Roomba Smart Mapping
Creating and launching a new experience for Roomba owners to map their homes, customize rooms, and clean selected spaces — designing the spatial intelligence that made Roomba i7 the first robot to truly know your home.
Overview
Teaching a robot to know your home
The Smart Mapping project was to create and launch a new experience for Roomba owners to send a robot to clean and map their homes, customize their maps, and clean selected rooms. I was a member of the core team that researched, designed, and launched the smart mapping functionality. We conducted several rounds of user research, iterated design, and led Alpha, Beta, and product launch in 2018.
My role
- Lead designer of the Smart Map program and the entire mobile App.
- Responsible for UX/UI design and robot behavior.
- Worked closely with a researcher, a hardware UX designer, a product manager, and engineering teams, including mobile, cloud, and robot software.
Existing robots could only show a one-time clean map per job. Smart Mapping robots remember a persistent map of the user's home.
Problem to solve
Robots that couldn't remember where they'd been
Existing Roombas could not be directed to clean specific areas or rooms. Users often wanted to clean high traffic areas, but those robots could only generate a one-time coverage clean map for each job. A clean map could only show where a robot visited during a job — making clean maps not helpful or easy to orient.
Roomba i7 was the first consumer robot that could remember a persistent map of a user's home. The Smart Mapping project goal was to create an intuitive map training experience and enable users to clean selected rooms.
End-to-end journey map identifying key user touchpoints, unknowns, and risky assumptions across the mapping experience.
Process
Journey map: end-to-end experience
We brainstormed the user journey based on the latest technology and identified unknowns and risky assumptions. I led UX/UI design for the entire mapping journey, but this case study selectively showcases the map customization design process.
Interviewed 8 Roomba 900 series owners in their homes to understand their mental models and pain points.
Discovery user study
Interviewed 8 Roomba 900 series (non-smart mapping robot) owners in their homes.
- Determined current owners' user pain points and user needs
- Understood users' mental model of their home space
- Evaluated the internal engineering prototype for map customization
Participants sketching their home layouts to reveal their spatial mental models.
Three persistent map visualizations generated from users' own clean map data.
How users comprehend the space
Evaluated usability of existing one-time clean maps. We utilized users' clean map data to generate 3 different persistent map visualizations for each participant — testing which representation felt most legible and actionable.
Engineering prototype testing
We tested the internal engineering prototype with users to identify friction before committing to a design direction. Four critical issues surfaced:
- No landmark: Some users struggled to judge the accuracy of auto-generated room divisions without spatial reference points (negative space clutter made orientation harder, not easier).
- Unintuitive interactions: The merge/split/label tool was not intuitive or easy to use — 7 out of 8 users expressed frustration, and several would have abandoned. Users expected room dividers to be tappable and directly editable (drag to move, spin to rotate, double-tap to save).
- Multiple actions at the same time: Being able to merge/divide and label simultaneously caused confusion — merging after labeling could undo prior work ("I don't want to label when I'm just going to need to re-label after I divide the room.").
- Auto-moving flow: Several users disliked the constrained flow where the app automatically moved them from room to room in seemingly random order, resulting in significant frustration.
Testing the internal engineering prototype revealed four critical usability issues before any design work began.
Step 1: Set up rooms with direct divider manipulation and spatial landmarks.
Step 2: Label rooms with a clear list and custom naming, separated from divider editing.
Explore ideal interaction
I created an initial prototype to test the ideal interaction model — addressing the pain points head-on before engineering constraints entered the picture.
- Display landmarks to help with orientation
- Break the flow into 3 clear steps
- Edit room dividers directly (batch editing)
😊 "The 3-step flow was easy to follow. The interaction of direct manipulation of the room dividers was intuitive. Participants appreciated the white clutter on the map."
⚠️ Engineering: "This is awesome! So much better!" — However, "we can't build this for MVP. The ideal interaction you want is out of scope."
The iterated design reduced interaction complexity while preserving the core UX insight: separate divider editing from room labeling.
Iterate the design to fit the dev timeline
With the ideal experience validated but out of MVP scope, I adapted the design to work within engineering constraints — preserving the key UX wins while fitting the build timeline.
- Reduced interaction: Saved one divider's editing at a time instead of the ideal batch editing. Used but refined existing UI components.
- List instead of tabs: More accessible and easier to build — the list pattern was also more scalable for homes with many rooms.
Final Design
The complete map customization flow
Outcomes
Shipped, patented, and celebrated
The work resulted in two US patents: Mapping interface for mobile robots and Map based training and interface for mobile robots.
Lessons learned
- High completion rate doesn't mean users are fully satisfied with the customization experience — some users still struggle with map comprehension and editing.
- Educational content is needed even in areas where technology is ahead of users' expectations.
- Visual mapping is a significant challenge for accessibility design.