ban on foreign-made robot vacuums
Why It Matters
DJI's first-generation Romo suffered catastrophic security flaws that compromised user privacy on a massive scale, making the Romo 2's Local Data Mode a critical response to rebuild trust in AI-powered home robotics. The emphasis on local-only operation reflects a growing industry trend where consumers and regulators demand privacy-preserving designs for devices that map and monitor private spaces. For AI practitioners, this highlights the importance of embedding security and privacy-by-design from the outset rather than retrofitting it after breaches occur.
Technical Details
- Navigation & Perception: Enhanced obstacle avoidance detects objects as small as 2mm and can identify transparent materials (mirrors, glass), addressing a common failure mode for LiDAR and camera-based systems
- Mobility: Two legs on the base enable climbing over single steps up to 4cm or two steps at approximately double that height, expanding the robot's operational range beyond flat surfaces
- Cleaning Performance: 36,000Pa suction on carpet, with extended brush and mop arms designed to reach sideboards and under furniture
- AI Reasoning: "Human-like reasoning and concept transfer" allows the robot to generalize from previously encountered obstacles to novel objects, suggesting a shift toward more adaptive, learning-based navigation rather than purely rule-based systems
- Local Data Mode: Disables internet sync with the DJI Home app, allows complete camera and microphone shutdown, and permits data deletion — a significant architectural change toward offline-first operation
- Models: Two variants — the transparent P2 and opaque A2 — both sharing the same core upgrades
Industry Insight
- Privacy-by-design is becoming a competitive differentiator in home robotics; DJI's Local Data Mode sets a benchmark that competitors will likely need to match as regulatory scrutiny intensifies around AI devices in domestic spaces
- The US market exclusion due to geopolitical restrictions represents a significant revenue opportunity lost for DJI and underscores how trade policy directly shapes AI hardware deployment timelines
- The integration of adaptive reasoning capabilities ("concept transfer") signals that robot vacuums are evolving from pre-programmed cleaners into genuinely learning systems, which could accelerate consumer adoption if reliability keeps pace with the claims
Disclaimer: The above content is generated by AI and is for reference only.